AM5 Noctua Cooling

Benchmarks for a future article by Michael Larabel.

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2301314-NE-AM5NOCTUA33
Jump To Table - Results

View

Do Not Show Noisy Results
Do Not Show Results With Incomplete Data
Do Not Show Results With Little Change/Spread
List Notable Results
Show Result Confidence Charts

Limit displaying results to tests within:

Web Browsers 1 Tests
C/C++ Compiler Tests 2 Tests
CPU Massive 3 Tests
Creator Workloads 6 Tests
Database Test Suite 3 Tests
Encoding 2 Tests
HPC - High Performance Computing 3 Tests
Common Kernel Benchmarks 2 Tests
Machine Learning 3 Tests
Multi-Core 5 Tests
Intel oneAPI 2 Tests
Python Tests 4 Tests
Server 3 Tests
Video Encoding 2 Tests

Statistics

Show Overall Harmonic Mean(s)
Show Overall Geometric Mean
Show Geometric Means Per-Suite/Category
Show Wins / Losses Counts (Pie Chart)
Normalize Results
Remove Outliers Before Calculating Averages

Graph Settings

Force Line Graphs Where Applicable
Convert To Scalar Where Applicable
Prefer Vertical Bar Graphs
No Box Plots
On Line Graphs With Missing Data, Connect The Line Gaps

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs
Condense Test Profiles With Multiple Version Results Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Highlight
Result
Hide
Result
Result
Identifier
View Logs
Performance Per
Dollar
Date
Run
  Test
  Duration
Wraith
January 29 2023
  9 Hours, 21 Minutes
Noctua NH-L9a-AM5
January 30 2023
  8 Hours, 34 Minutes
Invert Hiding All Results Option
  8 Hours, 58 Minutes
Only show results matching title/arguments (delimit multiple options with a comma):
Do not show results matching title/arguments (delimit multiple options with a comma):

Wraith

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203
Python Notes: Python 3.10.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional IBRS_FW STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Noctua NH-L9a-AM5

Processor: AMD Ryzen 9 7900 12-Core @ 3.70GHz (12 Cores / 24 Threads), Motherboard: Gigabyte B650M DS3H (F4h BIOS), Chipset: AMD Device 14d8, Memory: 32GB, Disk: 1000GB Sabrent Rocket 4.0 Plus, Graphics: Gigabyte AMD Raphael 512MB (2200/2400MHz), Audio: AMD Rembrandt Radeon HD Audio, Monitor: ASUS VP28U, Network: Realtek RTL8125 2.5GbE

OS: Ubuntu 22.10, Kernel: 6.2.0-060200rc5daily20230129-generic (x86_64), Desktop: GNOME Shell 43.0, Display Server: X Server 1.21.1.4 + Wayland, OpenGL: 4.6 Mesa 23.0.0-devel (git-e20564c 2022-12-12 kinetic-oibaf-ppa) (LLVM 15.0.5 DRM 3.49), Vulkan: 1.3.235, Compiler: GCC 12.2.0, File-System: ext4, Screen Resolution: 3840x2160

AM5 Noctua CoolingOpenBenchmarking.orgPhoronix Test SuiteAMD Ryzen 9 7900 12-Core @ 3.70GHz (12 Cores / 24 Threads)Gigabyte B650M DS3H (F4h BIOS)AMD Device 14d832GB1000GB Sabrent Rocket 4.0 PlusGigabyte AMD Raphael 512MB (2200/2400MHz)AMD Rembrandt Radeon HD AudioASUS VP28URealtek RTL8125 2.5GbEUbuntu 22.106.2.0-060200rc5daily20230129-generic (x86_64)GNOME Shell 43.0X Server 1.21.1.4 + Wayland4.6 Mesa 23.0.0-devel (git-e20564c 2022-12-12 kinetic-oibaf-ppa) (LLVM 15.0.5 DRM 3.49)1.3.235GCC 12.2.0ext43840x2160ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerOpenGLVulkanCompilerFile-SystemScreen ResolutionAM5 Noctua Cooling BenchmarksSystem Logs- Transparent Huge Pages: madvise- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v - Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203- Python 3.10.7- itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional IBRS_FW STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Wraith vs. Noctua NH-L9a-AM5 ComparisonPhoronix Test SuiteBaseline+0.8%+0.8%+1.6%+1.6%+2.4%+2.4%3.1%3.1%2.9%2.7%2.4%2%2%N.T.C.D.m - S.S.SN.T.C.D.m - S.S.S100 - 800 - Read Only - Average Latency100 - 800 - Read Only100 - 100 - Read Only - Average Latency2.6%100 - 100 - Read Only2.6%A.G.R.R.0.F - CPU100 - 250 - Read Only2.4%1.R.H.D.F.R.C.C2.3%100 - 500 - Read Only - Average Latency2.3%100 - 500 - Read Only2.3%100 - 250 - Read Only - Average Latency2.1%N.T.C.D.m - A.M.SN.T.C.D.m - A.M.SNeural Magic DeepSparseNeural Magic DeepSparsePostgreSQLPostgreSQLPostgreSQLPostgreSQLOpenVINOPostgreSQLClickHousePostgreSQLPostgreSQLPostgreSQLNeural Magic DeepSparseNeural Magic DeepSparseWraithNoctua NH-L9a-AM5

AM5 Noctua Coolingdeepsparse: NLP Text Classification, DistilBERT mnli - Synchronous Single-Streamdeepsparse: NLP Text Classification, DistilBERT mnli - Synchronous Single-Streampgbench: 100 - 800 - Read Only - Average Latencypgbench: 100 - 800 - Read Onlypgbench: 100 - 100 - Read Only - Average Latencypgbench: 100 - 100 - Read Onlyopenvino: Age Gender Recognition Retail 0013 FP16 - CPUpgbench: 100 - 250 - Read Onlyclickhouse: 100M Rows Hits Dataset, First Run / Cold Cachepgbench: 100 - 500 - Read Only - Average Latencypgbench: 100 - 500 - Read Onlypgbench: 100 - 250 - Read Only - Average Latencydeepsparse: NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Streamdeepsparse: NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Streamclickhouse: 100M Rows Hits Dataset, Second Runselenium: Maze Solver - Google Chromerocksdb: Rand Fill Syncopenvino: Person Detection FP32 - CPUpgbench: 100 - 100 - Read Write - Average Latencybrl-cad: VGR Performance Metricopenvino: Person Detection FP32 - CPUonednn: Convolution Batch Shapes Auto - bf16bf16bf16 - CPUpgbench: 100 - 100 - Read Writerocksdb: Seq Fillrocksdb: Read While Writingdeepsparse: CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Streamdeepsparse: CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Streamdeepsparse: CV Detection, YOLOv5s COCO - Asynchronous Multi-Streamdeepsparse: CV Detection, YOLOv5s COCO - Asynchronous Multi-Streamopenvino: Person Vehicle Bike Detection FP16 - CPUopenems: openEMS MSL_NotchFilteropenvino: Person Vehicle Bike Detection FP16 - CPUkvazaar: Bosphorus 1080p - Slowonednn: Deconvolution Batch shapes_1d - bf16bf16bf16 - CPUrocksdb: Read Rand Write Randopenvino: Person Detection FP16 - CPUonednn: IP Shapes 1D - bf16bf16bf16 - CPUopenvino: Vehicle Detection FP16 - CPUpgbench: 100 - 1000 - Read Only - Average Latencyrocksdb: Update Randopenvino: Vehicle Detection FP16 - CPUopenvino: Vehicle Detection FP16-INT8 - CPUpgbench: 100 - 1000 - Read Onlydeepsparse: NLP Text Classification, BERT base uncased SST2 - Asynchronous Multi-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2 - Asynchronous Multi-Streamopenvino: Vehicle Detection FP16-INT8 - CPUonednn: Recurrent Neural Network Inference - bf16bf16bf16 - CPUpgbench: 100 - 800 - Read Writepgbench: 100 - 800 - Read Write - Average Latencydeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Streamkvazaar: Bosphorus 1080p - Mediumopenvino: Age Gender Recognition Retail 0013 FP16 - CPUpgbench: 100 - 250 - Read Write - Average Latencypgbench: 100 - 250 - Read Writedeepsparse: CV Classification, ResNet-50 ImageNet - Synchronous Single-Streamdeepsparse: CV Classification, ResNet-50 ImageNet - Synchronous Single-Streamopenvino: Weld Porosity Detection FP16 - CPUonednn: Deconvolution Batch shapes_3d - bf16bf16bf16 - CPUkvazaar: Bosphorus 4K - Very Fastopenvino: Weld Porosity Detection FP16 - CPUdeepsparse: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Synchronous Single-Streamopenvino: Age Gender Recognition Retail 0013 FP16-INT8 - CPUdeepsparse: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Synchronous Single-Streamopenvino: Person Detection FP16 - CPUdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2 - Synchronous Single-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2 - Synchronous Single-Streamonednn: Recurrent Neural Network Training - bf16bf16bf16 - CPUopenvino: Weld Porosity Detection FP16-INT8 - CPUpgbench: 100 - 1000 - Read Writeopenvino: Face Detection FP16 - CPUopenvino: Weld Porosity Detection FP16-INT8 - CPUdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Streamdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Streamrocksdb: Rand Filldeepsparse: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Asynchronous Multi-Streamdeepsparse: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Asynchronous Multi-Streamopenvino: Face Detection FP16 - CPUpgbench: 100 - 500 - Read Writeuvg266: Bosphorus 1080p - Slowselenium: Kraken - Google Chromekvazaar: Bosphorus 4K - Ultra Fastkvazaar: Bosphorus 1080p - Ultra Fastuvg266: Bosphorus 1080p - Very Fastkvazaar: Bosphorus 4K - Mediumuvg266: Bosphorus 1080p - Mediumdeepsparse: CV Detection, YOLOv5s COCO - Synchronous Single-Streamopenvino: Face Detection FP16-INT8 - CPUdeepsparse: CV Detection, YOLOv5s COCO - Synchronous Single-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Streamuvg266: Bosphorus 4K - Mediumopenvino: Machine Translation EN To DE FP16 - CPUpgbench: 100 - 500 - Read Write - Average Latencyopenvino: Face Detection FP16-INT8 - CPUonednn: Matrix Multiply Batch Shapes Transformer - bf16bf16bf16 - CPUopenvino: Machine Translation EN To DE FP16 - CPUonednn: IP Shapes 3D - bf16bf16bf16 - CPUuvg266: Bosphorus 4K - Super Fastuvg266: Bosphorus 1080p - Ultra Fastdeepsparse: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Asynchronous Multi-Streamdeepsparse: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Asynchronous Multi-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Streamuvg266: Bosphorus 4K - Slowdeepsparse: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Synchronous Single-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Streamdeepsparse: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Synchronous Single-Streampgbench: 100 - 1000 - Read Write - Average Latencykvazaar: Bosphorus 1080p - Super Fastkvazaar: Bosphorus 4K - Slowuvg266: Bosphorus 4K - Very Fastkvazaar: Bosphorus 4K - Super Fastclickhouse: 100M Rows Hits Dataset, Third Rundeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Streamuvg266: Bosphorus 4K - Ultra Fastuvg266: Bosphorus 1080p - Super Fastdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Streamrocksdb: Rand Readselenium: WASM collisionDetection - Google Chromeopenvino: Age Gender Recognition Retail 0013 FP16-INT8 - CPUkvazaar: Bosphorus 1080p - Very Fastopenems: pyEMS Couplerselenium: WASM imageConvolute - Google Chromeselenium: PSPDFKit WASM - Google ChromeWraithNoctua NH-L9a-AM511.624086.00130.64812356640.07713049280.421295403205.190.39412683750.193128.066746.8367228.146.3157221177.316.4602851945.062.235701551912260773144715202.321029.641273.622281.42591101.3363.765.4460.875.4301725199541172.391.10886605.520.7987686559.95.02125294692.369964.94931195.04834.0301552051.55227.8248215.606563.5928590.2514.928167487.8579127.15386.352.0639536.28944.8619.320416963.5951.74175.10123.49108.097522.737243.97691632.361886.01154859.456.368.0881123.6341116661377.956476.9390631.511643745.49376.759.63239.06120.5715.0850.6864.2223321.9815.550124.606640.624310.6764.0230.42318.610.30182693.681.7617832.58157.72298.633420.0754684.28199.54138.84558.73627.196864.734177.2514.7730.4645.75235.24684.932138.51133.178.7343105090210230.810.7117.1743.8318.2411.270488.69540.63012693510.07912719180.411265126200.540.40312403650.197130.621345.9209232.456.2159531193.036.3782817955.002.212471568112141113114168204.133229.378574.217180.77461110.0163.295.4061.315.3929625035671164.821.10190609.330.8037734619.844.99124575091.846365.31691201.74829.4701560551.27227.9707214.484963.9228734.2314.854168317.8962126.53796.322.0548536.44948.9919.242017032.3251.95105.12123.03748.127322.654744.13701626.601892.32154359.486.348.1135123.2474116296678.192176.7125629.791647845.38377.659.49238.53120.3115.1150.5864.0981321.3615.580024.653440.547310.6963.9130.37218.640.30134693.821.7640532.62157.91298.989620.0515685.07469.53138.70578.72757.203964.793177.0914.7830.4445.78235.09684.523138.49133.138.7367105081082230.820.7115.1444.2918.03OpenBenchmarking.org

