GM-1000, Karbon 700, CompuLab Airtop PCs

Tests 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 2011011-FI-GM100056276
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

Limit displaying results to tests within:

AV1 3 Tests
Bioinformatics 2 Tests
BLAS (Basic Linear Algebra Sub-Routine) Tests 2 Tests
C++ Boost Tests 2 Tests
Timed Code Compilation 5 Tests
C/C++ Compiler Tests 21 Tests
Compression Tests 3 Tests
CPU Massive 30 Tests
Creator Workloads 24 Tests
Cryptography 2 Tests
Database Test Suite 5 Tests
Encoding 7 Tests
Fortran Tests 6 Tests
Game Development 3 Tests
HPC - High Performance Computing 24 Tests
Imaging 6 Tests
Java 2 Tests
Common Kernel Benchmarks 6 Tests
Machine Learning 12 Tests
Molecular Dynamics 5 Tests
MPI Benchmarks 5 Tests
Multi-Core 27 Tests
NVIDIA GPU Compute 7 Tests
OCR 2 Tests
Intel oneAPI 3 Tests
OpenCV Tests 2 Tests
OpenMPI Tests 5 Tests
Productivity 2 Tests
Programmer / Developer System Benchmarks 9 Tests
Python 4 Tests
Raytracing 2 Tests
Renderers 3 Tests
Scientific Computing 11 Tests
Server 6 Tests
Server CPU Tests 17 Tests
Single-Threaded 6 Tests
Speech 2 Tests
Telephony 2 Tests
Video Encoding 7 Tests
Vulkan Compute 3 Tests
Common Workstation Benchmarks 3 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

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
Karbon 700
October 30 2020
  22 Hours, 2 Minutes
CompuLab Airtop
October 31 2020
  19 Hours, 53 Minutes
CompuLab Airtop3
October 29 2020
  12 Hours, 25 Minutes
GM-1000
October 28 2020
  18 Hours
Invert Hiding All Results Option
  18 Hours, 5 Minutes

Only show results where is faster than
Only show results matching title/arguments (delimit multiple options with a comma):


GM-1000, Karbon 700, CompuLab Airtop PCsProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerDisplay DriverOpenGLVulkanCompilerFile-SystemScreen ResolutionOpenCLKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000Intel Xeon E-2278GEL @ 3.90GHz (8 Cores / 16 Threads)Logic Supply RXM-181 (Z01-0001A027 BIOS)Intel Cannon Lake PCH16GB512GB TS512GMTE510TIntel UHD P630 3GB (1150MHz)Realtek ALC233VA2431Intel I219-LM + 2 x Intel I210Ubuntu 20.105.8.0-26-generic (x86_64)GNOME Shell 3.38.1X Server 1.20.9modesetting 1.20.94.6 Mesa 20.2.11.2.145GCC 10.2.0ext41920x1080Intel Core i7-5775C @ 3.70GHz (4 Cores / 8 Threads)CompuLab v1.0 (ARTP-3.1.0.637.0.0 X64 BIOS)Intel Broadwell-U DMI256GB ADATA SP310eVGA NVIDIA GeForce GTX 950 2GB (1151/3304MHz)Realtek ALC886G237HLIntel I218-LM + Intel I210 + Intel 7260NVIDIA 455.284.6.0OpenCL 1.2 CUDA 11.1.961.2.142Intel Xeon E-2288G @ 5.00GHz (8 Cores / 16 Threads)Compulab SBC-ATCFL v1.2 (ATOP3.PRD.0.29.2 BIOS)Intel Cannon Lake PCH64GBSamsung SSD 970 EVO Plus 250GBNVIDIA Quadro RTX 4000 8GB (1005/6500MHz)Intel Cannon Lake PCH cAVSVE228Intel I219-LM + Intel I210Intel Core i7-9700TE @ 3.80GHz (8 Cores)C OZE GM-1000 v0.0.01.001 (5.13 BIOS)32GBCIE M8 T405 512GBNVIDIA Quadro P2000 4GB (1455/3003MHz)Realtek ALC888-VDDELL S2409WIntel I219-LM + 2 x Intel 10G X550T + Intel I2105.8.0-25-generic (x86_64)OpenBenchmarking.orgCompiler Details- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,brig,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-targets=nvptx-none=/build/gcc-10-JvwpWM/gcc-10-10.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-10-JvwpWM/gcc-10-10.2.0/debian/tmp-gcn/usr,hsa --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 Disk Details- Karbon 700: NONE / errors=remount-ro,relatime,rw- CompuLab Airtop: MQ-DEADLINE / errors=remount-ro,relatime,rw- CompuLab Airtop3: NONE / errors=remount-ro,relatime,rw- GM-1000: NONE / errors=remount-ro,relatime,rwProcessor Details- Karbon 700: Scaling Governor: intel_pstate powersave - CPU Microcode: 0xd6 - Thermald 2.3- CompuLab Airtop: Scaling Governor: intel_cpufreq ondemand - CPU Microcode: 0x22 - Thermald 2.3- CompuLab Airtop3: Scaling Governor: intel_pstate powersave - CPU Microcode: 0xd6 - Thermald 2.3- GM-1000: Scaling Governor: intel_pstate powersave - CPU Microcode: 0xd6 - Thermald 2.3Java Details- Karbon 700: OpenJDK Runtime Environment (build 11.0.9+11-Ubuntu-0ubuntu1)- CompuLab Airtop: OpenJDK Runtime Environment (build 11.0.9+11-Ubuntu-0ubuntu1)- CompuLab Airtop3: OpenJDK Runtime Environment (build 11.0.9+11-Ubuntu-0ubuntu1)- GM-1000: OpenJDK Runtime Environment (build 11.0.9+10-post-Ubuntu-0ubuntu1)Python Details- Python 3.8.6Security Details- Karbon 700: itlb_multihit: KVM: Mitigation of VMX unsupported + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced IBRS IBPB: conditional RSB filling + srbds: Mitigation of TSX disabled + tsx_async_abort: Mitigation of TSX disabled - CompuLab Airtop: itlb_multihit: KVM: Mitigation of VMX disabled + l1tf: Mitigation of PTE Inversion; VMX: conditional cache flushes SMT vulnerable + mds: Mitigation of Clear buffers; SMT vulnerable + meltdown: Mitigation of PTI + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Full generic retpoline IBPB: conditional IBRS_FW STIBP: conditional RSB filling + srbds: Mitigation of Microcode + tsx_async_abort: Mitigation of Clear buffers; SMT vulnerable - CompuLab Airtop3: itlb_multihit: KVM: Mitigation of VMX disabled + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced IBRS IBPB: conditional RSB filling + srbds: Mitigation of TSX disabled + tsx_async_abort: Mitigation of TSX disabled - GM-1000: itlb_multihit: KVM: Mitigation of VMX disabled + l1tf: Not affected + mds: Not affected + meltdown: Not affected + spec_store_bypass: Mitigation of SSB disabled via prctl and seccomp + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced IBRS IBPB: conditional RSB filling + srbds: Mitigation of TSX disabled + tsx_async_abort: Mitigation of TSX disabled OpenCL Details- CompuLab Airtop: GPU Compute Cores: 768- CompuLab Airtop3: GPU Compute Cores: 2304- GM-1000: GPU Compute Cores: 768

