AMD Ryzen zen4 Linux
Benchmarks for a future article.
HTML result view exported from: https://openbenchmarking.org/result/2301094-PTS-EXTRANEW01&grs&rdt.
oneDNN
Harness: IP Shapes 3D - Data Type: u8s8f32 - Engine: CPU
NCNN
Target: CPU - Model: blazeface
Mobile Neural Network
Model: mobilenet-v1-1.0
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPU
NCNN
Target: CPU - Model: FastestDet
NCNN
Target: CPU - Model: shufflenet-v2
ONNX Runtime
Model: bertsquad-12 - Device: CPU - Executor: Standard
ONNX Runtime
Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard
NCNN
Target: CPU - Model: regnety_400m
ONNX Runtime
Model: super-resolution-10 - Device: CPU - Executor: Standard
NCNN
Target: CPU-v3-v3 - Model: mobilenet-v3
Mobile Neural Network
Model: mobilenetV3
NCNN
Target: CPU - Model: mnasnet
Mobile Neural Network
Model: squeezenetv1.1
Mobile Neural Network
Model: nasnet
Mobile Neural Network
Model: MobileNetV2_224
C-Ray
Total Time - 4K, 16 Rays Per Pixel
NAS Parallel Benchmarks
Test / Class: EP.C
Stockfish
Total Time
Mobile Neural Network
Model: resnet-v2-50
OpenSSL
Algorithm: RSA4096
Zstd Compression
Compression Level: 8 - Compression Speed
OpenSSL
Algorithm: RSA4096
Cpuminer-Opt
Algorithm: Deepcoin
NAS Parallel Benchmarks
Test / Class: EP.D
oneDNN
Harness: Deconvolution Batch shapes_1d - Data Type: bf16bf16bf16 - Engine: CPU
Cpuminer-Opt
Algorithm: Magi
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: bf16bf16bf16 - Engine: CPU
OpenSSL
Algorithm: SHA256
NCNN
Target: CPU-v2-v2 - Model: mobilenet-v2
JPEG XL Decoding libjxl
CPU Threads: All
Cpuminer-Opt
Algorithm: Quad SHA-256, Pyrite
OpenVINO
Model: Vehicle Detection FP16 - Device: CPU
Coremark
CoreMark Size 666 - Iterations Per Second
Cpuminer-Opt
Algorithm: x25x
Cpuminer-Opt
Algorithm: scrypt
Cpuminer-Opt
Algorithm: Ringcoin
7-Zip Compression
Test: Decompression Rating
Cpuminer-Opt
Algorithm: Blake-2 S
Neural Magic DeepSparse
Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream
Cpuminer-Opt
Algorithm: LBC, LBRY Credits
IndigoBench
Acceleration: CPU - Scene: Supercar
Neural Magic DeepSparse
Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream
Neural Magic DeepSparse
Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream
Cpuminer-Opt
Algorithm: Triple SHA-256, Onecoin
Neural Magic DeepSparse
Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream
Cpuminer-Opt
Algorithm: Skeincoin
Stargate Digital Audio Workstation
Sample Rate: 44100 - Buffer Size: 1024
Xmrig
Variant: Wownero - Hash Count: 1M
Blender
Blend File: Barbershop - Compute: CPU-Only
IndigoBench
Acceleration: CPU - Scene: Bedroom
Tachyon
Total Time
Blender
Blend File: BMW27 - Compute: CPU-Only
Blender
Blend File: Classroom - Compute: CPU-Only
Xmrig
Variant: Monero - Hash Count: 1M
Stargate Digital Audio Workstation
Sample Rate: 480000 - Buffer Size: 1024
asmFish
1024 Hash Memory, 26 Depth
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: bf16bf16bf16 - Engine: CPU
ASTC Encoder
Preset: Exhaustive
OpenVINO
Model: Face Detection FP16-INT8 - Device: CPU
Chaos Group V-RAY
Mode: CPU
ASTC Encoder
Preset: Fast
ASTC Encoder
Preset: Medium
Stargate Digital Audio Workstation
Sample Rate: 44100 - Buffer Size: 512
ASTC Encoder
Preset: Thorough
ONNX Runtime
Model: fcn-resnet101-11 - Device: CPU - Executor: Standard
Neural Magic DeepSparse
Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Asynchronous Multi-Stream
OpenVINO