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM53691215SE +/- 0.07, N = 3SE +/- 0.06, N = 311.6211.27
MinAvgMaxNoctua NH-L9a-AM5300047735439Wraith300048395439OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.579.692.2Noctua NH-L9a-AM510.677.792.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3289 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3288 Joules per run.
MinAvgMaxWraith54.873.077.3Noctua NH-L9a-AM553.069.273.5OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM520406080100SE +/- 0.54, N = 3SE +/- 0.45, N = 386.0088.70

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 800 - Mode: Read Only - Average LatencyWraithNoctua NH-L9a-AM50.14580.29160.43740.58320.729SE +/- 0.001, N = 3SE +/- 0.003, N = 30.6480.6301. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxWraith300046915440Noctua NH-L9a-AM5300046925439OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.580.192.2Noctua NH-L9a-AM510.480.092.2OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 11235 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 11175 Joules per run.
MinAvgMaxWraith48.376.283.1Noctua NH-L9a-AM547.574.580.8OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 800 - Mode: Read OnlyWraithNoctua NH-L9a-AM5300K600K900K1200K1500KSE +/- 2024.35, N = 3SE +/- 5877.09, N = 3123566412693511. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 100 - Mode: Read Only - Average LatencyNoctua NH-L9a-AM5Wraith0.01780.03560.05340.07120.089SE +/- 0.000, N = 3SE +/- 0.001, N = 30.0790.0771. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxNoctua NH-L9a-AM5300046885439Wraith300046965440OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM59.279.9118.2Wraith10.379.992.2OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor40801201602001. Noctua NH-L9a-AM5: Approximate power consumption of 11138 Joules per run.2. Wraith: Approximate power consumption of 11134 Joules per run.
MinAvgMaxWraith44.075.582.4Noctua NH-L9a-AM544.574.180.0OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 100 - Mode: Read OnlyNoctua NH-L9a-AM5Wraith300K600K900K1200K1500KSE +/- 1401.49, N = 3SE +/- 10312.47, N = 3127191813049281. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Age Gender Recognition Retail 0013 FP16 - Device: CPUWraithNoctua NH-L9a-AM50.09450.1890.28350.3780.4725SE +/- 0.00, N = 3SE +/- 0.00, N = 30.420.41MIN: 0.24 / MAX: 8.88MIN: 0.23 / MAX: 8.841. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300043075190Noctua NH-L9a-AM5370043245219OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith12.085.1116.1Noctua NH-L9a-AM511.284.792.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor40801201602001. Wraith: Approximate power consumption of 5448 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5451 Joules per run.
MinAvgMaxWraith59.977.179.6Noctua NH-L9a-AM558.474.476.6OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 250 - Mode: Read OnlyNoctua NH-L9a-AM5Wraith300K600K900K1200K1500KSE +/- 14728.11, N = 3SE +/- 2942.90, N = 3126512612954031. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

ClickHouse

ClickHouse is an open-source, high performance OLAP data management system. This test profile uses ClickHouse's standard benchmark recommendations per https://clickhouse.com/docs/en/operations/performance-test/ / https://github.com/ClickHouse/ClickBench/tree/main/clickhouse with the 100 million rows web analytics dataset. The reported value is the query processing time using the geometric mean of all separate queries performed as an aggregate. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgQueries Per Minute, Geo Mean, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, First Run / Cold CacheNoctua NH-L9a-AM5Wraith50100150200250SE +/- 1.23, N = 3SE +/- 2.04, N = 5200.54205.19MIN: 14 / MAX: 8571.43MIN: 13.88 / MAX: 8571.43

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 500 - Mode: Read Only - Average LatencyNoctua NH-L9a-AM5Wraith0.09070.18140.27210.36280.4535SE +/- 0.004, N = 3SE +/- 0.003, N = 30.4030.3941. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxWraith300046885440Noctua NH-L9a-AM5300047035439OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.180.196.4Wraith10.579.992.2OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 11162 Joules per run.2. Wraith: Approximate power consumption of 11193 Joules per run.
MinAvgMaxWraith45.376.383.3Noctua NH-L9a-AM544.173.781.3OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 500 - Mode: Read OnlyNoctua NH-L9a-AM5Wraith300K600K900K1200K1500KSE +/- 12175.03, N = 3SE +/- 11732.69, N = 3124036512683751. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 250 - Mode: Read Only - Average LatencyNoctua NH-L9a-AM5Wraith0.04430.08860.13290.17720.2215SE +/- 0.002, N = 3SE +/- 0.001, N = 30.1970.1931. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxNoctua NH-L9a-AM5300046785439Wraith300046975440OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.079.992.3Wraith10.479.892.2OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 11186 Joules per run.2. Wraith: Approximate power consumption of 11151 Joules per run.
MinAvgMaxWraith44.375.982.9Noctua NH-L9a-AM544.073.880.4OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM5306090120150SE +/- 0.37, N = 3SE +/- 0.44, N = 3128.07130.62

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM51122334455SE +/- 0.13, N = 3SE +/- 0.15, N = 346.8445.92
MinAvgMaxNoctua NH-L9a-AM5300046435439Wraith300046675440OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.179.692.3Noctua NH-L9a-AM56.179.291.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3291 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3301 Joules per run.
MinAvgMaxWraith56.674.177.5Noctua NH-L9a-AM555.071.074.1OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

ClickHouse

ClickHouse is an open-source, high performance OLAP data management system. This test profile uses ClickHouse's standard benchmark recommendations per https://clickhouse.com/docs/en/operations/performance-test/ / https://github.com/ClickHouse/ClickBench/tree/main/clickhouse with the 100 million rows web analytics dataset. The reported value is the query processing time using the geometric mean of all separate queries performed as an aggregate. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgQueries Per Minute, Geo Mean, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, Second RunWraithNoctua NH-L9a-AM550100150200250SE +/- 1.81, N = 5SE +/- 1.43, N = 3228.14232.45MIN: 14.25 / MAX: 10000MIN: 14.21 / MAX: 10000

Selenium

This test profile uses the Selenium WebDriver for running various browser benchmarks in different available web browsers such as Mozilla Firefox and Google Chrome. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterSeleniumBenchmark: Maze Solver - Browser: Google ChromeWraithNoctua NH-L9a-AM5246810SE +/- 0.08, N = 15SE +/- 0.03, N = 46.36.21. chrome 109.0.5414.119
MinAvgMaxWraith300037705440Noctua NH-L9a-AM5300037945414OpenBenchmarking.orgMegahertz, More Is BetterSeleniumCPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM59.917.132.7Wraith10.116.338.5OpenBenchmarking.orgWatts, Fewer Is BetterSeleniumCPU Power Consumption Monitor11223344551. Noctua NH-L9a-AM5: Approximate power consumption of 338 Joules per run.2. Wraith: Approximate power consumption of 324 Joules per run.
MinAvgMaxNoctua NH-L9a-AM550.055.561.4Wraith44.150.564.4OpenBenchmarking.orgCelsius, Fewer Is BetterSeleniumCPU Temperature Monitor20406080100

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Random Fill SyncWraithNoctua NH-L9a-AM53K6K9K12K15KSE +/- 147.40, N = 3SE +/- 67.67, N = 315722159531. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Random Fill SyncNoctua NH-L9a-AM5Wraith120240360480600541.09553.97
MinAvgMaxNoctua NH-L9a-AM5300047305387Wraith300047325239OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.429.535.5Wraith9.328.439.9OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor11223344551. Noctua NH-L9a-AM5: Approximate power consumption of 1926 Joules per run.2. Wraith: Approximate power consumption of 1835 Joules per run.
MinAvgMaxWraith50.554.562.5Noctua NH-L9a-AM550.453.560.8OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Person Detection FP32 - Device: CPUNoctua NH-L9a-AM5Wraith30060090012001500SE +/- 10.05, N = 3SE +/- 4.20, N = 31193.031177.31MIN: 1010.43 / MAX: 1377.34MIN: 677.05 / MAX: 1380.591. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith370045195389Noctua NH-L9a-AM5370045715141OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.784.492.3Wraith12.184.392.3OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5653 Joules per run.2. Wraith: Approximate power consumption of 5620 Joules per run.
MinAvgMaxWraith60.580.588.0Noctua NH-L9a-AM558.177.783.4OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 100 - Mode: Read Write - Average LatencyWraithNoctua NH-L9a-AM5246810SE +/- 0.099, N = 12SE +/- 0.047, N = 36.4606.3781. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxNoctua NH-L9a-AM5299239275439Wraith297139595440OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.036.690.0Wraith6.035.390.1OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5160 Joules per run.2. Wraith: Approximate power consumption of 4974 Joules per run.
MinAvgMaxNoctua NH-L9a-AM547.356.576.4Wraith45.955.276.4OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