Karbon 700CompuLab AirtopCompuLab Airtop3GM-1000Logarithmic Result OverviewPhoronix Test Suite 10.6.1KripkeIORNCNNApache CassandraLAMMPS Molecular Dynamics SimulatorIncompact3DOpenVKLLeelaChessZeroFacebook RocksDBOpenSSLrays1benchTensorFlow LiteNAMDJohn The RipperKvazaarNumenta Anomaly BenchmarkYafaRayTimed Linux Kernel CompilationTimed LLVM CompilationKeyDBIntel Open Image Denoise7-Zip CompressionSVT-AV1LuxCoreRenderSVT-VP9PostgreSQL pgbenchZstd CompressionRealSR-NCNNWaifu2x-NCNN Vulkanx265OCRMyPDFdav1doneDNNTimed GDB GNU Debugger Compilationlibavif avifencTimed PHP CompilationASTC EncoderHigh Performance Conjugate GradientGROMACSFFTETimed Apache CompilationMonte Carlo Simulations of Ionised NebulaeVP9 libvpx EncodingWireGuard + Linux Networking Stack Stress TestPostMarkAI Benchmark AlphaHuginDaCapo BenchmarkRawTherapeeLevelDBeSpeak-NG Speech EngineC-BloscMlpack BenchmarkNumpy BenchmarklibrsvgTimed MAFFT AlignmentMobile Neural NetworkPyPerformanceJava Gradle BuildTesseract OCRDolfynRNNoiseBYTE Unix BenchmarkGitWebP Image EncodeGNU Octave BenchmarkTNNTimed HMMer SearchHierarchical INTegrationLibRawCaffe