Model: Weld Porosity Detection FP16 - Device: CPU
Stargate Digital Audio Workstation
Sample Rate: 96000 - Buffer Size: 1024
Timed Linux Kernel Compilation
Build: allmodconfig
Appleseed
Scene: Disney Material
Aircrack-ng
Stargate Digital Audio Workstation
Sample Rate: 192000 - Buffer Size: 1024
Liquid-DSP
Threads: 24 - Buffer Length: 256 - Filter Length: 57
Blender
Blend File: Pabellon Barcelona - Compute: CPU-Only
OpenVINO
Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPU
OpenVINO
Model: Weld Porosity Detection FP16-INT8 - Device: CPU
Blender
Blend File: Fishy Cat - Compute: CPU-Only
NAMD
ATPase Simulation - 327,506 Atoms
Liquid-DSP
Threads: 16 - Buffer Length: 256 - Filter Length: 57
oneDNN
Harness: Recurrent Neural Network Training - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: IP Shapes 3D - Data Type: bf16bf16bf16 - Engine: CPU
Timed LLVM Compilation
Build System: Ninja
Stargate Digital Audio Workstation
Sample Rate: 480000 - Buffer Size: 512
oneDNN
Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPU
oneDNN
Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPU
OpenVINO
Model: Face Detection FP16 - Device: CPU
OpenVINO
Model: Vehicle Detection FP16-INT8 - Device: CPU
Neural Magic DeepSparse
Model: CV Detection,YOLOv5s COCO - Scenario: Asynchronous Multi-Stream
SVT-HEVC
Tuning: 1 - Input: Bosphorus 4K
Rodinia
Test: OpenMP LavaMD
x265
Video Input: Bosphorus 4K
OpenVINO
Model: Person Detection FP16 - Device: CPU
Stargate Digital Audio Workstation
Sample Rate: 96000 - Buffer Size: 512
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPU
Appleseed
Scene: Emily
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPU
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: u8s8f32 - Engine: CPU
LAMMPS Molecular Dynamics Simulator
Model: Rhodopsin Protein
SVT-HEVC
Tuning: 1 - Input: Bosphorus 1080p
Neural Magic DeepSparse
Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Asynchronous Multi-Stream
Stargate Digital Audio Workstation
Sample Rate: 192000 - Buffer Size: 512
NAS Parallel Benchmarks
Test / Class: SP.B
Neural Magic DeepSparse
Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-Stream
Neural Magic DeepSparse
Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-Stream
OpenVINO
Model: Person Detection FP32 - Device: CPU
Neural Magic DeepSparse
Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-Stream
Neural Magic DeepSparse
Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-Stream
NCNN
Target: CPU - Model: efficientnet-b0
OpenVINO
Model: Age Gender Recognition Retail 0013 FP16 - Device: CPU
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: Deconvolution Batch shapes_1d - Data Type: u8s8f32 - Engine: CPU
Timed MPlayer Compilation
Time To Compile
Primesieve
Length: 1e13
Timed LLVM Compilation
Build System: Unix Makefiles
Primesieve
Length: 1e12
Timed Linux Kernel Compilation
Build: defconfig
Build2
Time To Compile
Rodinia
Test: OpenMP CFD Solver
NAS Parallel Benchmarks
Test / Class: BT.C
7-Zip Compression
Test: Compression Rating
GROMACS
Implementation: MPI CPU - Input: water_GMX50_bare
OpenVINO
Model: Machine Translation EN To DE FP16 - Device: CPU
SVT-VP9
Tuning: Visual Quality Optimized - Input: Bosphorus 1080p
PyPerformance
Benchmark: python_startup
Timed Godot Game Engine Compilation
Time To Compile
Timed FFmpeg Compilation
Time To Compile
SVT-HEVC
Tuning: 7 - Input: Bosphorus 4K
oneDNN