BRL-CAD

BRL-CAD is a cross-platform, open-source solid modeling system with built-in benchmark mode. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgVGR Performance Metric, More Is BetterBRL-CAD 7.34VGR Performance MetricNoctua NH-L9a-AM5Wraith60K120K180K240K300K2817952851941. (CXX) g++ options: -std=c++14 -pipe -fvisibility=hidden -fno-strict-aliasing -fno-common -fexceptions -ftemplate-depth-128 -m64 -ggdb3 -O3 -fipa-pta -fstrength-reduce -finline-functions -flto -ltcl8.6 -lregex_brl -lz_brl -lnetpbm -ldl -lm -ltk8.6
OpenBenchmarking.orgVGR Performance Metric Per Watt, More Is BetterBRL-CAD 7.34VGR Performance MetricNoctua NH-L9a-AM5Wraith70014002100280035003159.133196.96
MinAvgMaxWraith370044645330Noctua NH-L9a-AM5370044785239OpenBenchmarking.orgMegahertz, More Is BetterBRL-CAD 7.34CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.889.292.2Noctua NH-L9a-AM511.889.292.2OpenBenchmarking.orgWatts, Fewer Is BetterBRL-CAD 7.34CPU Power Consumption Monitor20406080100
MinAvgMaxWraith55.184.191.3Noctua NH-L9a-AM553.081.688.1OpenBenchmarking.orgCelsius, Fewer Is BetterBRL-CAD 7.34CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Person Detection FP32 - Device: CPUNoctua NH-L9a-AM5Wraith1.13852.2773.41554.5545.6925SE +/- 0.04, N = 3SE +/- 0.02, N = 35.005.061. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: Convolution Batch Shapes Auto - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM50.5031.0061.5092.0122.515SE +/- 0.01794, N = 9SE +/- 0.00682, N = 72.235702.21247MIN: 2.12MIN: 2.131. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxNoctua NH-L9a-AM5300042855439Wraith319543655028OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith12.058.991.0Noctua NH-L9a-AM511.658.192.2OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 589 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 589 Joules per run.
MinAvgMaxWraith60.470.974.4Noctua NH-L9a-AM558.668.472.3OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 100 - Mode: Read WriteWraithNoctua NH-L9a-AM53K6K9K12K15KSE +/- 234.08, N = 12SE +/- 116.23, N = 315519156811. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Sequential FillNoctua NH-L9a-AM5Wraith300K600K900K1200K1500KSE +/- 642.22, N = 3SE +/- 3237.54, N = 3121411112260771. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Sequential FillNoctua NH-L9a-AM5Wraith3K6K9K12K15K15945.3916241.97
MinAvgMaxNoctua NH-L9a-AM5370046674928Wraith370046955140OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.476.190.1Wraith11.575.590.1OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1802 Joules per run.2. Wraith: Approximate power consumption of 1812 Joules per run.
MinAvgMaxWraith58.378.484.0Noctua NH-L9a-AM557.076.480.1OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Read While WritingNoctua NH-L9a-AM5Wraith700K1400K2100K2800K3500KSE +/- 7283.63, N = 3SE +/- 43430.73, N = 3311416831447151. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Read While WritingNoctua NH-L9a-AM5Wraith8K16K24K32K40K36740.5636978.18
MinAvgMaxWraith300046974844Noctua NH-L9a-AM5300046975102OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.485.090.1Noctua NH-L9a-AM511.584.890.1OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5443 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5453 Joules per run.
MinAvgMaxWraith59.080.483.6Noctua NH-L9a-AM557.177.580.9OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM54080120160200SE +/- 0.27, N = 3SE +/- 0.25, N = 3202.32204.13

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM5714212835SE +/- 0.04, N = 3SE +/- 0.04, N = 329.6429.38
MinAvgMaxNoctua NH-L9a-AM5331446815098Wraith370046914986OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.281.092.2Wraith11.080.992.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3215 Joules per run.2. Wraith: Approximate power consumption of 3209 Joules per run.
MinAvgMaxWraith57.875.579.0Noctua NH-L9a-AM556.172.876.1OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: CV Detection, YOLOv5s COCO - Scenario: Asynchronous Multi-StreamNoctua NH-L9a-AM5Wraith1632486480SE +/- 0.44, N = 3SE +/- 0.23, N = 374.2273.62
MinAvgMaxWraith300047725193Noctua NH-L9a-AM5370047795165OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.881.092.2Wraith11.780.692.8OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3239 Joules per run.2. Wraith: Approximate power consumption of 3252 Joules per run.
MinAvgMaxWraith59.076.280.3Noctua NH-L9a-AM557.474.277.6OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: CV Detection, YOLOv5s COCO - Scenario: Asynchronous Multi-StreamNoctua NH-L9a-AM5Wraith20406080100SE +/- 0.47, N = 3SE +/- 0.27, N = 380.7781.43

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Person Vehicle Bike Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM52004006008001000SE +/- 1.48, N = 3SE +/- 1.08, N = 31101.331110.011. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

OpenEMS

OpenEMS is a free and open electromagnetic field solver using the FDTD method. This test profile runs OpenEMS and pyEMS benchmark demos. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMCells/s, More Is BetterOpenEMS 0.0.35-86Test: openEMS MSL_NotchFilterNoctua NH-L9a-AM5Wraith1428425670SE +/- 0.25, N = 3SE +/- 0.59, N = 363.2963.761. (CXX) g++ options: -O3 -rdynamic -ltinyxml -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -lexpat
OpenBenchmarking.orgMCells/s Per Watt, More Is BetterOpenEMS 0.0.35-86Test: openEMS MSL_NotchFilterWraithNoctua NH-L9a-AM50.32650.6530.97951.3061.63251.3771.451
MinAvgMaxNoctua NH-L9a-AM5299337525438Wraith299339325438OpenBenchmarking.orgMegahertz, More Is BetterOpenEMS 0.0.35-86CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith10.946.362.4Noctua NH-L9a-AM511.843.661.4OpenBenchmarking.orgWatts, Fewer Is BetterOpenEMS 0.0.35-86CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3041 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 2879 Joules per run.
MinAvgMaxWraith59.466.280.3Noctua NH-L9a-AM558.063.977.8OpenBenchmarking.orgCelsius, Fewer Is BetterOpenEMS 0.0.35-86CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Person Vehicle Bike Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM51.2242.4483.6724.8966.12SE +/- 0.01, N = 3SE +/- 0.01, N = 35.445.40MIN: 4 / MAX: 15.22MIN: 3.98 / MAX: 14.511. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith370044975160Noctua NH-L9a-AM5370045055181OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM58.884.993.1Wraith12.284.992.9OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5462 Joules per run.2. Wraith: Approximate power consumption of 5460 Joules per run.
MinAvgMaxWraith60.578.480.9Noctua NH-L9a-AM558.175.577.6OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: SlowWraithNoctua NH-L9a-AM51428425670SE +/- 0.32, N = 5SE +/- 0.38, N = 560.8761.311. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: SlowWraithNoctua NH-L9a-AM50.2090.4180.6270.8361.0450.9250.929
MinAvgMaxNoctua NH-L9a-AM5300045925268Wraith370046455439OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.066.091.3Wraith10.565.891.4OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 924 Joules per run.2. Wraith: Approximate power consumption of 908 Joules per run.
MinAvgMaxWraith56.378.084.1Noctua NH-L9a-AM555.075.981.6OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: Deconvolution Batch shapes_1d - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM51.22182.44363.66544.88726.109SE +/- 0.02649, N = 3SE +/- 0.00123, N = 35.430175.39296MIN: 4.91MIN: 4.911. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith370045075438Noctua NH-L9a-AM5370045275438OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.074.392.1Wraith11.374.192.2OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1856 Joules per run.2. Wraith: Approximate power consumption of 1878 Joules per run.
MinAvgMaxWraith55.475.481.6Noctua NH-L9a-AM553.973.178.9OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Read Random Write RandomNoctua NH-L9a-AM5Wraith500K1000K1500K2000K2500KSE +/- 12706.45, N = 3SE +/- 10913.09, N = 3250356725199541. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Read Random Write RandomNoctua NH-L9a-AM5Wraith6K12K18K24K30K29467.4529720.72
MinAvgMaxWraith300047074876Noctua NH-L9a-AM5300047214877OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor12002400360048006000
MinAvgMaxNoctua NH-L9a-AM59.485.090.1Wraith11.384.890.1OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5409 Joules per run.2. Wraith: Approximate power consumption of 5455 Joules per run.
MinAvgMaxWraith59.981.384.9Noctua NH-L9a-AM558.478.682.1OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Person Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM530060090012001500SE +/- 1.18, N = 3SE +/- 4.76, N = 31172.391164.82MIN: 669.11 / MAX: 1373.05MIN: 731.65 / MAX: 1351.61. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxNoctua NH-L9a-AM5300045315252Wraith364445495165OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith12.184.392.3Noctua NH-L9a-AM511.984.192.3OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5622 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5636 Joules per run.
MinAvgMaxWraith60.080.285.6Noctua NH-L9a-AM557.877.187.3OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: IP Shapes 1D - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM50.24950.4990.74850.9981.2475SE +/- 0.00091, N = 4SE +/- 0.00191, N = 41.108861.10190MIN: 1MIN: 0.991. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith300045015039Noctua NH-L9a-AM5300045075042OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM55.773.692.3Wraith11.873.392.3OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1399 Joules per run.2. Wraith: Approximate power consumption of 1375 Joules per run.
MinAvgMaxWraith52.369.374.0Noctua NH-L9a-AM550.567.071.5OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Vehicle Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM5130260390520650SE +/- 3.68, N = 3SE +/- 4.07, N = 3605.52609.331. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 1000 - Mode: Read Only - Average LatencyNoctua NH-L9a-AM5Wraith0.18070.36140.54210.72280.9035SE +/- 0.005, N = 3SE +/- 0.005, N = 30.8030.7981. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxNoctua NH-L9a-AM5300046645439Wraith300046765440OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM59.679.998.6Wraith10.179.692.2OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 11179 Joules per run.2. Wraith: Approximate power consumption of 11225 Joules per run.
MinAvgMaxWraith50.676.882.5Noctua NH-L9a-AM550.175.582.0OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Update RandomWraithNoctua NH-L9a-AM5170K340K510K680K850KSE +/- 904.07, N = 3SE +/- 1466.13, N = 37686557734611. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Update RandomWraithNoctua NH-L9a-AM52K4K6K8K10K9049.689119.03
MinAvgMaxNoctua NH-L9a-AM5345547425077Wraith300047454953OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.984.990.1Noctua NH-L9a-AM511.384.891.2OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5408 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5457 Joules per run.
MinAvgMaxWraith60.183.587.4Noctua NH-L9a-AM558.980.884.3OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Vehicle Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM53691215SE +/- 0.06, N = 3SE +/- 0.07, N = 39.909.84MIN: 4.2 / MAX: 21.61MIN: 4.88 / MAX: 21.381. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300045644861Noctua NH-L9a-AM5333745914888OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor12002400360048006000
MinAvgMaxNoctua NH-L9a-AM510.084.992.3Wraith11.784.892.3OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5435 Joules per run.2. Wraith: Approximate power consumption of 5424 Joules per run.
MinAvgMaxWraith60.877.880.3Noctua NH-L9a-AM558.375.177.6OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Vehicle Detection FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM51.12952.2593.38854.5185.6475SE +/- 0.01, N = 3SE +/- 0.01, N = 35.024.99MIN: 3.04 / MAX: 14.5MIN: 3.62 / MAX: 13.471. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300043774526Noctua NH-L9a-AM5300044005366OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith10.385.092.2Noctua NH-L9a-AM511.884.792.3OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5439 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5451 Joules per run.
MinAvgMaxWraith60.477.880.3Noctua NH-L9a-AM558.374.977.3OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 1000 - Mode: Read OnlyNoctua NH-L9a-AM5Wraith300K600K900K1200K1500KSE +/- 8627.67, N = 3SE +/- 8347.19, N = 3124575012529461. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM520406080100SE +/- 0.55, N = 3SE +/- 0.09, N = 392.3791.85
MinAvgMaxWraith300046145440Noctua NH-L9a-AM5370046255439OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.574.592.2Wraith11.673.592.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3577 Joules per run.2. Wraith: Approximate power consumption of 3554 Joules per run.
MinAvgMaxWraith52.870.775.4Noctua NH-L9a-AM551.068.172.9OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM51530456075SE +/- 0.39, N = 3SE +/- 0.06, N = 364.9565.32