GM-1000, Karbon 700, CompuLab Airtop PCsncnn: Vulkan GPU - yolov4-tinyncnn: Vulkan GPU - squeezenetncnn: Vulkan GPU - googlenetglmark2: 1920 x 1080ncnn: Vulkan GPU - efficientnet-b0ncnn: Vulkan GPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU-v2-v2 - mobilenet-v2ncnn: Vulkan GPU - mobilenetior: Read Testonednn: Deconvolution Batch deconv_3d - u8s8f32 - CPUcassandra: Writeswaifu2x-ncnn: 2x - 3 - Norealsr-ncnn: 4x - Yesincompact3d: Cylinderleveldb: Seq Fillonednn: IP Batch All - u8s8f32 - CPUpgbench: 1 - 50 - Read Onlypgbench: 1 - 50 - Read Only - Average Latencyopenvkl: vklBenchmarkonednn: Deconvolution Batch deconv_3d - f32 - CPUlczero: BLASlczero: Eigenjohn-the-ripper: Blowfishleveldb: Hot Readtensorflow-lite: Inception ResNet V2tensorflow-lite: Inception V4tensorflow-lite: SqueezeNettensorflow-lite: Mobilenet Floatrocksdb: Read While Writingrocksdb: Seq Fillonednn: Deconvolution Batch deconv_1d - u8s8f32 - CPUrays1bench: Large Scenerocksdb: Rand Readbrl-cad: VGR Performance Metricastcenc: Thoroughonednn: Matrix Multiply Batch Shapes Transformer - u8s8f32 - CPUastcenc: Exhaustivenumenta-nab: Relative Entropysvt-av1: Enc Mode 4 - 1080pkvazaar: Bosphorus 4K - Ultra Fasttensorflow-lite: Mobilenet Quantnamd: ATPase Simulation - 327,506 Atomsnumenta-nab: Earthgecko Skylinekvazaar: Bosphorus 4K - Very Fastnumenta-nab: Bayesian Changepointcompress-zstd: 3numenta-nab: Windowed Gaussianbuild-linux-kernel: Time To Compilekvazaar: Bosphorus 1080p - Mediumyafaray: Total Time For Sample Scenekvazaar: Bosphorus 1080p - Very Fastkeydb: build-llvm: Time To Compilekvazaar: Bosphorus 4K - Mediumoidn: Memorialcompress-7zip: Compress Speed Testdacapobench: Tradesoapluxcorerender: DLSCdav1d: Chimera 1080pluxcorerender: Rainbow Colors and Prismdav1d: Summer Nature 1080pjohn-the-ripper: MD5onednn: Recurrent Neural Network Training - f32 - CPUcompress-zstd: 19avifenc: 0avifenc: 2x265: Bosphorus 4Kncnn: Vulkan GPU - vgg16realsr-ncnn: 4x - Norocksdb: Rand Fill Syncncnn: Vulkan GPU - resnet18waifu2x-ncnn: 2x - 3 - Yesncnn: Vulkan GPU - resnet50ncnn: Vulkan GPU - alexnetocrmypdf: Processing 60 Page PDF Documentsvt-av1: Enc Mode 0 - 1080px265: Bosphorus 1080ptensorflow-lite: NASNet Mobileonednn: Convolution Batch Shapes Auto - f32 - CPUonednn: Convolution Batch Shapes Auto - u8s8f32 - CPUdav1d: Summer Nature 4Kbuild-gdb: Time To Compilebuild-php: Time To Compileai-benchmark: Device Inference Scoremlpack: scikit_linearridgeregressiondav1d: Chimera 1080p 10-bitncnn: CPU - vgg16onednn: Matrix Multiply Batch Shapes Transformer - f32 - CPUhpcg: avifenc: 8gromacs: Water Benchmarkonednn: Recurrent Neural Network Inference - f32 - CPUffte: N=256, 3D Complex FFT Routineonednn: IP Batch All - f32 - CPUbuild-apache: Time To Compilevpxenc: Speed 0avifenc: 10mocassin: Dust 2D tau100.0mnn: SqueezeNetV1.0wireguard: leveldb: Rand Deletencnn: CPU - alexnetpostmark: Disk Transaction Performancevpxenc: Speed 5ncnn: CPU - resnet50hugin: Panorama Photo Assistant + Stitching Timeai-benchmark: Device AI Scoreleveldb: Seq Fillrawtherapee: Total Benchmark Timeai-benchmark: Device Training Scorepyperformance: pathlibncnn: CPU - mobilenetncnn: CPU - resnet18pgbench: 1 - 1 - Read Onlyncnn: CPU - squeezenetblosc: blosclzastcenc: Fastpgbench: 1 - 1 - Read Only - Average Latencymlpack: scikit_icanumpy: ncnn: CPU-v2-v2 - mobilenet-v2pyperformance: floatpyperformance: python_startupncnn: CPU - yolov4-tinydacapobench: Jythonrsvg: SVG Files To PNGcaffe: AlexNet - CPU - 100pyperformance: crypto_pyaescaffe: AlexNet - CPU - 200pyperformance: pickle_pure_pythonmafft: Multiple Sequence Alignment - LSU RNAmnn: inception-v3caffe: GoogleNet - CPU - 100pyperformance: raytracecaffe: GoogleNet - CPU - 200pyperformance: 2to3pyperformance: regex_compilepyperformance: django_templatemnn: resnet-v2-50pyperformance: chaospyperformance: gowebp: Defaultjava-gradle-perf: Reactortesseract-ocr: Time To OCR 7 Imagespyperformance: nbodydolfyn: Computational Fluid Dynamicsmlpack: scikit_svmwebp: Quality 100, Lossless, Highest Compressionrnnoise: byte: Dhrystone 2webp: Quality 100tnn: CPU - SqueezeNet v1.1webp: Quality 100, Highest Compressiongit: Time To Complete Common Git Commandspyperformance: json_loadsoctave-benchmark: webp: Quality 100, Losslesshmmer: Pfam Database Searchhint: FLOATtnn: CPU - MobileNet v2libraw: Post-Processing Benchmarkdacapobench: H2kripke: mlpack: scikit_qdarocksdb: Rand Fillncnn: Vulkan GPU - blazefacencnn: Vulkan GPU - mnasnetncnn: Vulkan GPU - shufflenet-v2ncnn: CPU - googlenetastcenc: Mediumpgbench: 1 - 50 - Read Write - Average Latencypgbench: 1 - 50 - Read Writepgbench: 1 - 1 - Read Write - Average Latencypgbench: 1 - 1 - Read Writeleveldb: Seek Randleveldb: Rand Readleveldb: Rand Fillleveldb: Rand Fillleveldb: Overwriteleveldb: Overwriteleveldb: Fill Syncleveldb: Fill Syncopenssl: RSA 4096-bit Performanceespeak: Text-To-Speech Synthesissvt-vp9: Visual Quality Optimized - Bosphorus 1080psvt-vp9: PSNR/SSIM Optimized - Bosphorus 1080pkvazaar: Bosphorus 1080p - Ultra Fastonednn: Deconvolution Batch deconv_1d - f32 - CPUonednn: IP Batch 1D - u8s8f32 - CPUonednn: IP Batch 1D - f32 - CPUdacapobench: Tradebeanslammps: Rhodopsin Proteinior: Write TestKarbon 700CompuLab AirtopCompuLab Airtop3GM-100065.8727.5823.5452416.4710.198.1726.19517.184.634955012410.7721764.690512.73032640.541.12752266750.22198.397.886713353061122710.18039526604371977298283202055152245910335417.7204831.933153797246.093.62375384.6629.8702.11516.002051262.80992192.6168.6171.5561609.016.805134.66913.81246.29634.69570969.821051.6003.035.273313551721.05450.531.15432.27594105539.07612.5134.32577.8149.00151.46223.462917523.8579.33550.2225.9228.1900.10441.4426200732.003230.8841100.79117.55283.9257453.9487.55118.127.045872.314996.3540.517251.01217775.746492801108.94326.4286.355.9992659.170207.69244.01323.87694422.8743.4958.573143043.65685.42968518.227.5724.172654423.506843.67.230.03860.81334.997.821148.5242.36488122.5645473310911125043412.44562.30814637947929423833217849.359.9831092551.673256.35025.65113121.02527.2543.02527.46338072622.22.647343.2218.13356.44927.18.56919.972131.201371786567.42537369.08627.1232675741904130.535263582.419.2610.1022.406.2067.1687561.14594113.67310.22954.60733.155.25032.91309.9681.41931.631.67596.13112.7167.536.618032.555646.2734239565.116652.8820.0112.358.4837185.703.733.1810.74232.4511.2904201043.292405.7741024.4666319.873.58281119150.44753.7517.536778473762334.1567087053783620754273236703686406655730213.237620.17186250434486972.575.48320597.5051.0131.2909.743551854.44112282.7965.3382.8842511.427.591217.4088.63391.49621.55346852.161666.6061.983.502267378030.67307.340.73277.97399692605.87023.7203.068120.4616.1033.7447.2706246.1719.92714.098.7042.8320.06827.9138831920.398719.580182.58173.464118.9446154.4763.5181.924.323192.710618.9830.466167.47919794.106635888116.97836.0644.528.25933110.775255.81543.79523.75500017.2146.6075.990135144.66399.21973623.027.0624.242109223.056601.68.680.04760.10276.247.0313610.238.90542726.7195759612511534248912.31563.20414721153829427637519655.254.8201212791.823251.68827.89214022.54921.4245.70925.33136078067.22.804292.5858.44959.82928.08.60121.025117.867370080894.79875301.68228.323456371669476.4646445913.322.5920.3911.14156.8663193.3203015.4794.15243.38520.443.32620.410317.6820.11048.437.78663.9980.9139.0010.29305.623218.6432849442.48877.358.503.633.222.641.661.434.591052.003.313566782081.525378.57991552.827.74572965450.169141.936.65640840765152726.83629461273260697226512153645192337313215665.6421546.114346559610056633.232.56113274.4922.2152.94322.211564961.96010132.85412.0543.0792955.512.45498.50919.02177.73147.21751370.01769.6624.267.534875236921.41644.691.53582.07836454322.59425.799.60959.41212.368.3112.62623441.685.6013.752.1121.6870.13454.9119970817.024816.5236157.4191.59164.68911282.47114.5866.283.964934.090165.0860.819143.88830993.63518867867.690520.9107.774.8052026.581160.38733.73814.23833328.6128.5145.839223633.54560.833110814.319.0015.193355416.2010397.85.530.0342.14420.295.1589.96.8229.06368918.2203930985.9795823398.62844.44410286937620662226413838.942.58185.92001.313196.07320.27110316.67822.4334.79621.58248614938.02.082269.2756.38044.96721.46.48516.022100.192471438878.81023283.96033.3529282126409362.688717380.611.481.315.008.5283.7315972.3724348.9377.66336.18349.036.68748.36472.0620.32539.923.705131.83169.1291.465.046731.889894.4973827916.786253.1716.2810.277.8434425.233.393.009.31476.156.69517207323.214276.624706.78615332.567.76541283730.39080.398.8837178475876544.8545441300601118040636927839380873698109112.820319.74246163314313876.985.91239631.6949.5021.44610.382795244.17868300.5445.6297.1353601.726.115212.6489.01374.75423.54458524.511604.0591.993.702411066800.67354.510.75344.07430834670.34123.3190.816110.7376.5333.3337.87534735.8016.86813.228.1739.6080.07431.4232124616.622416.123692.95155.395114.7966403.1476.1783.074.385023.708316.6720.533231.38630890.906674731114.29132.1155.096.28334411.117269.93226.18722.53657819.6547.3375.788137227.20495.48173219.330.3323.562486225.548600.47.930.04064.72322.076.611229.0143.08483324.631442451168942446211.26863.68212614050525531935218652.351.8871172701.759271.42827.10813822.14528.9346.98229.10036200967.72.780362.1308.57260.37228.48.00721.197131.248360309999.31612369.90827.2633392088958387.816347391.043.102.3823.999.9168.8967261.9885035.9284.74828.45931.128.73330.82260.1460.41492.733.37780.06107.7944.4510.809214.840059.5332436214.870461.45OpenBenchmarking.org