Harness: IP Shapes 1D - Data Type: f32 - Engine: CPU
GNU Radio
Test: Five Back to Back FIR Filters
SVT-VP9
Tuning: Visual Quality Optimized - Input: Bosphorus 4K
ONNX Runtime
Model: yolov4 - Device: CPU - Executor: Standard
Mobile Neural Network
Model: SqueezeNetV1.0
SVT-HEVC
Tuning: 7 - Input: Bosphorus 1080p
Timed Mesa Compilation
Time To Compile
Mobile Neural Network
Model: inception-v3
OpenVINO
Model: Person Vehicle Bike Detection FP16 - Device: CPU
libavif avifenc
Encoder Speed: 6
Cpuminer-Opt
Algorithm: Myriad-Groestl
x264
Video Input: Bosphorus 4K
oneDNN
Harness: Deconvolution Batch shapes_3d - Data Type: f32 - Engine: CPU
oneDNN
Harness: Deconvolution Batch shapes_1d - Data Type: f32 - Engine: CPU
OpenFOAM
Input: drivaerFastback, Small Mesh Size - Execution Time
SVT-VP9
Tuning: VMAF Optimized - Input: Bosphorus 1080p
SVT-VP9
Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p
Neural Magic DeepSparse
Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-Stream
Neural Magic DeepSparse
Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-Stream
x264
Video Input: Bosphorus 1080p
oneDNN
Harness: Deconvolution Batch shapes_3d - Data Type: u8s8f32 - Engine: CPU
Appleseed
Scene: Material Tester
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Medium
SVT-HEVC
Tuning: 10 - Input: Bosphorus 1080p
oneDNN
Harness: Deconvolution Batch shapes_3d - Data Type: bf16bf16bf16 - Engine: CPU
Neural Magic DeepSparse
Model: CV Detection,YOLOv5s COCO - Scenario: Synchronous Single-Stream
Neural Magic DeepSparse
Model: CV Detection,YOLOv5s COCO - Scenario: Synchronous Single-Stream
NCNN
Target: CPU - Model: googlenet
ONNX Runtime
Model: ArcFace ResNet-100 - Device: CPU - Executor: Parallel
SVT-AV1
Encoder Mode: Preset 8 - Input: Bosphorus 4K
Darktable
Test: Boat - Acceleration: CPU-only
SVT-HEVC
Tuning: 10 - Input: Bosphorus 4K
libavif avifenc
Encoder Speed: 0
Zstd Compression
Compression Level: 19 - Compression Speed
SVT-AV1
Encoder Mode: Preset 8 - Input: Bosphorus 1080p
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Very Fast
Neural Magic DeepSparse
Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Synchronous Single-Stream
Neural Magic DeepSparse
Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Synchronous Single-Stream
SVT-VP9
Tuning: VMAF Optimized - Input: Bosphorus 4K
Rodinia
Test: OpenMP Leukocyte
SVT-AV1
Encoder Mode: Preset 4 - Input: Bosphorus 4K
libavif avifenc
Encoder Speed: 6, Lossless
SVT-VP9
Tuning: PSNR/SSIM Optimized - Input: Bosphorus 4K
Neural Magic DeepSparse
Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-Stream
Neural Magic DeepSparse
Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-Stream
Rodinia
Test: OpenMP Streamcluster
ONNX Runtime
Model: super-resolution-10 - Device: CPU - Executor: Parallel
oneDNN
Harness: IP Shapes 3D - Data Type: f32 - Engine: CPU
Zstd Compression
Compression Level: 19, Long Mode - Compression Speed
SVT-AV1
Encoder Mode: Preset 12 - Input: Bosphorus 4K
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Ultra Fast
libavif avifenc
Encoder Speed: 2
Y-Cruncher
Pi Digits To Calculate: 1B
Timed PHP Compilation
Time To Compile
GPAW
Input: Carbon Nanotube
VP9 libvpx Encoding
Speed: Speed 5 - Input: Bosphorus 4K
Kvazaar
Video Input: Bosphorus 1080p - Video Preset: Medium
Darktable
Test: Server Rack - Acceleration: CPU-only
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: u8s8f32 - Engine: CPU
