EPYC 2021 Benchmarks
Tests for a future article.
HTML result view exported from: https://openbenchmarking.org/result/2102219-HA-EB716339316&sro&grr.
Quantum ESPRESSO
Input: AUSURF112
NWChem
Input: C240 Buckyball
LeelaChessZero
Backend: Eigen
LeelaChessZero
Backend: BLAS
LAMMPS Molecular Dynamics Simulator
Model: 20k Atoms
Caffe
Model: GoogleNet - Acceleration: CPU - Iterations: 200
OpenFOAM
Input: Motorbike 60M
Timed LLVM Compilation
Time To Compile
BlogBench
Test: Read
OpenVKL
Benchmark: vklBenchmark
Hierarchical INTegration
Test: FLOAT
AI Benchmark Alpha
Device AI Score
AI Benchmark Alpha
Device Training Score
AI Benchmark Alpha
Device Inference Score
Crypto++
Test: Keyed Algorithms
Blender
Blend File: Barbershop - Compute: CPU-Only
JPEG XL
Input: PNG - Encode Speed: 8
ONNX Runtime
Model: super-resolution-10 - Device: OpenMP CPU
Ngspice
Circuit: C7552
WebP2 Image Encode
Encode Settings: Quality 95, Compression Effort 7
Numpy Benchmark
Tinymembench
Standard Memset
BRL-CAD
VGR Performance Metric
Blender
Blend File: Pabellon Barcelona - Compute: CPU-Only
Crypto++
Test: Integer + Elliptic Curve Public Key Algorithms
ONNX Runtime
Model: shufflenet-v2-10 - Device: OpenMP CPU
Incompact3D
Input: Cylinder
ONNX Runtime
Model: bertsquad-10 - Device: OpenMP CPU
Blender
Blend File: Classroom - Compute: CPU-Only
Monte Carlo Simulations of Ionised Nebulae
Input: Dust 2D tau100.0
High Performance Conjugate Gradient
Ngspice
Circuit: C2670
Mobile Neural Network
Model: mobilenet-v1-1.0
Mobile Neural Network
Model: MobileNetV2_224
Mobile Neural Network
Model: resnet-v2-50
Mobile Neural Network
Model: SqueezeNetV1.0
Darmstadt Automotive Parallel Heterogeneous Suite
Backend: OpenMP - Kernel: Points2Image
Caffe
Model: AlexNet - Acceleration: CPU - Iterations: 200
asmFish
1024 Hash Memory, 26 Depth
WebP2 Image Encode
Encode Settings: Quality 75, Compression Effort 7
Appleseed
Scene: Emily
YafaRay
Total Time For Sample Scene
ONNX Runtime
Model: fcn-resnet101-11 - Device: OpenMP CPU
Rodinia
Test: OpenMP Leukocyte
OSPray
Demo: San Miguel - Renderer: Path Tracer
GPAW
Input: Carbon Nanotube
ONNX Runtime
Model: yolov4 - Device: OpenMP CPU
Rodinia
Test: OpenMP LavaMD
PlaidML
FP16: No - Mode: Inference - Network: VGG19 - Device: CPU
C-Blosc
Compressor: blosclz
Stress-NG
Test: CPU Cache
TensorFlow Lite
Model: NASNet Mobile
Chaos Group V-RAY
Mode: CPU
ASTC Encoder
Preset: Exhaustive
Blender
Blend File: Fishy Cat - Compute: CPU-Only
Cpuminer-Opt
Algorithm: Garlicoin
Cpuminer-Opt
Algorithm: Deepcoin
LZ4 Compression
Compression Level: 9 - Decompression Speed
LZ4 Compression
Compression Level: 9 - Compression Speed
Apache Cassandra
Test: Writes
PlaidML
FP16: No - Mode: Inference - Network: VGG16 - Device: CPU
Numenta Anomaly Benchmark
Detector: Earthgecko Skyline
dav1d
Video Input: Chimera 1080p 10-bit
FinanceBench
Benchmark: Bonds OpenMP
JPEG XL
Input: PNG - Encode Speed: 7
Timed MrBayes Analysis
Primate Phylogeny Analysis
Apache CouchDB
Bulk Size: 100 - Inserts: 1000 - Rounds: 24
Chaos Group V-RAY
Mode: CPU
Montage Astronomical Image Mosaic Engine
Mosaic of M17, K band, 1.5 deg x 1.5 deg
NAS Parallel Benchmarks
Test / Class: EP.D
oneDNN
Harness: Recurrent Neural Network Training - Data Type: u8s8f32 - Engine: CPU
InfluxDB
Concurrent Streams: 4 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
KeyDB
Facebook RocksDB