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Vehicle Detection FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM530060090012001500SE +/- 2.68, N = 3SE +/- 2.53, N = 31195.041201.741. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM52004006008001000SE +/- 1.85, N = 3SE +/- 1.64, N = 3834.03829.47MIN: 826.35MIN: 822.381. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith300045845227Noctua NH-L9a-AM5300046155314OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.483.592.2Wraith11.983.592.2OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 6788 Joules per run.2. Wraith: Approximate power consumption of 6815 Joules per run.
MinAvgMaxWraith59.480.084.6Noctua NH-L9a-AM557.577.782.9OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 800 - Mode: Read WriteWraithNoctua NH-L9a-AM53K6K9K12K15KSE +/- 118.84, N = 3SE +/- 111.95, N = 315520156051. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 800 - Mode: Read Write - Average LatencyWraithNoctua NH-L9a-AM51224364860SE +/- 0.40, N = 3SE +/- 0.37, N = 351.5551.271. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxWraith299241365440Noctua NH-L9a-AM5299342295439OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM59.848.490.3Wraith6.348.090.3OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 6808 Joules per run.2. Wraith: Approximate power consumption of 6742 Joules per run.
MinAvgMaxWraith51.662.878.8Noctua NH-L9a-AM551.361.777.6OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM5714212835SE +/- 0.10, N = 3SE +/- 0.03, N = 327.8227.97

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM550100150200250SE +/- 0.76, N = 3SE +/- 0.21, N = 3215.61214.48
MinAvgMaxWraith300047405396Noctua NH-L9a-AM5370047775432OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM59.675.492.2Wraith11.474.892.3OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3797 Joules per run.2. Wraith: Approximate power consumption of 3764 Joules per run.
MinAvgMaxWraith58.375.383.8Noctua NH-L9a-AM557.173.281.4OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: MediumWraithNoctua NH-L9a-AM51428425670SE +/- 0.47, N = 5SE +/- 0.35, N = 563.5963.921. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: MediumWraithNoctua NH-L9a-AM50.22030.44060.66090.88121.10150.9680.979
MinAvgMaxWraith300045505304Noctua NH-L9a-AM5300045745432OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith10.565.791.6Noctua NH-L9a-AM56.365.391.6OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 894 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 888 Joules per run.
MinAvgMaxWraith55.676.782.6Noctua NH-L9a-AM553.974.780.5OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Age Gender Recognition Retail 0013 FP16 - Device: CPUWraithNoctua NH-L9a-AM56K12K18K24K30KSE +/- 57.74, N = 3SE +/- 41.32, N = 328590.2528734.231. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 250 - Mode: Read Write - Average LatencyWraithNoctua NH-L9a-AM548121620SE +/- 0.08, N = 3SE +/- 0.08, N = 314.9314.851. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxNoctua NH-L9a-AM5299236165439Wraith299137125440OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.234.690.1Wraith11.333.790.1OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 4865 Joules per run.2. Wraith: Approximate power consumption of 4750 Joules per run.
MinAvgMaxWraith46.856.378.1Noctua NH-L9a-AM546.955.678.3OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 250 - Mode: Read WriteWraithNoctua NH-L9a-AM54K8K12K16K20KSE +/- 91.73, N = 3SE +/- 96.50, N = 316748168311. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-StreamNoctua NH-L9a-AM5Wraith246810SE +/- 0.0066, N = 3SE +/- 0.0144, N = 37.89627.8579
MinAvgMaxNoctua NH-L9a-AM5300048175304Wraith300048235439OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.680.992.3Noctua NH-L9a-AM510.879.292.1OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3234 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3246 Joules per run.
MinAvgMaxWraith57.375.679.0Noctua NH-L9a-AM555.672.075.9OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-StreamNoctua NH-L9a-AM5Wraith306090120150SE +/- 0.10, N = 3SE +/- 0.23, N = 3126.54127.15

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Weld Porosity Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM5246810SE +/- 0.01, N = 3SE +/- 0.01, N = 36.356.32MIN: 3.25 / MAX: 14.93MIN: 3.23 / MAX: 14.441. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300044095214Noctua NH-L9a-AM5300044454564OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.385.392.7Noctua NH-L9a-AM511.885.192.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5462 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5419 Joules per run.
MinAvgMaxWraith60.478.280.5Noctua NH-L9a-AM558.075.177.5OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: Deconvolution Batch shapes_3d - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM50.46440.92881.39321.85762.322SE +/- 0.00663, N = 9SE +/- 0.00572, N = 92.063952.05485MIN: 1.92MIN: 1.911. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith300039984664Noctua NH-L9a-AM5300040264707OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor12002400360048006000
MinAvgMaxWraith11.945.792.1Noctua NH-L9a-AM510.145.292.2OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 320 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 322 Joules per run.
MinAvgMaxWraith57.164.568.3Noctua NH-L9a-AM555.462.466.3OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: Very FastWraithNoctua NH-L9a-AM5816243240SE +/- 0.02, N = 4SE +/- 0.01, N = 436.2836.441. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: Very FastWraithNoctua NH-L9a-AM50.11390.22780.34170.45560.56950.4970.506
MinAvgMaxWraith300044615304Noctua NH-L9a-AM5370045335350OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.773.089.7Noctua NH-L9a-AM510.472.089.7OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 1497 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 1457 Joules per run.
MinAvgMaxWraith58.177.982.1Noctua NH-L9a-AM556.175.379.8OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Weld Porosity Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM52004006008001000SE +/- 1.62, N = 3SE +/- 1.43, N = 3944.86948.991. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5510152025SE +/- 0.05, N = 3SE +/- 0.02, N = 319.3219.24
MinAvgMaxNoctua NH-L9a-AM5300048085422Wraith370048245415OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.176.292.2Noctua NH-L9a-AM59.476.092.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3377 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3397 Joules per run.
MinAvgMaxWraith57.873.577.3Noctua NH-L9a-AM556.571.476.8OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM54K8K12K16K20KSE +/- 22.38, N = 3SE +/- 28.68, N = 316963.5917032.321. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM51224364860SE +/- 0.13, N = 3SE +/- 0.06, N = 351.7451.95

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Person Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM51.1522.3043.4564.6085.76SE +/- 0.00, N = 3SE +/- 0.02, N = 35.105.121. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5306090120150SE +/- 0.23, N = 3SE +/- 0.08, N = 3123.49123.04
MinAvgMaxNoctua NH-L9a-AM5300047575430Wraith370048175440OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.776.392.2Noctua NH-L9a-AM511.173.892.1OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3486 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3493 Joules per run.
MinAvgMaxWraith58.474.777.8Noctua NH-L9a-AM555.871.174.9OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5246810SE +/- 0.0151, N = 3SE +/- 0.0053, N = 38.09758.1273

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5510152025SE +/- 0.18, N = 3SE +/- 0.17, N = 322.7422.65
MinAvgMaxNoctua NH-L9a-AM5300048295439Wraith300048885440OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.276.193.0Noctua NH-L9a-AM59.774.692.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3474 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3530 Joules per run.
MinAvgMaxWraith36.664.272.4Noctua NH-L9a-AM537.161.069.1OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM51020304050SE +/- 0.35, N = 3SE +/- 0.34, N = 343.9844.14

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM5400800120016002000SE +/- 3.38, N = 3SE +/- 1.28, N = 31632.361626.60MIN: 1623.03MIN: 1621.031. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith300046445340Noctua NH-L9a-AM5300046575308OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM59.283.593.1Wraith10.783.592.3OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 6902 Joules per run.2. Wraith: Approximate power consumption of 6902 Joules per run.
MinAvgMaxWraith54.578.684.4Noctua NH-L9a-AM552.876.382.1OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Weld Porosity Detection FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM5400800120016002000SE +/- 3.15, N = 3SE +/- 2.47, N = 31886.011892.321. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 1000 - Mode: Read WriteNoctua NH-L9a-AM5Wraith3K6K9K12K15KSE +/- 92.72, N = 3SE +/- 217.62, N = 1215435154851. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Face Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM53691215SE +/- 0.01, N = 3SE +/- 0.00, N = 39.459.481. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Weld Porosity Detection FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM5246810SE +/- 0.01, N = 3SE +/- 0.01, N = 36.366.34MIN: 3.54 / MAX: 14.82MIN: 3.48 / MAX: 13.921. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith370044124925Noctua NH-L9a-AM5300044164536OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.385.092.2Wraith12.084.892.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5443 Joules per run.2. Wraith: Approximate power consumption of 5454 Joules per run.
MinAvgMaxWraith61.178.180.6Noctua NH-L9a-AM558.375.077.5OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5246810SE +/- 0.0049, N = 3SE +/- 0.0095, N = 38.08818.1135