NCNN

NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: yolov4-tinyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10001530456075SE +/- 0.28, N = 3SE +/- 0.03, N = 3SE +/- 0.19, N = 3SE +/- 0.01, N = 365.8720.018.5016.28MIN: 50.88 / MAX: 68.67MIN: 15.5 / MAX: 21.72MIN: 8.02 / MAX: 9.19MIN: 15.64 / MAX: 17.641. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU - Model: yolov4-tinyKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000110220330440550526.9680.0468.00130.241. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU - Model: yolov4-tinyKarbon 700CompuLab AirtopCompuLab Airtop3GM-100020040060080010001053.92160.08136.00130.241. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU - Model: yolov4-tinyKarbon 700CompuLab AirtopCompuLab Airtop3GM-100060120180240300256.8974.0442.5061.861. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: yolov4-tinyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10001326395265Min: 65.32 / Avg: 65.87 / Max: 66.23Min: 19.96 / Avg: 20.01 / Max: 20.06Min: 8.31 / Avg: 8.5 / Max: 8.88Min: 16.26 / Avg: 16.28 / Max: 16.291. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: squeezenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000612182430SE +/- 0.17, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 327.5812.353.6310.27MIN: 27.17 / MAX: 31.88MIN: 11.58 / MAX: 15.04MIN: 3.57 / MAX: 3.76MIN: 9.98 / MAX: 10.781. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU - Model: squeezenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-100050100150200250220.6449.4029.0482.161. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU - Model: squeezenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000100200300400500441.2898.8058.0882.161. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU - Model: squeezenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-100020406080100107.5645.7018.1539.031. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: squeezenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000612182430Min: 27.36 / Avg: 27.58 / Max: 27.92Min: 12.33 / Avg: 12.35 / Max: 12.38Min: 3.61 / Avg: 3.63 / Max: 3.64Min: 10.26 / Avg: 10.27 / Max: 10.271. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: googlenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000612182430SE +/- 0.02, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.03, N = 323.548.483.227.84MIN: 23.24 / MAX: 23.83MIN: 8.04 / MAX: 10.85MIN: 3.2 / MAX: 3.32MIN: 7.62 / MAX: 10.981. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU - Model: googlenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-10004080120160200188.3233.9225.7662.721. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU - Model: googlenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-100080160240320400376.6467.8451.5262.721. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU - Model: googlenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002040608010091.8131.3816.1029.791. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: googlenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000510152025Min: 23.5 / Avg: 23.54 / Max: 23.56Min: 8.47 / Avg: 8.48 / Max: 8.48Min: 3.21 / Avg: 3.22 / Max: 3.22Min: 7.79 / Avg: 7.84 / Max: 7.881. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

GLmark2

This is a test of Linaro's glmark2 port, currently using the X11 OpenGL 2.0 target. GLmark2 is a basic OpenGL benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgScore, More Is BetterGLmark2 2020.04Resolution: 1920 x 1080Karbon 700CompuLab AirtopGM-1000800160024003200400052437183442

NCNN

NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: efficientnet-b0Karbon 700CompuLab AirtopCompuLab Airtop3GM-100048121620SE +/- 0.06, N = 3SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.02, N = 316.475.702.645.23MIN: 16.27 / MAX: 16.89MIN: 5.57 / MAX: 6.86MIN: 2.61 / MAX: 2.67MIN: 5.06 / MAX: 5.951. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU - Model: efficientnet-b0Karbon 700CompuLab AirtopCompuLab Airtop3GM-1000306090120150131.7622.8021.1241.841. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU - Model: efficientnet-b0Karbon 700CompuLab AirtopCompuLab Airtop3GM-100060120180240300263.5245.6042.2441.841. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU - Model: efficientnet-b0Karbon 700CompuLab AirtopCompuLab Airtop3GM-1000142842567064.2321.0913.2019.871. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: efficientnet-b0Karbon 700CompuLab AirtopCompuLab Airtop3GM-100048121620Min: 16.36 / Avg: 16.47 / Max: 16.55Min: 5.69 / Avg: 5.7 / Max: 5.71Min: 2.63 / Avg: 2.64 / Max: 2.64Min: 5.18 / Avg: 5.23 / Max: 5.261. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3Karbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215SE +/- 0.22, N = 3SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.03, N = 310.193.731.663.39MIN: 9.74 / MAX: 10.79MIN: 3.65 / MAX: 4.87MIN: 1.65 / MAX: 1.69MIN: 3.29 / MAX: 4.221. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3Karbon 700CompuLab AirtopCompuLab Airtop3GM-10002040608010081.5214.9213.2827.121. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3Karbon 700CompuLab AirtopCompuLab Airtop3GM-10004080120160200163.0429.8426.5627.121. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3Karbon 700CompuLab AirtopCompuLab Airtop3GM-100091827364539.74113.8018.30012.8821. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3Karbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215Min: 9.77 / Avg: 10.19 / Max: 10.49Min: 3.72 / Avg: 3.73 / Max: 3.73Min: 1.66 / Avg: 1.66 / Max: 1.67Min: 3.34 / Avg: 3.39 / Max: 3.421. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2Karbon 700CompuLab AirtopCompuLab Airtop3GM-1000246810SE +/- 0.07, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 3SE +/- 0.03, N = 38.173.181.433.00MIN: 7.83 / MAX: 8.44MIN: 3.11 / MAX: 4.26MIN: 1.4 / MAX: 9.39MIN: 2.89 / MAX: 3.841. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2Karbon 700CompuLab AirtopCompuLab Airtop3GM-1000153045607565.3612.7211.4424.001. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2Karbon 700CompuLab AirtopCompuLab Airtop3GM-1000306090120150130.7225.4422.8824.001. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2Karbon 700CompuLab AirtopCompuLab Airtop3GM-100071421283531.86311.7667.15011.4001. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2Karbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215Min: 8.03 / Avg: 8.17 / Max: 8.26Min: 3.17 / Avg: 3.18 / Max: 3.18Min: 1.41 / Avg: 1.43 / Max: 1.45Min: 2.95 / Avg: 3 / Max: 3.041. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: mobilenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000612182430SE +/- 0.29, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.00, N = 326.1910.744.599.31MIN: 23.7 / MAX: 35.85MIN: 9.23 / MAX: 12.04MIN: 4.54 / MAX: 12.63MIN: 9 / MAX: 10.281. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread
OpenBenchmarking.orgms x Core, Fewer Is BetterNCNN 20200916Performance Per Core - Target: Vulkan GPU - Model: mobilenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-100050100150200250209.5242.9636.7274.481. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterNCNN 20200916Performance Per Thread - Target: Vulkan GPU - Model: mobilenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-100090180270360450419.0485.9273.4474.481. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterNCNN 20200916Performance Per Clock - Target: Vulkan GPU - Model: mobilenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-100020406080100102.1439.7422.9535.381. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterNCNN 20200916Target: Vulkan GPU - Model: mobilenetKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000612182430Min: 25.63 / Avg: 26.19 / Max: 26.61Min: 10.72 / Avg: 10.74 / Max: 10.75Min: 4.57 / Avg: 4.59 / Max: 4.63Min: 9.31 / Avg: 9.31 / Max: 9.321. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