Test: Read While Writing
Timed Godot Game Engine Compilation
Time To Compile
LZ4 Compression
Compression Level: 3 - Decompression Speed
LZ4 Compression
Compression Level: 3 - Compression Speed
OSPray
Demo: XFrog Forest - Renderer: Path Tracer
oneDNN
Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPU
InfluxDB
Concurrent Streams: 64 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
TensorFlow Lite
Model: Inception ResNet V2
Build2
Time To Compile
GROMACS
Water Benchmark
Blender
Blend File: BMW27 - Compute: CPU-Only
TensorFlow Lite
Model: Inception V4
Kripke
Appleseed
Scene: Disney Material
Cpuminer-Opt
Algorithm: Skeincoin
oneDNN
Harness: Recurrent Neural Network Training - Data Type: bf16bf16bf16 - Engine: CPU
JPEG XL Decoding
CPU Threads: 1
GROMACS
Input: water_GMX50_bare
Stockfish
Total Time
OpenVINO
Model: Age Gender Recognition Retail 0013 FP16 - Device: CPU
OpenVINO
Model: Age Gender Recognition Retail 0013 FP16 - Device: CPU
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPU
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: bf16bf16bf16 - Engine: CPU
Perl Benchmarks
Test: Pod2html
Himeno Benchmark
Poisson Pressure Solver
OpenVINO
Model: Person Detection 0106 FP16 - Device: CPU
OpenVINO
Model: Person Detection 0106 FP16 - Device: CPU
Cpuminer-Opt
Algorithm: x25x
OpenVINO
Model: Person Detection 0106 FP32 - Device: CPU
OpenVINO
Model: Person Detection 0106 FP32 - Device: CPU
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 100 - Mode: Read Write - Average Latency
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 100 - Mode: Read Write
Coremark
CoreMark Size 666 - Iterations Per Second
OpenVINO
Model: Face Detection 0106 FP16 - Device: CPU
OpenVINO
Model: Face Detection 0106 FP16 - Device: CPU
OpenVINO
Model: Face Detection 0106 FP32 - Device: CPU
OpenVINO
Model: Face Detection 0106 FP32 - Device: CPU
Crypto++
Test: Unkeyed Algorithms
Facebook RocksDB
Test: Random Read
OpenVINO
Model: Age Gender Recognition Retail 0013 FP32 - Device: CPU
OpenVINO
Model: Age Gender Recognition Retail 0013 FP32 - Device: CPU
IndigoBench
Acceleration: CPU - Scene: Bedroom
Timed Linux Kernel Compilation
Time To Compile
IndigoBench
Acceleration: CPU - Scene: Supercar
LuxCoreRender
Scene: DLSC
LuxCoreRender
Scene: Rainbow Colors and Prism
TensorFlow Lite
Model: SqueezeNet
TensorFlow Lite
Model: Mobilenet Float
TensorFlow Lite
Model: Mobilenet Quant
Facebook RocksDB
Test: Random Fill Sync
John The Ripper
Test: MD5
simdjson
Throughput Test: LargeRandom
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Medium
FinanceBench
Benchmark: Repo OpenMP
RawTherapee
Total Benchmark Time
MBW
Test: Memory Copy - Array Size: 8192 MiB
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 250 - Mode: Read Write - Average Latency
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 250 - Mode: Read Write
NAMD
ATPase Simulation - 327,506 Atoms
simdjson
Throughput Test: PartialTweets
simdjson
Throughput Test: DistinctUserID
Hugin
Panorama Photo Assistant + Stitching Time
simdjson
Throughput Test: Kostya
OpenFOAM
Input: Motorbike 30M
JPEG XL Decoding
CPU Threads: All
Perl Benchmarks
Test: Interpreter
Timed PHP Compilation
Time To Compile
7-Zip Compression
Compress Speed Test
OSPray
Demo: XFrog Forest - Renderer: SciVis
eSpeak-NG Speech Engine
Text-To-Speech Synthesis
Redis
Test: SET
Redis
Test: GET
Tachyon
Total Time