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5306090120150SE +/- 0.08, N = 3SE +/- 0.14, N = 3123.63123.25
MinAvgMaxNoctua NH-L9a-AM5300047295439Wraith300047765414OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith10.773.392.2Noctua NH-L9a-AM58.571.492.3OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3521 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3522 Joules per run.
MinAvgMaxWraith57.573.978.1Noctua NH-L9a-AM554.971.376.3OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Random FillNoctua NH-L9a-AM5Wraith200K400K600K800K1000KSE +/- 3379.18, N = 3SE +/- 1263.47, N = 3116296611666131. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Random FillNoctua NH-L9a-AM5Wraith3K6K9K12K15K13679.4613731.33
MinAvgMaxNoctua NH-L9a-AM5300047844913Wraith370047944919OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.785.090.1Wraith11.485.090.1OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5441 Joules per run.2. Wraith: Approximate power consumption of 5409 Joules per run.
MinAvgMaxWraith57.883.190.0Noctua NH-L9a-AM556.581.085.8OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM520406080100SE +/- 0.06, N = 3SE +/- 0.49, N = 377.9678.19

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM520406080100SE +/- 0.06, N = 3SE +/- 0.48, N = 376.9476.71
MinAvgMaxWraith300045695440Noctua NH-L9a-AM5300045795380OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.878.892.9Wraith11.778.493.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3414 Joules per run.2. Wraith: Approximate power consumption of 3422 Joules per run.
MinAvgMaxWraith58.174.381.8Noctua NH-L9a-AM556.972.280.9OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Face Detection FP16 - Device: CPUWraithNoctua NH-L9a-AM5140280420560700SE +/- 0.55, N = 3SE +/- 0.23, N = 3631.51629.79MIN: 552.36 / MAX: 655.2MIN: 553.7 / MAX: 655.391. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300044275148Noctua NH-L9a-AM5370044525164OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith12.284.292.1Noctua NH-L9a-AM511.184.292.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5584 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5555 Joules per run.
MinAvgMaxWraith61.578.888.3Noctua NH-L9a-AM559.876.586.0OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 500 - Mode: Read WriteWraithNoctua NH-L9a-AM54K8K12K16K20KSE +/- 120.21, N = 3SE +/- 149.35, N = 1216437164781. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: SlowNoctua NH-L9a-AM5Wraith1020304050SE +/- 0.13, N = 4SE +/- 0.14, N = 445.3845.49
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: SlowNoctua NH-L9a-AM5Wraith0.14450.2890.43350.5780.72250.6390.642
MinAvgMaxNoctua NH-L9a-AM5370045995307Wraith370046385371OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.771.091.9Wraith10.070.892.0OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1225 Joules per run.2. Wraith: Approximate power consumption of 1221 Joules per run.
MinAvgMaxWraith46.871.579.4Noctua NH-L9a-AM546.070.377.9OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Selenium

This test profile uses the Selenium WebDriver for running various browser benchmarks in different available web browsers such as Mozilla Firefox and Google Chrome. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterSeleniumBenchmark: Kraken - Browser: Google ChromeNoctua NH-L9a-AM5Wraith80160240320400SE +/- 2.52, N = 3SE +/- 3.32, N = 3377.6376.71. chrome 109.0.5414.119
MinAvgMaxWraith300041655440Noctua NH-L9a-AM5300042145439OpenBenchmarking.orgMegahertz, More Is BetterSeleniumCPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.222.547.9Wraith11.021.646.5OpenBenchmarking.orgWatts, Fewer Is BetterSeleniumCPU Power Consumption Monitor14284256701. Noctua NH-L9a-AM5: Approximate power consumption of 466 Joules per run.2. Wraith: Approximate power consumption of 447 Joules per run.
MinAvgMaxNoctua NH-L9a-AM541.553.466.4Wraith40.052.165.0OpenBenchmarking.orgCelsius, Fewer Is BetterSeleniumCPU Temperature Monitor20406080100

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: Ultra FastNoctua NH-L9a-AM5Wraith1326395265SE +/- 0.10, N = 5SE +/- 0.09, N = 559.4959.631. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: Ultra FastNoctua NH-L9a-AM5Wraith0.20140.40280.60420.80561.0070.8940.895
MinAvgMaxNoctua NH-L9a-AM5300043915273Wraith300044205311OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.666.691.2Noctua NH-L9a-AM511.166.690.2OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 933 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 932 Joules per run.
MinAvgMaxWraith55.874.279.6Noctua NH-L9a-AM553.872.077.4OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Ultra FastNoctua NH-L9a-AM5Wraith50100150200250SE +/- 0.60, N = 10SE +/- 0.56, N = 10238.53239.061. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Ultra FastWraithNoctua NH-L9a-AM51.28722.57443.86165.14886.4365.6935.721
MinAvgMaxNoctua NH-L9a-AM5370041605113Wraith300041655128OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.342.090.2Noctua NH-L9a-AM59.341.790.2OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 273 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 275 Joules per run.
MinAvgMaxWraith50.463.969.0Noctua NH-L9a-AM549.162.367.4OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: Very FastNoctua NH-L9a-AM5Wraith306090120150SE +/- 0.10, N = 7SE +/- 0.07, N = 7120.31120.57
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: Very FastNoctua NH-L9a-AM5Wraith0.49770.99541.49311.99082.48852.1912.212
MinAvgMaxNoctua NH-L9a-AM5300042435330Wraith370043115354OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.254.990.7Wraith10.054.590.2OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 494 Joules per run.2. Wraith: Approximate power consumption of 490 Joules per run.
MinAvgMaxWraith53.869.374.5Noctua NH-L9a-AM552.667.973.0OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: MediumWraithNoctua NH-L9a-AM548121620SE +/- 0.03, N = 3SE +/- 0.02, N = 315.0815.111. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: MediumWraithNoctua NH-L9a-AM50.04210.08420.12630.16840.21050.1870.187
MinAvgMaxNoctua NH-L9a-AM5300045165298Wraith300045295265OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.280.890.8Wraith11.980.790.1OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3527 Joules per run.2. Wraith: Approximate power consumption of 3522 Joules per run.
MinAvgMaxWraith56.881.085.1Noctua NH-L9a-AM555.378.582.3OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: MediumNoctua NH-L9a-AM5Wraith1122334455SE +/- 0.07, N = 4SE +/- 0.12, N = 450.5850.68
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: MediumNoctua NH-L9a-AM5Wraith0.1640.3280.4920.6560.820.7260.729
MinAvgMaxWraith370045515298Noctua NH-L9a-AM5370045855298OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.669.792.2Wraith11.769.592.2OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1114 Joules per run.2. Wraith: Approximate power consumption of 1113 Joules per run.
MinAvgMaxWraith53.074.180.6Noctua NH-L9a-AM551.973.179.8OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: CV Detection, YOLOv5s COCO - Scenario: Synchronous Single-StreamNoctua NH-L9a-AM5Wraith1428425670SE +/- 0.11, N = 3SE +/- 0.06, N = 364.1064.22

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Face Detection FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM570140210280350SE +/- 0.89, N = 3SE +/- 0.37, N = 3321.98321.36MIN: 227.32 / MAX: 342.52MIN: 285.91 / MAX: 334.031. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300044795140Noctua NH-L9a-AM5300044975164OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith12.284.492.2Noctua NH-L9a-AM58.884.492.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5541 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5539 Joules per run.
MinAvgMaxWraith60.578.584.6Noctua NH-L9a-AM558.576.282.4OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: CV Detection, YOLOv5s COCO - Scenario: Synchronous Single-StreamNoctua NH-L9a-AM5Wraith48121620SE +/- 0.03, N = 3SE +/- 0.02, N = 315.5815.55
MinAvgMaxWraith300048045155Noctua NH-L9a-AM5300048085432OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM59.680.692.9Wraith11.879.292.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3226 Joules per run.2. Wraith: Approximate power consumption of 3220 Joules per run.
MinAvgMaxWraith58.176.580.5Noctua NH-L9a-AM556.974.778.0OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5612182430SE +/- 0.02, N = 3SE +/- 0.01, N = 324.6124.65

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Synchronous Single-StreamWraithNoctua NH-L9a-AM5918273645SE +/- 0.03, N = 3SE +/- 0.02, N = 340.6240.55
MinAvgMaxNoctua NH-L9a-AM5300047405439Wraith300047655398OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith10.975.892.3Noctua NH-L9a-AM56.275.692.3OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3612 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3606 Joules per run.
MinAvgMaxWraith58.174.278.4Noctua NH-L9a-AM557.171.975.9OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: MediumWraithNoctua NH-L9a-AM53691215SE +/- 0.01, N = 3SE +/- 0.01, N = 310.6710.69
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: MediumWraithNoctua NH-L9a-AM50.02880.05760.08640.11520.1440.1280.128
MinAvgMaxNoctua NH-L9a-AM5300044555300Wraith300044965439OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM58.683.791.4Wraith10.983.291.4OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5021 Joules per run.2. Wraith: Approximate power consumption of 5018 Joules per run.
MinAvgMaxWraith58.582.085.5Noctua NH-L9a-AM556.680.183.5OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Machine Translation EN To DE FP16 - Device: CPUWraithNoctua NH-L9a-AM51428425670SE +/- 0.15, N = 3SE +/- 0.06, N = 364.0263.91MIN: 52.86 / MAX: 75.56MIN: 41.47 / MAX: 78.261. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300045755276Noctua NH-L9a-AM5300045965225OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM511.884.592.6Wraith12.084.492.7OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5439 Joules per run.2. Wraith: Approximate power consumption of 5486 Joules per run.
MinAvgMaxWraith60.678.881.4Noctua NH-L9a-AM558.376.178.5OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 500 - Mode: Read Write - Average LatencyWraithNoctua NH-L9a-AM5714212835SE +/- 0.22, N = 3SE +/- 0.29, N = 1230.4230.371. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
MinAvgMaxWraith299237245440Noctua NH-L9a-AM5299037485439OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption MonitorNoctua NH-L9a-AM5Wraith20406080100Min: 5.98 / Avg: 37.96 / Max: 90.19Min: 8.91 / Avg: 36.93 / Max: 90.121. Noctua NH-L9a-AM5: Approximate power consumption of 5337 Joules per run.2. Wraith: Approximate power consumption of 5182 Joules per run.
MinAvgMaxWraith48.157.878.8Noctua NH-L9a-AM546.555.076.8OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Face Detection FP16-INT8 - Device: CPUWraithNoctua NH-L9a-AM5510152025SE +/- 0.06, N = 3SE +/- 0.03, N = 318.6118.641. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: Matrix Multiply Batch Shapes Transformer - Data Type: bf16bf16bf16 - Engine: CPUWraithNoctua NH-L9a-AM50.06790.13580.20370.27160.3395SE +/- 0.000157, N = 4SE +/- 0.000096, N = 40.3018260.301346MIN: 0.28MIN: 0.281. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith300044925117Noctua NH-L9a-AM5370045425269OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.868.792.1Wraith11.168.692.1OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1099 Joules per run.2. Wraith: Approximate power consumption of 1098 Joules per run.
MinAvgMaxWraith54.868.072.0Noctua NH-L9a-AM552.865.769.6OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFPS, More Is BetterOpenVINO 2022.3Model: Machine Translation EN To DE FP16 - Device: CPUWraithNoctua NH-L9a-AM520406080100SE +/- 0.23, N = 3SE +/- 0.09, N = 393.6893.821. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared

oneDNN

This is a test of the Intel oneDNN as an Intel-optimized library for Deep Neural Networks and making use of its built-in benchdnn functionality. The result is the total perf time reported. Intel oneDNN was formerly known as DNNL (Deep Neural Network Library) and MKL-DNN before being rebranded as part of the Intel oneAPI toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 3.0Harness: IP Shapes 3D - Data Type: bf16bf16bf16 - Engine: CPUNoctua NH-L9a-AM5Wraith0.39690.79381.19071.58761.9845SE +/- 0.00606, N = 5SE +/- 0.00435, N = 51.764051.76178MIN: 1.66MIN: 1.671. (CXX) g++ options: -O3 -march=native -fopenmp -msse4.1 -fPIC -pie -ldl -lpthread
MinAvgMaxWraith300046275291Noctua NH-L9a-AM5333446495198OpenBenchmarking.orgMegahertz, More Is BetteroneDNN 3.0CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.765.692.2Wraith11.865.492.1OpenBenchmarking.orgWatts, Fewer Is BetteroneDNN 3.0CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 866 Joules per run.2. Wraith: Approximate power consumption of 864 Joules per run.
MinAvgMaxWraith55.470.778.9Noctua NH-L9a-AM553.368.776.8OpenBenchmarking.orgCelsius, Fewer Is BetteroneDNN 3.0CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Super FastWraithNoctua NH-L9a-AM5816243240SE +/- 0.06, N = 3SE +/- 0.08, N = 332.5832.62
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Super FastWraithNoctua NH-L9a-AM50.09830.19660.29490.39320.49150.4350.437
MinAvgMaxNoctua NH-L9a-AM5300043905272Wraith300044235322OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith10.974.991.1Noctua NH-L9a-AM511.674.692.0OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 1673 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 1692 Joules per run.
MinAvgMaxWraith57.476.781.1Noctua NH-L9a-AM556.174.679.1OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: Ultra FastWraithNoctua NH-L9a-AM5306090120150SE +/- 0.15, N = 8SE +/- 0.11, N = 8157.72157.91
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: Ultra FastWraithNoctua NH-L9a-AM50.72071.44142.16212.88283.60353.1663.203
MinAvgMaxNoctua NH-L9a-AM5370042705265Wraith300043035349OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith11.149.890.2Noctua NH-L9a-AM57.749.390.7OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 386 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 388 Joules per run.
MinAvgMaxWraith50.965.771.3Noctua NH-L9a-AM550.064.369.8OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM570140210280350SE +/- 0.37, N = 3SE +/- 0.36, N = 3298.63298.99

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM5510152025SE +/- 0.03, N = 3SE +/- 0.02, N = 320.0820.05
MinAvgMaxNoctua NH-L9a-AM5300046345439Wraith300046415415OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM58.978.992.1Wraith11.777.892.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3391 Joules per run.2. Wraith: Approximate power consumption of 3425 Joules per run.
MinAvgMaxWraith58.375.385.4Noctua NH-L9a-AM556.873.282.4OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-StreamNoctua NH-L9a-AM5Wraith150300450600750SE +/- 0.17, N = 3SE +/- 0.62, N = 3685.07684.28
MinAvgMaxWraith300047785439Noctua NH-L9a-AM5300047935402OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.474.392.2Wraith10.572.192.7OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3643 Joules per run.2. Wraith: Approximate power consumption of 3653 Joules per run.
MinAvgMaxWraith58.174.278.0Noctua NH-L9a-AM556.572.876.5OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: SlowNoctua NH-L9a-AM5Wraith3691215SE +/- 0.01, N = 3SE +/- 0.01, N = 39.539.54
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: SlowWraithNoctua NH-L9a-AM50.02570.05140.07710.10280.12850.1140.114
MinAvgMaxWraith300045075309Noctua NH-L9a-AM5300045375375OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.883.891.5Noctua NH-L9a-AM510.983.790.4OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 5585 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 5609 Joules per run.
MinAvgMaxWraith49.979.584.8Noctua NH-L9a-AM549.077.782.6OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Scenario: Synchronous Single-StreamNoctua NH-L9a-AM5Wraith306090120150SE +/- 0.06, N = 3SE +/- 0.40, N = 3138.71138.85

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-StreamNoctua NH-L9a-AM5Wraith246810SE +/- 0.0051, N = 3SE +/- 0.0094, N = 38.72758.7362

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Scenario: Synchronous Single-StreamNoctua NH-L9a-AM5Wraith246810SE +/- 0.0033, N = 3SE +/- 0.0206, N = 37.20397.1968
MinAvgMaxNoctua NH-L9a-AM5300048385439Wraith300048595390OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith10.076.493.1Noctua NH-L9a-AM511.574.992.2OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3388 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3394 Joules per run.
MinAvgMaxWraith58.074.077.9Noctua NH-L9a-AM556.171.676.4OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL 15Scaling Factor: 100 - Clients: 1000 - Mode: Read Write - Average LatencyNoctua NH-L9a-AM5Wraith1428425670SE +/- 0.39, N = 3SE +/- 1.02, N = 1264.7964.731. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lm
OpenBenchmarking.orgMegahertz, More Is BetterPostgreSQL 15CPU Peak Freq (Highest CPU Core Frequency) MonitorWraithNoctua NH-L9a-AM59001800270036004500Min: 2988 / Avg: 4648.85 / Max: 5440Min: 2994 / Avg: 4736.81 / Max: 5440
MinAvgMaxNoctua NH-L9a-AM510.759.690.3Wraith11.059.390.7OpenBenchmarking.orgWatts, Fewer Is BetterPostgreSQL 15CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 8419 Joules per run.2. Wraith: Approximate power consumption of 8396 Joules per run.
OpenBenchmarking.orgCelsius, Fewer Is BetterPostgreSQL 15CPU Temperature MonitorWraithNoctua NH-L9a-AM51530456075Min: 52.75 / Avg: 68.33 / Max: 78.13Min: 54.13 / Avg: 67.33 / Max: 77.38

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Super FastNoctua NH-L9a-AM5Wraith4080120160200SE +/- 0.22, N = 9SE +/- 0.12, N = 9177.09177.251. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Super FastWraithNoctua NH-L9a-AM50.8461.6922.5383.3844.233.7523.760
MinAvgMaxWraith300041865196Noctua NH-L9a-AM5359342415249OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith10.747.290.2Noctua NH-L9a-AM57.347.190.7OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 352 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 351 Joules per run.
MinAvgMaxWraith52.566.972.4Noctua NH-L9a-AM551.365.170.4OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: SlowWraithNoctua NH-L9a-AM548121620SE +/- 0.04, N = 3SE +/- 0.05, N = 314.7714.781. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: SlowWraithNoctua NH-L9a-AM50.04140.08280.12420.16560.2070.1820.184
MinAvgMaxNoctua NH-L9a-AM5300045305433Wraith300045395282OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.581.090.2Noctua NH-L9a-AM511.380.390.2OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 3616 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 3585 Joules per run.
MinAvgMaxWraith48.977.683.5Noctua NH-L9a-AM547.875.281.0OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Very FastNoctua NH-L9a-AM5Wraith714212835SE +/- 0.02, N = 3SE +/- 0.05, N = 330.4430.46
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Very FastWraithNoctua NH-L9a-AM50.0920.1840.2760.3680.460.4020.409
MinAvgMaxNoctua NH-L9a-AM5300044065340Wraith370044385213OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxWraith12.075.890.1Noctua NH-L9a-AM59.774.591.0OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 1793 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 1738 Joules per run.
MinAvgMaxWraith58.978.082.1Noctua NH-L9a-AM557.576.180.4OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Kvazaar

This is a test of Kvazaar as a CPU-based H.265/HEVC video encoder written in the C programming language and optimized in Assembly. Kvazaar is the winner of the 2016 ACM Open-Source Software Competition and developed at the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: Super FastWraithNoctua NH-L9a-AM51020304050SE +/- 0.07, N = 4SE +/- 0.05, N = 445.7545.781. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 4K - Video Preset: Super FastWraithNoctua NH-L9a-AM50.14690.29380.44070.58760.73450.6510.653
MinAvgMaxWraith300043985368Noctua NH-L9a-AM5300044145439OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.070.390.2Noctua NH-L9a-AM59.770.191.4OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 1213 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 1210 Joules per run.
MinAvgMaxWraith56.975.580.5Noctua NH-L9a-AM555.073.278.1OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

ClickHouse

ClickHouse is an open-source, high performance OLAP data management system. This test profile uses ClickHouse's standard benchmark recommendations per https://clickhouse.com/docs/en/operations/performance-test/ / https://github.com/ClickHouse/ClickBench/tree/main/clickhouse with the 100 million rows web analytics dataset. The reported value is the query processing time using the geometric mean of all separate queries performed as an aggregate. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgQueries Per Minute, Geo Mean, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, Third RunNoctua NH-L9a-AM5Wraith50100150200250SE +/- 3.31, N = 3SE +/- 1.63, N = 5235.09235.24MIN: 14.22 / MAX: 10000MIN: 14.23 / MAX: 8571.43
OpenBenchmarking.orgQueries Per Minute, Geo Mean Per Watt, More Is BetterClickHouse 22.12.3.5100M Rows Hits Dataset, Third RunNoctua NH-L9a-AM5Wraith0.72921.45842.18762.91683.6463.1273.241
MinAvgMaxNoctua NH-L9a-AM5300050165397Wraith300050315440OpenBenchmarking.orgMegahertz, More Is BetterClickHouse 22.12.3.5CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM58.875.290.2Wraith11.472.690.2OpenBenchmarking.orgWatts, Fewer Is BetterClickHouse 22.12.3.5CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 18068 Joules per run.2. Wraith: Approximate power consumption of 18202 Joules per run.
OpenBenchmarking.orgCelsius, Fewer Is BetterClickHouse 22.12.3.5CPU Temperature MonitorWraithNoctua NH-L9a-AM520406080100Min: 51.13 / Avg: 84.97 / Max: 92.5Min: 50.13 / Avg: 83.27 / Max: 90.13