IOR

IOR is a parallel I/O storage benchmark. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMB/s, More Is BetterIOR 3.2.1Read TestKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002004006008001000SE +/- 0.53, N = 15SE +/- 0.82, N = 11SE +/- 4.11, N = 13SE +/- 4.42, N = 15517.18232.451052.00476.15MIN: 142.77 / MAX: 594.11MIN: 207.97 / MAX: 242.21MIN: 53.05 / MAX: 1439MIN: 10.65 / MAX: 548.821. (CC) gcc options: -O2 -lm -pthread -lmpi
OpenBenchmarking.orgMB/s, More Is BetterIOR 3.2.1Read TestKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002004006008001000Min: 511.99 / Avg: 517.18 / Max: 520.15Min: 228.43 / Avg: 232.45 / Max: 236.65Min: 1034.39 / Avg: 1052 / Max: 1069.98Min: 443.63 / Avg: 476.15 / Max: 499.021. (CC) gcc options: -O2 -lm -pthread -lmpi

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 oneAPI initiative. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 1.5Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215SE +/- 0.00916, N = 9SE +/- 0.00833, N = 9SE +/- 0.02023, N = 9SE +/- 0.01402, N = 94.6349511.290403.313566.69517MIN: 4.53MIN: 11.22MIN: 3.15MIN: 6.491. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -pie -lpthread -ldl
OpenBenchmarking.orgms x Core, Fewer Is BetteroneDNN 1.5Performance Per Core - Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000122436486037.0845.1626.5153.561. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetteroneDNN 1.5Performance Per Thread - Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002040608010074.1690.3253.0253.561. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetteroneDNN 1.5Performance Per Clock - Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000102030405018.0841.7716.5725.441. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 1.5Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215Min: 4.6 / Avg: 4.63 / Max: 4.67Min: 11.27 / Avg: 11.29 / Max: 11.35Min: 3.23 / Avg: 3.31 / Max: 3.45Min: 6.64 / Avg: 6.7 / Max: 6.741. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -pie -lpthread -ldl

Apache Cassandra

This is a benchmark of the Apache Cassandra NoSQL database management system making use of cassandra-stress. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgOp/s, More Is BetterApache Cassandra 3.11.4Test: WritesKarbon 700CompuLab AirtopCompuLab Airtop3GM-100015K30K45K60K75KSE +/- 581.94, N = 6SE +/- 299.90, N = 4SE +/- 744.79, N = 3SE +/- 177.03, N = 1550124201046782020732
OpenBenchmarking.orgOp/s Per Core, More Is BetterApache Cassandra 3.11.4Performance Per Core - Test: WritesKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002K4K6K8K10K6265.55026.08477.52591.51. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgOp/s Per Thread, More Is BetterApache Cassandra 3.11.4Performance Per Thread - Test: WritesKarbon 700CompuLab AirtopCompuLab Airtop3GM-100090018002700360045003132.752513.004238.752591.501. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgOp/s Per GHz, More Is BetterApache Cassandra 3.11.4Performance Per Clock - Test: WritesKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003K6K9K12K15K12852.315433.5113564.005455.791. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgOp/s, More Is BetterApache Cassandra 3.11.4Test: WritesKarbon 700CompuLab AirtopCompuLab Airtop3GM-100012K24K36K48K60KMin: 49081 / Avg: 50123.67 / Max: 52804Min: 19581 / Avg: 20104 / Max: 20743Min: 66893 / Avg: 67819.67 / Max: 69293Min: 19586 / Avg: 20732.33 / Max: 22089

Waifu2x-NCNN Vulkan

Waifu2x-NCNN is an NCNN neural network implementation of the Waifu2x converter project and accelerated using the Vulkan API. NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. This test profile times how long it takes to increase the resolution of a sample image with Vulkan. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterWaifu2x-NCNN Vulkan 20200818Scale: 2x - Denoise: 3 - TAA: NoKarbon 700CompuLab AirtopGM-10003691215SE +/- 0.003, N = 5SE +/- 0.030, N = 10SE +/- 0.007, N = 910.7723.2923.214
OpenBenchmarking.orgSeconds, Fewer Is BetterWaifu2x-NCNN Vulkan 20200818Scale: 2x - Denoise: 3 - TAA: NoKarbon 700CompuLab AirtopGM-10003691215Min: 10.77 / Avg: 10.77 / Max: 10.78Min: 3.25 / Avg: 3.29 / Max: 3.55Min: 3.19 / Avg: 3.21 / Max: 3.25

RealSR-NCNN

RealSR-NCNN is an NCNN neural network implementation of the RealSR project and accelerated using the Vulkan API. RealSR is the Real-World Super Resolution via Kernel Estimation and Noise Injection. NCNN is a high performance neural network inference framework optimized for mobile and other platforms developed by Tencent. This test profile times how long it takes to increase the resolution of a sample image by a scale of 4x with Vulkan. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterRealSR-NCNN 20200818Scale: 4x - TAA: YesKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000400800120016002000SE +/- 0.20, N = 3SE +/- 6.42, N = 9SE +/- 0.44, N = 3SE +/- 0.07, N = 31764.69405.7781.53276.62
OpenBenchmarking.orgSeconds, Fewer Is BetterRealSR-NCNN 20200818Scale: 4x - TAA: YesKarbon 700CompuLab AirtopCompuLab Airtop3GM-100030060090012001500Min: 1764.31 / Avg: 1764.69 / Max: 1764.95Min: 362.31 / Avg: 405.77 / Max: 418.21Min: 80.72 / Avg: 81.53 / Max: 82.23Min: 276.49 / Avg: 276.62 / Max: 276.75