Zstd Compression
Compression Level: 19
ACES DGEMM
Sustained Floating-Point Rate
PyPerformance
Benchmark: regex_compile
ASKAP
Test: tConvolve MPI - Gridding
ASKAP
Test: tConvolve MPI - Degridding
OSPray
Demo: San Miguel - Renderer: SciVis
QuantLib
Parboil
Test: OpenMP LBM
PHPBench
PHP Benchmark Suite
ebizzy
LibRaw
Post-Processing Benchmark
m-queens
Time To Solve
Timed FFmpeg Compilation
Time To Compile
Etcpak
Configuration: ETC2
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 250 - Mode: Read Only - Average Latency
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 250 - Mode: Read Only
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 100 - Mode: Read Only - Average Latency
PostgreSQL pgbench
Scaling Factor: 100 - Clients: 100 - Mode: Read Only
Numenta Anomaly Benchmark
Detector: Bayesian Changepoint
PyPerformance
Benchmark: pathlib
Stress-NG
Test: Socket Activity
Basis Universal
Settings: UASTC Level 3
miniFE
Problem Size: Small
C-Ray
Total Time - 4K, 16 Rays Per Pixel
NAS Parallel Benchmarks
Test / Class: LU.C
Stress-NG
Test: Context Switching
OSPray
Demo: NASA Streamlines - Renderer: Path Tracer
x265
Video Input: Bosphorus 4K
oneDNN
Harness: Deconvolution Batch shapes_1d - Data Type: f32 - Engine: CPU
CloverLeaf
Lagrangian-Eulerian Hydrodynamics
Zstd Compression
Compression Level: 3
Stress-NG
Test: CPU Stress
Pennant
Test: sedovbig
John The Ripper
Test: Blowfish
Aircrack-ng
Stress-NG
Test: Crypto
Stress-NG
Test: Matrix Math
Stress-NG
Test: Vector Math
Google SynthMark
Test: VoiceMark_100
NAS Parallel Benchmarks
Test / Class: IS.D
Darmstadt Automotive Parallel Heterogeneous Suite
Backend: OpenMP - Kernel: NDT Mapping
LULESH
Crafty
Elapsed Time
PyPerformance
Benchmark: django_template
POV-Ray
Trace Time
XZ Compression
Compressing ubuntu-16.04.3-server-i386.img, Compression Level 9
PyBench
Total For Average Test Times
Tungsten Renderer
Scene: Water Caustic
PyPerformance
Benchmark: nbody
PyPerformance
Benchmark: float
Algebraic Multi-Grid Benchmark
PyPerformance
Benchmark: crypto_pyaes
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Very Fast
Swet
Average
toyBrot Fractal Generator
Implementation: C++ Tasks
toyBrot Fractal Generator
Implementation: OpenMP
Botan
Test: AES-256
JPEG XL
Input: JPEG - Encode Speed: 5
toyBrot Fractal Generator
Implementation: C++ Threads
toyBrot Fractal Generator
Implementation: TBB
Darmstadt Automotive Parallel Heterogeneous Suite
Backend: OpenMP - Kernel: Euclidean Cluster
Timed MPlayer Compilation
Time To Compile
Pennant
Test: leblancbig
Basis Universal
Settings: UASTC Level 2
ASKAP
Test: tConvolve OpenMP - Degridding
ASKAP
Test: tConvolve OpenMP - Gridding
Stream-Dynamic
- Triad
Stream-Dynamic
- Add
Stream-Dynamic
- Scale
Stream-Dynamic
- Copy
oneDNN
Harness: IP Shapes 1D - Data Type: f32 - Engine: CPU
Timed ImageMagick Compilation
Time To Compile
ASTC Encoder
Preset: Thorough
OCRMyPDF
Processing 60 Page PDF Document
Rodinia
Test: OpenMP Streamcluster
oneDNN
Harness: Deconvolution Batch shapes_1d - Data Type: u8s8f32 - Engine: CPU
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Ultra Fast
SVT-AV1
Encoder Mode: Enc Mode 4 - Input: 1080p
Numenta Anomaly Benchmark
Detector: Relative Entropy
Botan
Test: Blowfish
Botan
Test: Twofish
OSPray
Demo: Magnetic Reconnection - Renderer: SciVis
Botan
Test: CAST-256
Botan
Test: KASUMI
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPU
Rodinia
Test: OpenMP CFD Solver
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPU
Intel Open Image Denoise
Scene: Memorial
Sysbench
Test: Memory
Tungsten Renderer
Scene: Hair
JPEG XL
Input: PNG - Encode Speed: 5
JPEG XL
Input: JPEG - Encode Speed: 7
Stream
Type: Copy
NAS Parallel Benchmarks
Test / Class: CG.C
NAS Parallel Benchmarks
Test / Class: FT.C
WebP2 Image Encode
Encode Settings: Quality 100, Compression Effort 5
Sysbench
Test: CPU
SVT-AV1
Encoder Mode: Enc Mode 8 - Input: 1080p
rays1bench
Large Scene
JPEG XL
Input: JPEG - Encode Speed: 8
oneDNN
Harness: IP Shapes 3D - Data Type: f32 - Engine: CPU
OSPray
Demo: NASA Streamlines - Renderer: SciVis
dav1d
Video Input: Chimera 1080p
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPU
Numenta Anomaly Benchmark
Detector: Windowed Gaussian
SVT-VP9
Tuning: Visual Quality Optimized - Input: Bosphorus 1080p
dav1d
Video Input: Summer Nature 4K
x264
H.264 Video Encoding
NAS Parallel Benchmarks
Test / Class: EP.C
NAS Parallel Benchmarks
Test / Class: MG.C
Etcpak
Configuration: DXT1
FFTE
N=256, 3D Complex FFT Routine
oneDNN
Harness: Deconvolution Batch shapes_3d - Data Type: f32 - Engine: CPU
oneDNN
Harness: Deconvolution Batch shapes_3d - Data Type: u8s8f32 - Engine: CPU
LAMMPS Molecular Dynamics Simulator
Model: Rhodopsin Protein
SVT-VP9
Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p
TSCP
AI Chess Performance
dav1d
Video Input: Summer Nature 1080p
ctx_clock
Context Switch Time
Stream-Dynamic
- Triad
JPEG XL Decoding
CPU Threads: All
JPEG XL Decoding
CPU Threads: 1
JPEG XL
Input: JPEG - Encode Speed: 8
JPEG XL
Input: JPEG - Encode Speed: 7
JPEG XL
Input: JPEG - Encode Speed: 5
JPEG XL
Input: PNG - Encode Speed: 8
JPEG XL
Input: PNG - Encode Speed: 7
JPEG XL
Input: PNG - Encode Speed: 5
GROMACS
Input: water_GMX50_bare
Chaos Group V-RAY
Mode: CPU
Stream
Type: Triad
Algebraic Multi-Grid Benchmark
Intel Open Image Denoise
Scene: Memorial
OpenVKL
Benchmark: vklBenchmark
Hierarchical INTegration
Test: FLOAT
asmFish
1024 Hash Memory, 26 Depth
Stockfish
Total Time
TSCP
AI Chess Performance
Crafty
Elapsed Time
Sysbench
Test: Memory
Sysbench
Test: CPU
Swet
Average
QuantLib
Etcpak
Configuration: DXT1
Etcpak
Configuration: ETC2
Google SynthMark
Test: VoiceMark_100
LibRaw
Post-Processing Benchmark
MBW
Test: Memory Copy - Array Size: 8192 MiB
Tinymembench
Standard Memset
Stream
Type: Triad
Stream
Type: Add
Stream
Type: Add
Stream
Type: Scale
Stream
Type: Copy
Stress-NG
Test: CPU Cache
Stress-NG
Test: Context Switching
Stress-NG
Test: Socket Activity
Stress-NG
Test: Matrix Math
Stress-NG
Test: Vector Math
Stress-NG
Test: Crypto
Stress-NG
Test: CPU Stress
Coremark
CoreMark Size 666 - Iterations Per Second
LZ4 Compression
Compression Level: 9 - Decompression Speed
LZ4 Compression
Compression Level: 3 - Decompression Speed
7-Zip Compression
Compress Speed Test
Zstd Compression
Compression Level: 19
Zstd Compression
Compression Level: 3
C-Blosc
Compressor: blosclz
Crypto++
Test: Integer + Elliptic Curve Public Key Algorithms
Crypto++
Test: Keyed Algorithms
Crypto++
Test: Unkeyed Algorithms
Botan
Test: Twofish
Botan
Test: KASUMI
Botan
Test: CAST-256
Botan
Test: Blowfish
Botan
Test: AES-256
BlogBench
Test: Read
ebizzy
Facebook RocksDB
Test: Read While Writing
Facebook RocksDB
Test: Random Read
Facebook RocksDB
Test: Random Fill Sync