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM5150300450600750SE +/- 0.53, N = 3SE +/- 0.58, N = 3684.93684.52
MinAvgMaxWraith319747715393Noctua NH-L9a-AM5300047795414OpenBenchmarking.orgMegahertz, More Is BetterNeural Magic DeepSparse 1.3.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.875.092.3Wraith10.273.994.4OpenBenchmarking.orgWatts, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 3677 Joules per run.2. Wraith: Approximate power consumption of 3622 Joules per run.
MinAvgMaxWraith58.375.279.5Noctua NH-L9a-AM557.073.678.0OpenBenchmarking.orgCelsius, Fewer Is BetterNeural Magic DeepSparse 1.3.2CPU Temperature Monitor20406080100

uvg266

uvg266 is an open-source VVC/H.266 (Versatile Video Coding) encoder based on Kvazaar as part of the Ultra Video Group, Tampere University, Finland. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Ultra FastNoctua NH-L9a-AM5Wraith918273645SE +/- 0.05, N = 4SE +/- 0.03, N = 438.4938.51
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 4K - Video Preset: Ultra FastNoctua NH-L9a-AM5Wraith0.11950.2390.35850.4780.59750.5290.531
MinAvgMaxNoctua NH-L9a-AM5356044205339Wraith300044335439OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM511.072.890.5Wraith12.072.691.4OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 1420 Joules per run.2. Wraith: Approximate power consumption of 1433 Joules per run.
MinAvgMaxWraith56.976.181.3Noctua NH-L9a-AM555.174.479.0OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

OpenBenchmarking.orgFrames Per Second, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: Super FastNoctua NH-L9a-AM5Wraith306090120150SE +/- 0.13, N = 8SE +/- 0.11, N = 8133.13133.17
OpenBenchmarking.orgFrames Per Second Per Watt, More Is Betteruvg266 0.4.1Video Input: Bosphorus 1080p - Video Preset: Super FastWraithNoctua NH-L9a-AM50.56881.13761.70642.27522.8442.5232.528
MinAvgMaxWraith300041875349Noctua NH-L9a-AM5300042935388OpenBenchmarking.orgMegahertz, More Is Betteruvg266 0.4.1CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxWraith11.652.890.9Noctua NH-L9a-AM59.452.790.8OpenBenchmarking.orgWatts, Fewer Is Betteruvg266 0.4.1CPU Power Consumption Monitor204060801001. Wraith: Approximate power consumption of 449 Joules per run.2. Noctua NH-L9a-AM5: Approximate power consumption of 448 Joules per run.
MinAvgMaxWraith51.567.172.6Noctua NH-L9a-AM550.865.871.0OpenBenchmarking.orgCelsius, Fewer Is Betteruvg266 0.4.1CPU Temperature Monitor20406080100

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.3.2Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-StreamWraithNoctua NH-L9a-AM5246810SE +/- 0.0075, N = 3SE +/- 0.0069, N = 38.73438.7367

RocksDB

This is a benchmark of Meta/Facebook's RocksDB as an embeddable persistent key-value store for fast storage based on Google's LevelDB. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterRocksDB 7.9.2Test: Random ReadNoctua NH-L9a-AM5Wraith20M40M60M80M100MSE +/- 438183.14, N = 3SE +/- 622147.23, N = 31050810821050902101. (CXX) g++ options: -O3 -march=native -pthread -fno-builtin-memcmp -fno-rtti -lpthread
OpenBenchmarking.orgOp/s Per Watt, More Is BetterRocksDB 7.9.2Test: Random ReadNoctua NH-L9a-AM5Wraith300K600K900K1200K1500K1240296.661243877.82
MinAvgMaxNoctua NH-L9a-AM5300044594605Wraith300045075299OpenBenchmarking.orgMegahertz, More Is BetterRocksDB 7.9.2CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.884.790.1Wraith11.284.591.1OpenBenchmarking.orgWatts, Fewer Is BetterRocksDB 7.9.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5450 Joules per run.2. Wraith: Approximate power consumption of 5435 Joules per run.
MinAvgMaxWraith47.079.585.4Noctua NH-L9a-AM546.577.382.8OpenBenchmarking.orgCelsius, Fewer Is BetterRocksDB 7.9.2CPU Temperature Monitor20406080100

Selenium

This test profile uses the Selenium WebDriver for running various browser benchmarks in different available web browsers such as Mozilla Firefox and Google Chrome. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterSeleniumBenchmark: WASM collisionDetection - Browser: Google ChromeNoctua NH-L9a-AM5Wraith50100150200250SE +/- 0.03, N = 5SE +/- 0.02, N = 5230.82230.811. chrome 109.0.5414.119
MinAvgMaxNoctua NH-L9a-AM5300042985440Wraith300043485440OpenBenchmarking.orgMegahertz, More Is BetterSeleniumCPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.521.339.2Wraith11.020.036.2OpenBenchmarking.orgWatts, Fewer Is BetterSeleniumCPU Power Consumption Monitor11223344551. Noctua NH-L9a-AM5: Approximate power consumption of 333 Joules per run.2. Wraith: Approximate power consumption of 308 Joules per run.
MinAvgMaxNoctua NH-L9a-AM541.953.863.8Wraith39.852.364.6OpenBenchmarking.orgCelsius, Fewer Is BetterSeleniumCPU Temperature Monitor20406080100

OpenVINO

This is a test of the Intel OpenVINO, a toolkit around neural networks, using its built-in benchmarking support and analyzing the throughput and latency for various models. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterOpenVINO 2022.3Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPUNoctua NH-L9a-AM5Wraith0.15750.3150.47250.630.7875SE +/- 0.00, N = 3SE +/- 0.00, N = 30.70.7MIN: 0.39 / MAX: 8.93MIN: 0.35 / MAX: 9.231. (CXX) g++ options: -fPIC -fsigned-char -ffunction-sections -fdata-sections -O3 -fno-strict-overflow -fwrapv -shared
MinAvgMaxWraith300043995083Noctua NH-L9a-AM5300044105267OpenBenchmarking.orgMegahertz, More Is BetterOpenVINO 2022.3CPU Peak Freq (Highest CPU Core Frequency) Monitor13002600390052006500
MinAvgMaxNoctua NH-L9a-AM510.485.392.2Wraith12.085.192.2OpenBenchmarking.orgWatts, Fewer Is BetterOpenVINO 2022.3CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 5430 Joules per run.2. Wraith: Approximate power consumption of 5447 Joules per run.
MinAvgMaxWraith59.877.380.3Noctua NH-L9a-AM557.874.677.4OpenBenchmarking.orgCelsius, Fewer Is BetterOpenVINO 2022.3CPU Temperature Monitor20406080100

CPU Temperature Monitor

OpenBenchmarking.orgCelsiusCPU Temperature MonitorPhoronix Test Suite System MonitoringWraithNoctua NH-L9a-AM520406080100Min: 36.63 / Avg: 73.44 / Max: 92.5Min: 37.13 / Avg: 73.09 / Max: 90.13

CPU Power Consumption Monitor

OpenBenchmarking.orgWattsCPU Power Consumption MonitorPhoronix Test Suite System MonitoringNoctua NH-L9a-AM5Wraith20406080100Min: 5.72 / Avg: 68.56 / Max: 118.2Min: 5.96 / Avg: 65.84 / Max: 116.13

CPU Peak Freq (Highest CPU Core Frequency) Monitor

OpenBenchmarking.orgMegahertzCPU Peak Freq (Highest CPU Core Frequency) MonitorPhoronix Test Suite System MonitoringNoctua NH-L9a-AM5Wraith9001800270036004500Min: 2990 / Avg: 4541.6 / Max: 5440Min: 2971 / Avg: 4519.91 / Max: 5443

Kvazaar

MinAvgMaxWraith54.671.977.0Noctua NH-L9a-AM553.370.275.1OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

MinAvgMaxNoctua NH-L9a-AM510.855.291.4Wraith11.154.991.4OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor20406080100

MinAvgMaxNoctua NH-L9a-AM5300044055359Wraith300044175391OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000

OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Very FastNoctua NH-L9a-AM5Wraith0.47990.95981.43971.91962.39952.0852.133

OpenBenchmarking.orgFrames Per Second, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Very FastNoctua NH-L9a-AM5Wraith306090120150SE +/- 2.13, N = 15SE +/- 2.03, N = 15115.14117.171. (CC) gcc options: -pthread -ftree-vectorize -fvisibility=hidden -O2 -lpthread -lm -lrt
OpenBenchmarking.orgFrames Per Second Per Watt, More Is BetterKvazaar 2.2Video Input: Bosphorus 1080p - Video Preset: Very FastNoctua NH-L9a-AM5Wraith0.47990.95981.43971.91962.39952.0852.133
MinAvgMaxNoctua NH-L9a-AM5300044055359Wraith300044175391OpenBenchmarking.orgMegahertz, More Is BetterKvazaar 2.2CPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM510.855.291.4Wraith11.154.991.4OpenBenchmarking.orgWatts, Fewer Is BetterKvazaar 2.2CPU Power Consumption Monitor204060801001. Noctua NH-L9a-AM5: Approximate power consumption of 508 Joules per run.2. Wraith: Approximate power consumption of 498 Joules per run.
MinAvgMaxWraith54.671.977.0Noctua NH-L9a-AM553.370.275.1OpenBenchmarking.orgCelsius, Fewer Is BetterKvazaar 2.2CPU Temperature Monitor20406080100

PostgreSQL

This is a benchmark of PostgreSQL using the integrated pgbench for facilitating the database benchmarks. Learn more via the OpenBenchmarking.org test page.