Incompact3D

Incompact3d is a Fortran-MPI based, finite difference high-performance code for solving the incompressible Navier-Stokes equation and as many as you need scalar transport equations. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgSeconds, Fewer Is BetterIncompact3D 2020-09-17Input: CylinderKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002004006008001000SE +/- 1.81, N = 3SE +/- 4.69, N = 3SE +/- 1.24, N = 3SE +/- 1.81, N = 3512.731024.47378.58706.791. (F9X) gfortran options: -cpp -funroll-loops -floop-optimize -fcray-pointer -fbacktrace -pthread -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -ldl -levent -levent_pthreads -lutil -lm -lrt -lz
OpenBenchmarking.orgSeconds x Core, Fewer Is BetterIncompact3D 2020-09-17Performance Per Core - Input: CylinderKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000120024003600480060004101.844097.873028.645654.291. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgSeconds x Thread, Fewer Is BetterIncompact3D 2020-09-17Performance Per Thread - Input: CylinderKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002K4K6K8K10K8203.698195.736057.285654.291. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgSeconds x GHz, Fewer Is BetterIncompact3D 2020-09-17Performance Per Clock - Input: CylinderKarbon 700CompuLab AirtopCompuLab Airtop3GM-100080016002400320040001999.653790.531892.902685.791. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgSeconds, Fewer Is BetterIncompact3D 2020-09-17Input: CylinderKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002004006008001000Min: 509.11 / Avg: 512.73 / Max: 514.73Min: 1017.2 / Avg: 1024.47 / Max: 1033.23Min: 376.18 / Avg: 378.58 / Max: 380.34Min: 703.74 / Avg: 706.79 / Max: 710.011. (F9X) gfortran options: -cpp -funroll-loops -floop-optimize -fcray-pointer -fbacktrace -pthread -lmpi_usempif08 -lmpi_mpifh -lmpi -lopen-rte -lopen-pal -lhwloc -ldl -levent -levent_pthreads -lutil -lm -lrt -lz

LevelDB

LevelDB is a key-value storage library developed by Google that supports making use of Snappy for data compression and has other modern features. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMB/s, More Is BetterLevelDB 1.22Benchmark: Sequential FillKarbon 700CompuLab AirtopCompuLab Airtop3GM-10001224364860SE +/- 0.15, N = 3SE +/- 0.07, N = 3SE +/- 0.78, N = 3SE +/- 0.41, N = 540.519.852.832.51. (CXX) g++ options: -O3 -lsnappy -lpthread
OpenBenchmarking.orgMB/s Per Core, More Is BetterLevelDB 1.22Performance Per Core - Benchmark: Sequential FillKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002468105.064.956.604.061. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgMB/s Per Thread, More Is BetterLevelDB 1.22Performance Per Thread - Benchmark: Sequential FillKarbon 700CompuLab AirtopCompuLab Airtop3GM-10000.91351.8272.74053.6544.56752.532.483.304.061. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgMB/s Per GHz, More Is BetterLevelDB 1.22Performance Per Clock - Benchmark: Sequential FillKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000369121510.385.3510.568.551. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgMB/s, More Is BetterLevelDB 1.22Benchmark: Sequential FillKarbon 700CompuLab AirtopCompuLab Airtop3GM-10001122334455Min: 40.3 / Avg: 40.53 / Max: 40.8Min: 19.7 / Avg: 19.83 / Max: 19.9Min: 51.4 / Avg: 52.77 / Max: 54.1Min: 31.4 / Avg: 32.52 / Max: 33.81. (CXX) g++ options: -O3 -lsnappy -lpthread

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 oneAPI initiative. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 1.5Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-10001632486480SE +/- 0.05, N = 3SE +/- 0.24, N = 3SE +/- 0.06, N = 3SE +/- 0.33, N = 341.1373.5827.7567.77MIN: 35.58MIN: 72.88MIN: 25.65MIN: 59.541. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -pie -lpthread -ldl
OpenBenchmarking.orgms x Core, Fewer Is BetteroneDNN 1.5Performance Per Core - Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000120240360480600329.02294.33221.97542.121. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetteroneDNN 1.5Performance Per Thread - Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000140280420560700658.04588.66443.93542.121. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetteroneDNN 1.5Performance Per Clock - Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-100060120180240300160.40272.26138.73257.511. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 1.5Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-10001428425670Min: 41.04 / Avg: 41.13 / Max: 41.2Min: 73.17 / Avg: 73.58 / Max: 74Min: 27.68 / Avg: 27.75 / Max: 27.86Min: 67.23 / Avg: 67.77 / Max: 68.361. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -pie -lpthread -ldl

PostgreSQL pgbench

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

OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 13.0Scaling Factor: 1 - Clients: 50 - Mode: Read OnlyKarbon 700CompuLab AirtopCompuLab Airtop3GM-100060K120K180K240K300KSE +/- 470.82, N = 3SE +/- 1589.17, N = 3SE +/- 2332.06, N = 3SE +/- 1251.67, N = 152266751119152965451283731. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm
OpenBenchmarking.orgTPS Per Core, More Is BetterPostgreSQL pgbench 13.0Performance Per Core - Scaling Factor: 1 - Clients: 50 - Mode: Read OnlyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10008K16K24K32K40K28334.3827978.7537068.1316046.631. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgTPS Per Thread, More Is BetterPostgreSQL pgbench 13.0Performance Per Thread - Scaling Factor: 1 - Clients: 50 - Mode: Read OnlyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10004K8K12K16K20K14167.1913989.3818534.0616046.631. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgTPS Per GHz, More Is BetterPostgreSQL pgbench 13.0Performance Per Clock - Scaling Factor: 1 - Clients: 50 - Mode: Read OnlyKarbon 700CompuLab AirtopCompuLab Airtop3GM-100013K26K39K52K65K58121.7930247.3059309.0033782.371. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgTPS, More Is BetterPostgreSQL pgbench 13.0Scaling Factor: 1 - Clients: 50 - Mode: Read OnlyKarbon 700CompuLab AirtopCompuLab Airtop3GM-100050K100K150K200K250KMin: 226050.55 / Avg: 226674.84 / Max: 227597.49Min: 109589.91 / Avg: 111915.28 / Max: 114954.37Min: 292821.58 / Avg: 296544.71 / Max: 300839.21Min: 125521.71 / Avg: 128373.27 / Max: 144975.271. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL pgbench 13.0Scaling Factor: 1 - Clients: 50 - Mode: Read Only - Average LatencyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10000.10060.20120.30180.40240.503SE +/- 0.000, N = 3SE +/- 0.006, N = 3SE +/- 0.001, N = 3SE +/- 0.003, N = 150.2210.4470.1690.3901. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm
OpenBenchmarking.orgms x Core, Fewer Is BetterPostgreSQL pgbench 13.0Performance Per Core - Scaling Factor: 1 - Clients: 50 - Mode: Read Only - Average LatencyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10000.7021.4042.1062.8083.511.7681.7881.3523.1201. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetterPostgreSQL pgbench 13.0Performance Per Thread - Scaling Factor: 1 - Clients: 50 - Mode: Read Only - Average LatencyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10000.80461.60922.41383.21844.0233.5363.5762.7043.1201. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetterPostgreSQL pgbench 13.0Performance Per Clock - Scaling Factor: 1 - Clients: 50 - Mode: Read Only - Average LatencyKarbon 700CompuLab AirtopCompuLab Airtop3GM-10000.37220.74441.11661.48881.8610.8621.6540.8451.4821. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetterPostgreSQL pgbench 13.0Scaling Factor: 1 - Clients: 50 - Mode: Read Only - Average LatencyKarbon 700CompuLab AirtopCompuLab Airtop3GM-100012345Min: 0.22 / Avg: 0.22 / Max: 0.22Min: 0.44 / Avg: 0.45 / Max: 0.46Min: 0.17 / Avg: 0.17 / Max: 0.17Min: 0.35 / Avg: 0.39 / Max: 0.41. (CC) gcc options: -fno-strict-aliasing -fwrapv -O2 -lpgcommon -lpgport -lpq -lpthread -lrt -ldl -lm