simdjson
Throughput Test: DistinctUserID
simdjson
Throughput Test: Kostya
simdjson
Throughput Test: LargeRandom
simdjson
Throughput Test: PartialTweets
PHPBench
PHP Benchmark Suite
Redis
Test: GET
Redis
Test: SET
InfluxDB
Concurrent Streams: 64 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
InfluxDB
Concurrent Streams: 4 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
KeyDB
Apache Cassandra
Test: Writes
BRL-CAD
VGR Performance Metric
Aircrack-ng
x265
Video Input: Bosphorus 4K
x264
H.264 Video Encoding
SVT-VP9
Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p
SVT-VP9
Tuning: Visual Quality Optimized - Input: Bosphorus 1080p
SVT-AV1
Encoder Mode: Enc Mode 4 - Input: 1080p
SVT-AV1
Encoder Mode: Enc Mode 8 - Input: 1080p
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Ultra Fast
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Very Fast
Kvazaar
Video Input: Bosphorus 4K - Video Preset: Medium
dav1d
Video Input: Chimera 1080p 10-bit
dav1d
Video Input: Chimera 1080p
dav1d
Video Input: Summer Nature 4K
dav1d
Video Input: Summer Nature 1080p
John The Ripper
Test: Blowfish
John The Ripper
Test: MD5
Cpuminer-Opt
Algorithm: Skeincoin
Cpuminer-Opt
Algorithm: Deepcoin
Cpuminer-Opt
Algorithm: Garlicoin
Cpuminer-Opt
Algorithm: x25x
PlaidML
FP16: No - Mode: Inference - Network: VGG19 - Device: CPU
PlaidML
FP16: No - Mode: Inference - Network: VGG16 - Device: CPU
ONNX Runtime
Model: bertsquad-10 - Device: OpenMP CPU
ONNX Runtime
Model: super-resolution-10 - Device: OpenMP CPU
ONNX Runtime
Model: shufflenet-v2-10 - Device: OpenMP CPU
ONNX Runtime
Model: fcn-resnet101-11 - Device: OpenMP CPU
ONNX Runtime
Model: yolov4 - Device: OpenMP CPU
LeelaChessZero
Backend: Eigen
LeelaChessZero
Backend: BLAS
AI Benchmark Alpha
Device AI Score
Numpy Benchmark
NAS Parallel Benchmarks
Test / Class: CG.C
NAS Parallel Benchmarks
Test / Class: MG.C
NAS Parallel Benchmarks
Test / Class: IS.D
NAS Parallel Benchmarks
Test / Class: LU.C
NAS Parallel Benchmarks
Test / Class: FT.C
NAS Parallel Benchmarks
Test / Class: EP.D
NAS Parallel Benchmarks
Test / Class: EP.C
GROMACS
Water Benchmark
miniFE
Problem Size: Small
LULESH
LAMMPS Molecular Dynamics Simulator
Model: 20k Atoms
LAMMPS Molecular Dynamics Simulator
Model: Rhodopsin Protein
Kripke
Himeno Benchmark
Poisson Pressure Solver
High Performance Conjugate Gradient
FFTE
N=256, 3D Complex FFT Routine
Darmstadt Automotive Parallel Heterogeneous Suite
Backend: OpenMP - Kernel: Points2Image
Darmstadt Automotive Parallel Heterogeneous Suite
Backend: OpenMP - Kernel: NDT Mapping
Darmstadt Automotive Parallel Heterogeneous Suite
Backend: OpenMP - Kernel: Euclidean Cluster
ASKAP
Test: tConvolve OpenMP - Degridding
ASKAP
Test: tConvolve MPI - Gridding
ACES DGEMM
Sustained Floating-Point Rate
IndigoBench
Acceleration: CPU - Scene: Bedroom
IndigoBench
Acceleration: CPU - Scene: Supercar
Chaos Group V-RAY
Mode: CPU
LuxCoreRender
Scene: Rainbow Colors and Prism
LuxCoreRender
Scene: DLSC
rays1bench
Large Scene
OSPray
Demo: San Miguel - Renderer: Path Tracer
OSPray
Demo: San Miguel - Renderer: SciVis
OSPray
Demo: NASA Streamlines - Renderer: Path Tracer
OSPray
Demo: NASA Streamlines - Renderer: SciVis
OSPray
Demo: XFrog Forest - Renderer: Path Tracer
OSPray
Demo: XFrog Forest - Renderer: SciVis
OSPray
Demo: Magnetic Reconnection - Renderer: SciVis
Phoronix Test Suite v10.8.4