Scaling Factor: 100 - Clients: 5000 - Mode: Read Only

Wraith: The test run did not produce a result. E: pgbench: error: need at least 5003 open files, but system limit is 1024

Noctua NH-L9a-AM5: The test run did not produce a result. E: pgbench: error: need at least 5003 open files, but system limit is 1024

Scaling Factor: 100 - Clients: 5000 - Mode: Read Write

Wraith: The test run did not produce a result. E: pgbench: error: need at least 5003 open files, but system limit is 1024

Noctua NH-L9a-AM5: The test run did not produce a result. E: pgbench: error: need at least 5003 open files, but system limit is 1024

OpenEMS

OpenBenchmarking.orgCelsius, Fewer Is BetterOpenEMS 0.0.35-86CPU Temperature MonitorWraithNoctua NH-L9a-AM520406080100Min: 41.13 / Avg: 85.54 / Max: 90.38Min: 42.5 / Avg: 85.02 / Max: 89.5

OpenBenchmarking.orgWatts, Fewer Is BetterOpenEMS 0.0.35-86CPU Power Consumption MonitorNoctua NH-L9a-AM5Wraith20406080100Min: 10.56 / Avg: 83.38 / Max: 92.52Min: 10.74 / Avg: 80.94 / Max: 90.83

OpenBenchmarking.orgMegahertz, More Is BetterOpenEMS 0.0.35-86CPU Peak Freq (Highest CPU Core Frequency) MonitorWraithNoctua NH-L9a-AM59001800270036004500Min: 3000 / Avg: 5104.38 / Max: 5432Min: 2994 / Avg: 5110.21 / Max: 5439

OpenBenchmarking.orgMCells/s Per Watt, More Is BetterOpenEMS 0.0.35-86Test: pyEMS CouplerNoctua NH-L9a-AM5Wraith0.1220.2440.3660.4880.610.5310.542

OpenBenchmarking.orgMCells/s, More Is BetterOpenEMS 0.0.35-86Test: pyEMS CouplerWraithNoctua NH-L9a-AM51020304050SE +/- 1.40, N = 6SE +/- 0.55, N = 943.8344.291. (CXX) g++ options: -O3 -rdynamic -ltinyxml -lcrypto -lcurl -lpthread -lsz -lz -ldl -lm -lexpat
OpenBenchmarking.orgMCells/s Per Watt, More Is BetterOpenEMS 0.0.35-86Test: pyEMS CouplerNoctua NH-L9a-AM5Wraith0.1220.2440.3660.4880.610.5310.542
OpenBenchmarking.orgMegahertz, More Is BetterOpenEMS 0.0.35-86CPU Peak Freq (Highest CPU Core Frequency) MonitorWraithNoctua NH-L9a-AM59001800270036004500Min: 3000 / Avg: 5104.38 / Max: 5432Min: 2994 / Avg: 5110.21 / Max: 5439
OpenBenchmarking.orgWatts, Fewer Is BetterOpenEMS 0.0.35-86CPU Power Consumption MonitorNoctua NH-L9a-AM5Wraith20406080100Min: 10.56 / Avg: 83.38 / Max: 92.52Min: 10.74 / Avg: 80.94 / Max: 90.831. Noctua NH-L9a-AM5: Approximate power consumption of 43348 Joules per run.2. Wraith: Approximate power consumption of 42641 Joules per run.
OpenBenchmarking.orgCelsius, Fewer Is BetterOpenEMS 0.0.35-86CPU Temperature MonitorWraithNoctua NH-L9a-AM520406080100Min: 41.13 / Avg: 85.54 / Max: 90.38Min: 42.5 / Avg: 85.02 / Max: 89.5

Selenium

MinAvgMaxNoctua NH-L9a-AM542.551.861.4Wraith41.550.160.3OpenBenchmarking.orgCelsius, Fewer Is BetterSeleniumCPU Temperature Monitor20406080100

MinAvgMaxNoctua NH-L9a-AM58.515.929.9Wraith10.415.628.5OpenBenchmarking.orgWatts, Fewer Is BetterSeleniumCPU Power Consumption Monitor918273645

MinAvgMaxWraith300038725443Noctua NH-L9a-AM5300039375439OpenBenchmarking.orgMegahertz, More Is BetterSeleniumCPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000

OpenBenchmarking.orgms, Fewer Is BetterSeleniumBenchmark: WASM imageConvolute - Browser: Google ChromeWraithNoctua NH-L9a-AM548121620SE +/- 0.31, N = 15SE +/- 0.16, N = 618.2418.031. chrome 109.0.5414.119
MinAvgMaxWraith300038725443Noctua NH-L9a-AM5300039375439OpenBenchmarking.orgMegahertz, More Is BetterSeleniumCPU Peak Freq (Highest CPU Core Frequency) Monitor14002800420056007000
MinAvgMaxNoctua NH-L9a-AM58.515.929.9Wraith10.415.628.5OpenBenchmarking.orgWatts, Fewer Is BetterSeleniumCPU Power Consumption Monitor9182736451. Noctua NH-L9a-AM5: Approximate power consumption of 164 Joules per run.2. Wraith: Approximate power consumption of 161 Joules per run.
MinAvgMaxNoctua NH-L9a-AM542.551.861.4Wraith41.550.160.3OpenBenchmarking.orgCelsius, Fewer Is BetterSeleniumCPU Temperature Monitor20406080100

Benchmark: PSPDFKit WASM - Browser: Google Chrome

Wraith: The test quit with a non-zero exit status. E: NameError: name 'StaleElementReferenceException' is not defined

Noctua NH-L9a-AM5: The test quit with a non-zero exit status. E: NameError: name 'StaleElementReferenceException' is not defined

139 Results Shown

Neural Magic DeepSparse:
  NLP Text Classification, DistilBERT mnli - Synchronous Single-Stream:
    ms/batch
    items/sec
PostgreSQL:
  100 - 800 - Read Only - Average Latency
  100 - 800 - Read Only
  100 - 100 - Read Only - Average Latency
  100 - 100 - Read Only
OpenVINO
PostgreSQL
ClickHouse
PostgreSQL:
  100 - 500 - Read Only - Average Latency
  100 - 500 - Read Only
  100 - 250 - Read Only - Average Latency
Neural Magic DeepSparse:
  NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Stream:
    items/sec
    ms/batch
ClickHouse
Selenium
RocksDB
OpenVINO
PostgreSQL
BRL-CAD
OpenVINO
oneDNN
PostgreSQL
RocksDB:
  Seq Fill
  Read While Writing
Neural Magic DeepSparse:
  CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Stream:
    items/sec
    ms/batch
  CV Detection, YOLOv5s COCO - Asynchronous Multi-Stream:
    ms/batch
    items/sec
OpenVINO
OpenEMS
OpenVINO
Kvazaar
oneDNN
RocksDB
OpenVINO
oneDNN
OpenVINO
PostgreSQL
RocksDB
OpenVINO:
  Vehicle Detection FP16 - CPU
  Vehicle Detection FP16-INT8 - CPU
PostgreSQL
Neural Magic DeepSparse:
  NLP Text Classification, BERT base uncased SST2 - Asynchronous Multi-Stream:
    ms/batch
    items/sec
OpenVINO
oneDNN
PostgreSQL:
  100 - 800 - Read Write
  100 - 800 - Read Write - Average Latency
Neural Magic DeepSparse:
  CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Stream:
    items/sec
    ms/batch
Kvazaar
OpenVINO
PostgreSQL:
  100 - 250 - Read Write - Average Latency
  100 - 250 - Read Write
Neural Magic DeepSparse:
  CV Classification, ResNet-50 ImageNet - Synchronous Single-Stream:
    ms/batch
    items/sec
OpenVINO
oneDNN
Kvazaar
OpenVINO
Neural Magic DeepSparse
OpenVINO
Neural Magic DeepSparse
OpenVINO
Neural Magic DeepSparse:
  NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Stream:
    ms/batch
    items/sec
  NLP Text Classification, BERT base uncased SST2 - Synchronous Single-Stream:
    ms/batch
    items/sec
oneDNN
OpenVINO
PostgreSQL
OpenVINO:
  Face Detection FP16 - CPU
  Weld Porosity Detection FP16-INT8 - CPU
Neural Magic DeepSparse:
  NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Stream:
    items/sec
    ms/batch
RocksDB
Neural Magic DeepSparse:
  NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Asynchronous Multi-Stream:
    items/sec
    ms/batch
OpenVINO
PostgreSQL
uvg266
Selenium
Kvazaar:
  Bosphorus 4K - Ultra Fast
  Bosphorus 1080p - Ultra Fast
uvg266
Kvazaar
uvg266
Neural Magic DeepSparse
OpenVINO
Neural Magic DeepSparse:
  CV Detection, YOLOv5s COCO - Synchronous Single-Stream
  CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Stream
  CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Stream
uvg266
OpenVINO
PostgreSQL
OpenVINO
oneDNN
OpenVINO
oneDNN
uvg266:
  Bosphorus 4K - Super Fast
  Bosphorus 1080p - Ultra Fast
Neural Magic DeepSparse:
  NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Asynchronous Multi-Stream:
    items/sec
    ms/batch
  NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Stream:
    ms/batch
uvg266
Neural Magic DeepSparse:
  NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Synchronous Single-Stream
  NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Stream
  NLP Sentiment Analysis, 80% Pruned Quantized BERT Base Uncased - Synchronous Single-Stream
PostgreSQL
Kvazaar:
  Bosphorus 1080p - Super Fast
  Bosphorus 4K - Slow
uvg266
Kvazaar
ClickHouse
Neural Magic DeepSparse
uvg266:
  Bosphorus 4K - Ultra Fast
  Bosphorus 1080p - Super Fast
Neural Magic DeepSparse
RocksDB
Selenium
OpenVINO
CPU Temperature Monitor:
  Phoronix Test Suite System Monitoring:
    Celsius
    Watts
    Megahertz
  CPU Temp Monitor:
    Celsius
  CPU Power Consumption Monitor:
    Watts
  CPU Peak Freq (Highest CPU Core Frequency) Monitor:
    Megahertz
  Bosphorus 1080p - Very Fast:
    Frames Per Second Per Watt
Kvazaar
OpenEMS:
  CPU Temp Monitor
  CPU Power Consumption Monitor
  CPU Peak Freq (Highest CPU Core Frequency) Monitor
  pyEMS Coupler
OpenEMS
Selenium:
  CPU Temp Monitor
  CPU Power Consumption Monitor
  CPU Peak Freq (Highest CPU Core Frequency) Monitor
Selenium

Wraith

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203
Python Notes: Python 3.10.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional IBRS_FW STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 29 January 2023 13:04 by user phoronix.

Noctua NH-L9a-AM5

Processor: AMD Ryzen 9 7900 12-Core @ 3.70GHz (12 Cores / 24 Threads), Motherboard: Gigabyte B650M DS3H (F4h BIOS), Chipset: AMD Device 14d8, Memory: 32GB, Disk: 1000GB Sabrent Rocket 4.0 Plus, Graphics: Gigabyte AMD Raphael 512MB (2200/2400MHz), Audio: AMD Rembrandt Radeon HD Audio, Monitor: ASUS VP28U, Network: Realtek RTL8125 2.5GbE

OS: Ubuntu 22.10, Kernel: 6.2.0-060200rc5daily20230129-generic (x86_64), Desktop: GNOME Shell 43.0, Display Server: X Server 1.21.1.4 + Wayland, OpenGL: 4.6 Mesa 23.0.0-devel (git-e20564c 2022-12-12 kinetic-oibaf-ppa) (LLVM 15.0.5 DRM 3.49), Vulkan: 1.3.235, Compiler: GCC 12.2.0, File-System: ext4, Screen Resolution: 3840x2160

Kernel Notes: Transparent Huge Pages: madvise
Compiler Notes: --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-12-U8K4Qv/gcc-12-12.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v
Processor Notes: Scaling Governor: acpi-cpufreq schedutil (Boost: Enabled) - CPU Microcode: 0xa601203
Python Notes: Python 3.10.7
Security Notes: itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Retpolines IBPB: conditional IBRS_FW STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected

Testing initiated at 30 January 2023 15:01 by user phoronix.