OpenVKL

OpenVKL is the Intel Open Volume Kernel Library that offers high-performance volume computation kernels and part of the Intel oneAPI rendering toolkit. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgItems / Sec, More Is BetterOpenVKL 0.9Benchmark: vklBenchmarkKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000306090120150SE +/- 0.29, N = 3SE +/- 0.05, N = 3SE +/- 0.98, N = 3SE +/- 0.39, N = 398.3953.75141.9380.39MIN: 1 / MAX: 378MIN: 1 / MAX: 212MIN: 1 / MAX: 580MIN: 1 / MAX: 345
OpenBenchmarking.orgItems / Sec Per Core, More Is BetterOpenVKL 0.9Performance Per Core - Benchmark: vklBenchmarkKarbon 700CompuLab AirtopCompuLab Airtop3GM-10004812162012.3013.4417.7410.051. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgItems / Sec Per Thread, More Is BetterOpenVKL 0.9Performance Per Thread - Benchmark: vklBenchmarkKarbon 700CompuLab AirtopCompuLab Airtop3GM-100036912156.156.728.8710.051. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgItems / Sec Per GHz, More Is BetterOpenVKL 0.9Performance Per Clock - Benchmark: vklBenchmarkKarbon 700CompuLab AirtopCompuLab Airtop3GM-100071421283525.2314.5328.3921.161. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgItems / Sec, More Is BetterOpenVKL 0.9Benchmark: vklBenchmarkKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000306090120150Min: 97.92 / Avg: 98.39 / Max: 98.92Min: 53.67 / Avg: 53.75 / Max: 53.83Min: 140.83 / Avg: 141.93 / Max: 143.89Min: 79.75 / Avg: 80.39 / Max: 81.08

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 oneAPI initiative. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgms, Fewer Is BetteroneDNN 1.5Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-100048121620SE +/- 0.02235, N = 9SE +/- 0.04063, N = 9SE +/- 0.02344, N = 9SE +/- 0.08092, N = 157.8867117.536706.656408.88371MIN: 7.8MIN: 17.19MIN: 6.4MIN: 8.111. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -pie -lpthread -ldl
OpenBenchmarking.orgms x Core, Fewer Is BetteroneDNN 1.5Performance Per Core - Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000163248648063.0970.1553.2571.071. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgms x Thread, Fewer Is BetteroneDNN 1.5Performance Per Thread - Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000306090120150126.19140.29106.5071.071. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgms x GHz, Fewer Is BetteroneDNN 1.5Performance Per Clock - Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000142842567030.7664.8933.2833.761. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgms, Fewer Is BetteroneDNN 1.5Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPUKarbon 700CompuLab AirtopCompuLab Airtop3GM-100048121620Min: 7.83 / Avg: 7.89 / Max: 8.06Min: 17.38 / Avg: 17.54 / Max: 17.76Min: 6.6 / Avg: 6.66 / Max: 6.82Min: 8.34 / Avg: 8.88 / Max: 9.441. (CXX) g++ options: -O3 -march=native -std=c++11 -fopenmp -msse4.1 -fPIC -pie -lpthread -ldl

LeelaChessZero

LeelaChessZero (lc0 / lczero) is a chess engine automated vian neural networks. This test profile can be used for OpenCL, CUDA + cuDNN, and BLAS (CPU-based) benchmarking. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.26Backend: BLASKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002004006008001000SE +/- 10.93, N = 3SE +/- 8.54, N = 3SE +/- 6.44, N = 33357848407841. (CXX) g++ options: -flto -pthread
OpenBenchmarking.orgNodes Per Second Per Core, More Is BetterLeelaChessZero 0.26Performance Per Core - Backend: BLASKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000408012016020041.88196.00105.0098.001. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgNodes Per Second Per Thread, More Is BetterLeelaChessZero 0.26Performance Per Thread - Backend: BLASKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002040608010020.9498.0052.5098.001. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgNodes Per Second Per GHz, More Is BetterLeelaChessZero 0.26Performance Per Clock - Backend: BLASKarbon 700CompuLab AirtopCompuLab Airtop3GM-10005010015020025085.90211.89168.00206.321. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.26Backend: BLASKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000150300450600750Min: 762 / Avg: 783.67 / Max: 797Min: 823 / Avg: 840 / Max: 850Min: 771 / Avg: 783.67 / Max: 7921. (CXX) g++ options: -flto -pthread

OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.26Backend: EigenKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000160320480640800SE +/- 4.33, N = 3SE +/- 3.38, N = 3SE +/- 7.36, N = 3SE +/- 2.52, N = 33067377657581. (CXX) g++ options: -flto -pthread
OpenBenchmarking.orgNodes Per Second Per Core, More Is BetterLeelaChessZero 0.26Performance Per Core - Backend: EigenKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000408012016020038.25184.2595.6394.751. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgNodes Per Second Per Thread, More Is BetterLeelaChessZero 0.26Performance Per Thread - Backend: EigenKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002040608010019.1392.1347.8194.751. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgNodes Per Second Per GHz, More Is BetterLeelaChessZero 0.26Performance Per Clock - Backend: EigenKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000408012016020078.46199.19153.00199.471. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgNodes Per Second, More Is BetterLeelaChessZero 0.26Backend: EigenKarbon 700CompuLab AirtopCompuLab Airtop3GM-1000130260390520650Min: 299 / Avg: 306.33 / Max: 314Min: 733 / Avg: 737.33 / Max: 744Min: 757 / Avg: 765.33 / Max: 780Min: 755 / Avg: 758 / Max: 7631. (CXX) g++ options: -flto -pthread