NAS Parallel Benchmarks
Test / Class: LU.C
ONNX Runtime
Model: bertsquad-12 - Device: CPU - Executor: Parallel
SVT-AV1
Encoder Mode: Preset 13 - Input: Bosphorus 4K
Timed Wasmer Compilation
Time To Compile
Darktable
Test: Masskrug - Acceleration: CPU-only
Xcompact3d Incompact3d
Input: input.i3d 129 Cells Per Direction
Y-Cruncher
Pi Digits To Calculate: 500M
oneDNN
Harness: IP Shapes 1D - Data Type: bf16bf16bf16 - Engine: CPU
GIMP
Test: resize
NAS Parallel Benchmarks
Test / Class: FT.C
NAS Parallel Benchmarks
Test / Class: CG.C
ONNX Runtime
Model: fcn-resnet101-11 - Device: CPU - Executor: Parallel
Kvazaar
Video Input: Bosphorus 1080p - Video Preset: Ultra Fast
Timed GDB GNU Debugger Compilation
Time To Compile
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPU
OpenVINO
Model: Machine Translation EN To DE FP16 - Device: CPU
VP9 libvpx Encoding
Speed: Speed 0 - Input: Bosphorus 1080p
VP9 libvpx Encoding
Speed: Speed 0 - Input: Bosphorus 4K
Cpuminer-Opt
Algorithm: Garlicoin
OpenFOAM
Input: drivaerFastback, Small Mesh Size - Mesh Time
OpenVINO
Model: Vehicle Detection FP16 - Device: CPU
nekRS
Input: TurboPipe Periodic
NCNN
Target: CPU - Model: mobilenet
Liquid-DSP
Threads: 8 - Buffer Length: 256 - Filter Length: 57
OpenVINO
Model: Face Detection FP16-INT8 - Device: CPU
SVT-AV1
Encoder Mode: Preset 4 - Input: Bosphorus 1080p
SVT-AV1
Encoder Mode: Preset 12 - Input: Bosphorus 1080p
Zstd Compression
Compression Level: 3 - Compression Speed
ONNX Runtime
Model: yolov4 - Device: CPU - Executor: Parallel
OpenVINO
Model: Vehicle Detection FP16-INT8 - Device: CPU
OpenVINO
Model: Weld Porosity Detection FP16 - Device: CPU
Kvazaar
Video Input: Bosphorus 1080p - Video Preset: Very Fast
Neural Magic DeepSparse
Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Synchronous Single-Stream
OpenVINO
Model: Face Detection FP16 - Device: CPU
Neural Magic DeepSparse
Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Synchronous Single-Stream
Darktable
Test: Server Room - Acceleration: CPU-only
NCNN
Target: CPU - Model: squeezenet_ssd
OpenVINO
Model: Person Detection FP16 - Device: CPU
NCNN
Target: CPU - Model: resnet18
Rodinia
Test: OpenMP HotSpot3D
SVT-AV1
Encoder Mode: Preset 13 - Input: Bosphorus 1080p
oneDNN
Harness: IP Shapes 1D - Data Type: u8s8f32 - Engine: CPU
OpenVINO
Model: Person Detection FP32 - Device: CPU
Timed CPython Compilation
Build Configuration: Default
x265
Video Input: Bosphorus 1080p
libavif avifenc
Encoder Speed: 10, Lossless
Zstd Compression
Compression Level: 8, Long Mode - Compression Speed
NCNN
Target: CPU - Model: yolov4-tiny
DaCapo Benchmark
Java Test: Tradebeans
NCNN
Target: CPU - Model: resnet50
NCNN
Target: CPU - Model: vision_transformer
GIMP
Test: unsharp-mask
GNU Radio
Test: Signal Source (Cosine)
OpenVINO
Model: Person Vehicle Bike Detection FP16 - Device: CPU
NAS Parallel Benchmarks
Test / Class: MG.C
OpenVINO
Model: Age Gender Recognition Retail 0013 FP16 - Device: CPU
DaCapo Benchmark
Java Test: Tradesoap
Neural Magic DeepSparse
Model: NLP Question Answering, BERT base uncased SQuaD 12layer Pruned90 - Scenario: Asynchronous Multi-Stream
ONNX Runtime
Model: GPT-2 - Device: CPU - Executor: Standard
NAS Parallel Benchmarks
Test / Class: SP.C
Neural Magic DeepSparse
Model: CV Detection,YOLOv5s COCO - Scenario: Asynchronous Multi-Stream
VP9 libvpx Encoding
Speed: Speed 5 - Input: Bosphorus 1080p
OpenVINO
Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPU
OpenVINO
Model: Weld Porosity Detection FP16-INT8 - Device: CPU
Neural Magic DeepSparse
Model: NLP Text Classification, BERT base uncased SST2 - Scenario: Asynchronous Multi-Stream
NAS Parallel Benchmarks
Test / Class: IS.D
TNN
Target: CPU - Model: SqueezeNet v2
DaCapo Benchmark
Java Test: H2
Neural Magic DeepSparse
Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream
WebP Image Encode
Encode Settings: Default
Neural Magic DeepSparse
Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream
JPEG XL libjxl
Input: JPEG - Quality: 100
Algebraic Multi-Grid Benchmark
Timed Apache Compilation
Time To Compile
Neural Magic DeepSparse
Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream
LeelaChessZero
Backend: Eigen
Neural Magic DeepSparse
Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream
JPEG XL libjxl
Input: JPEG - Quality: 90
JPEG XL libjxl
Input: PNG - Quality: 90
JPEG XL libjxl
Input: JPEG - Quality: 80
NCNN
Target: CPU - Model: vgg16
GIMP
Test: auto-levels
JPEG XL libjxl
Input: PNG - Quality: 80
simdjson
Throughput Test: TopTweet
Ngspice
Circuit: C7552
ONNX Runtime
Model: GPT-2 - Device: CPU - Executor: Parallel
Timed CPython Compilation
Build Configuration: Released Build, PGO + LTO Optimized
DaCapo Benchmark
Java Test: Jython
Zstd Compression
Compression Level: 3, Long Mode - Compression Speed
NCNN
Target: CPU - Model: alexnet
PyPerformance
Benchmark: crypto_pyaes
GNU Radio
Test: FIR Filter
GNU Radio
Test: Hilbert Transform
TNN
Target: CPU - Model: DenseNet
Zstd Compression
Compression Level: 8, Long Mode - Decompression Speed
Liquid-DSP
Threads: 2 - Buffer Length: 256 - Filter Length: 57
Zstd Compression
Compression Level: 3, Long Mode - Decompression Speed
Ngspice
Circuit: C2670
PyPerformance
Benchmark: nbody
PHPBench
PHP Benchmark Suite
WebP Image Encode
Encode Settings: Quality 100, Lossless
Zstd Compression
Compression Level: 19 - Decompression Speed
Zstd Compression
Compression Level: 8 - Decompression Speed
LeelaChessZero
Backend: BLAS
JPEG XL Decoding libjxl
CPU Threads: 1
Zstd Compression
Compression Level: 19, Long Mode - Decompression Speed
simdjson
Throughput Test: DistinctUserID
Liquid-DSP
Threads: 1 - Buffer Length: 256 - Filter Length: 57
QuantLib
PyPerformance
Benchmark: go
GNU Radio
Test: IIR Filter
PyPerformance
Benchmark: float
WebP Image Encode
Encode Settings: Quality 100, Lossless, Highest Compression
Crafty
Elapsed Time
WebP Image Encode
Encode Settings: Quality 100, Highest Compression
GIMP
Test: rotate
PyPerformance
Benchmark: 2to3
PyPerformance
Benchmark: pathlib
TNN
Target: CPU - Model: SqueezeNet v1.1
GNU Radio
Test: FM Deemphasis Filter
PyBench
Total For Average Test Times
Liquid-DSP
Threads: 4 - Buffer Length: 256 - Filter Length: 57
WebP Image Encode
Encode Settings: Quality 100
Git
Time To Complete Common Git Commands
PyPerformance
Benchmark: regex_compile
PyPerformance
Benchmark: django_template
PyPerformance
Benchmark: raytrace
JPEG XL libjxl
Input: PNG - Quality: 100
simdjson
Throughput Test: LargeRandom
PyPerformance
Benchmark: json_loads
PyPerformance
Benchmark: pickle_pure_python
simdjson
Throughput Test: PartialTweets
TNN
Target: CPU - Model: MobileNet v2
FLAC Audio Encoding
WAV To FLAC
PyPerformance
Benchmark: chaos
Zstd Compression
Compression Level: 3 - Decompression Speed
simdjson
Throughput Test: Kostya
LAME MP3 Encoding
WAV To MP3
Phoronix Test Suite v10.8.5