John The Ripper

This is a benchmark of John The Ripper, which is a password cracker. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 1.9.0-jumbo-1Test: BlowfishKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003K6K9K12K15KSE +/- 103.18, N = 10SE +/- 227.62, N = 4SE +/- 60.95, N = 13112276233152727654-lbz21. (CC) gcc options: -m64 -lssl -lcrypto -fopenmp -pthread -lm -lz -ldl -lcrypt
OpenBenchmarking.orgReal C/S Per Core, More Is BetterJohn The Ripper 1.9.0-jumbo-1Performance Per Core - Test: BlowfishKarbon 700CompuLab AirtopCompuLab Airtop3GM-10004008001200160020001403.381558.251909.00956.751. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgReal C/S Per Thread, More Is BetterJohn The Ripper 1.9.0-jumbo-1Performance Per Thread - Test: BlowfishKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002004006008001000701.69779.13954.50956.751. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgReal C/S Per GHz, More Is BetterJohn The Ripper 1.9.0-jumbo-1Performance Per Clock - Test: BlowfishKarbon 700CompuLab AirtopCompuLab Airtop3GM-100070014002100280035002878.721684.593054.402014.211. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgReal C/S, More Is BetterJohn The Ripper 1.9.0-jumbo-1Test: BlowfishKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003K6K9K12K15KMin: 11074 / Avg: 11227.1 / Max: 12154Min: 15019 / Avg: 15271.75 / Max: 15954Min: 7579 / Avg: 7653.77 / Max: 83851. (CC) gcc options: -m64 -lssl -lcrypto -fopenmp -pthread -lm -lz -ldl -lcrypt

LevelDB

LevelDB is a key-value storage library developed by Google that supports making use of Snappy for data compression and has other modern features. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMicroseconds Per Op, Fewer Is BetterLevelDB 1.22Benchmark: Hot ReadKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215SE +/- 0.133, N = 5SE +/- 0.008, N = 8SE +/- 0.037, N = 6SE +/- 0.045, N = 1510.1804.1566.8364.8541. (CXX) g++ options: -O3 -lsnappy -lpthread
OpenBenchmarking.orgMicroseconds Per Op x Core, Fewer Is BetterLevelDB 1.22Performance Per Core - Benchmark: Hot ReadKarbon 700CompuLab AirtopCompuLab Airtop3GM-10002040608010081.4416.6254.6938.831. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgMicroseconds Per Op x Thread, Fewer Is BetterLevelDB 1.22Performance Per Thread - Benchmark: Hot ReadKarbon 700CompuLab AirtopCompuLab Airtop3GM-10004080120160200162.8833.25109.3838.831. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgMicroseconds Per Op x GHz, Fewer Is BetterLevelDB 1.22Performance Per Clock - Benchmark: Hot ReadKarbon 700CompuLab AirtopCompuLab Airtop3GM-100091827364539.7015.3834.1818.451. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgMicroseconds Per Op, Fewer Is BetterLevelDB 1.22Benchmark: Hot ReadKarbon 700CompuLab AirtopCompuLab Airtop3GM-10003691215Min: 9.89 / Avg: 10.18 / Max: 10.6Min: 4.11 / Avg: 4.16 / Max: 4.18Min: 6.73 / Avg: 6.84 / Max: 7Min: 4.59 / Avg: 4.85 / Max: 5.131. (CXX) g++ options: -O3 -lsnappy -lpthread

TensorFlow Lite

This is a benchmark of the TensorFlow Lite implementation. The current Linux support is limited to running on CPUs. This test profile is measuring the average inference time. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2020-08-23Model: Inception ResNet V2Karbon 700CompuLab AirtopCompuLab Airtop3GM-10001.5M3M4.5M6M7.5MSE +/- 17665.88, N = 3SE +/- 615.40, N = 3SE +/- 1078.35, N = 3SE +/- 40177.63, N = 33952660708705329461275441300
OpenBenchmarking.orgMicroseconds x Core, Fewer Is BetterTensorFlow Lite 2020-08-23Performance Per Core - Model: Inception ResNet V2Karbon 700CompuLab AirtopCompuLab Airtop3GM-10009M18M27M36M45M316212802834821223569016435304001. Karbon 700: Detected core count of 82. CompuLab Airtop: Detected core count of 43. CompuLab Airtop3: Detected core count of 84. GM-1000: Detected core count of 8
OpenBenchmarking.orgMicroseconds x Thread, Fewer Is BetterTensorFlow Lite 2020-08-23Performance Per Thread - Model: Inception ResNet V2Karbon 700CompuLab AirtopCompuLab Airtop3GM-100014M28M42M56M70M632425605669642447138032435304001. Karbon 700: Detected thread count of 162. CompuLab Airtop: Detected thread count of 83. CompuLab Airtop3: Detected thread count of 164. GM-1000: Detected thread count of 8
OpenBenchmarking.orgMicroseconds x GHz, Fewer Is BetterTensorFlow Lite 2020-08-23Performance Per Clock - Model: Inception ResNet V2Karbon 700CompuLab AirtopCompuLab Airtop3GM-10006M12M18M24M30M15415374.026222096.114730635.020676940.01. Karbon 700: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.902. CompuLab Airtop: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.70 3. CompuLab Airtop3: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 5.00 4. GM-1000: Detected GHz base clock speed (use PTS sensors for real-time frequency/sensor reporting) count of 3.80
OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2020-08-23Model: Inception ResNet V2Karbon 700CompuLab AirtopCompuLab Airtop3GM-10001.2M2.4M3.6M4.8M6MMin: 3917330 / Avg: 3952660 / Max: 3970630Min: 7086240 / Avg: 7087053.33 / Max: 7088260Min: 2943980 / Avg: 2946126.67 / Max: 2947380Min: 5360970 / Avg: 5441300 / Max: 5483210

OpenBenchmarking.orgMicroseconds, Fewer Is BetterTensorFlow Lite 2020-08-23Model: Inception V4Karbon 700CompuLab AirtopCompuLab Airtop3GM-10002M4M6M8M10MSE +/- 20514.24, N = 3SE +/- 604.55, N = 3SE +/- 1363.50, N = 3SE +/- 34182.46, N = 34371977783620732606976011180