AMD 5th Gen EPYC 9005 Turin vs. NVIDIA GH200 Performance Benchmarks

AMD 5th Gen EPYC "Turin" Linux benchmarks by Michael Larabel for a future article on launch day of EPYC 9005 series Linux performance.

HTML result view exported from: https://openbenchmarking.org/result/2410298-NE-NVIDIAGRA46&grs.

OpenSSL

Algorithm: ChaCha20-Poly1305

Liquid-DSP

Threads: 128 - Buffer Length: 256 - Filter Length: 512

OpenVINO

Model: Vehicle Detection FP16-INT8 - Device: CPU

OpenVINO

Model: Weld Porosity Detection FP16-INT8 - Device: CPU

Liquid-DSP

Threads: 1 - Buffer Length: 256 - Filter Length: 512

OpenVINO

Model: Face Detection Retail FP16-INT8 - Device: CPU

miniBUDE

Implementation: OpenMP - Input Deck: BM2

miniBUDE

Implementation: OpenMP - Input Deck: BM2

NAS Parallel Benchmarks

Test / Class: LU.C

OpenVINO

Model: Road Segmentation ADAS FP16-INT8 - Device: CPU

NAS Parallel Benchmarks

Test / Class: SP.C

OpenVINO

Model: Noise Suppression Poconet-Like FP16 - Device: CPU

OpenVINO

Model: Person Vehicle Bike Detection FP16 - Device: CPU

BRL-CAD

VGR Performance Metric

OpenSSL

Algorithm: RSA4096

John The Ripper

Test: bcrypt

John The Ripper

Test: Blowfish

OpenVINO

Model: Weld Porosity Detection FP16-INT8 - Device: CPU

OpenVINO

Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPU

srsRAN Project

Test: PUSCH Processor Benchmark, Throughput Total

Memcached

Set To Get Ratio: 1:100

ASKAP

Test: tConvolve MPI - Gridding

OpenSSL

Algorithm: AES-256-GCM

OpenSSL

Algorithm: AES-128-GCM

OpenVINO

Model: Face Detection FP16-INT8 - Device: CPU

NAS Parallel Benchmarks

Test / Class: IS.D

OpenVINO

Model: Machine Translation EN To DE FP16 - Device: CPU

GROMACS

Implementation: MPI CPU - Input: water_GMX50_bare

Stockfish

Chess Benchmark

OpenSSL

Algorithm: RSA4096

Liquid-DSP

Threads: 256 - Buffer Length: 256 - Filter Length: 57

ASKAP

Test: tConvolve MPI - Degridding

OpenVINO

Model: Vehicle Detection FP16-INT8 - Device: CPU

OpenVINO

Model: Road Segmentation ADAS FP16-INT8 - Device: CPU

John The Ripper

Test: WPA PSK

OpenVINO

Model: Person Vehicle Bike Detection FP16 - Device: CPU

srsRAN Project

Test: PUSCH Processor Benchmark, Throughput Thread

ASTC Encoder

Preset: Very Thorough

ASTC Encoder

Preset: Exhaustive

ASTC Encoder

Preset: Thorough

libavif avifenc

Encoder Speed: 0

OpenSSL

Algorithm: SHA256

7-Zip Compression

Test: Decompression Rating

7-Zip Compression

Test: Compression Rating

QuantLib

Configuration: Multi-Threaded

OSPRay Studio

Camera: 1 - Resolution: 4K - Samples Per Pixel: 32 - Renderer: Path Tracer - Acceleration: CPU

OpenFOAM

Input: drivaerFastback, Medium Mesh Size - Execution Time

OSPRay Studio

Camera: 3 - Resolution: 4K - Samples Per Pixel: 32 - Renderer: Path Tracer - Acceleration: CPU

OpenVINO

Model: Person Detection FP16 - Device: CPU

OpenVINO

Model: Age Gender Recognition Retail 0013 FP16-INT8 - Device: CPU

Liquid-DSP

Threads: 256 - Buffer Length: 256 - Filter Length: 512

OpenVINO

Model: Face Detection Retail FP16-INT8 - Device: CPU

OpenVINO

Model: Face Detection FP16-INT8 - Device: CPU

OpenVINO

Model: Person Detection FP16 - Device: CPU

OpenVINO

Model: Machine Translation EN To DE FP16 - Device: CPU

OpenVINO

Model: Noise Suppression Poconet-Like FP16 - Device: CPU

OpenSSL

Algorithm: ChaCha20

Primesieve

Length: 1e13

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon Obj

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon

OpenSSL

Algorithm: SHA512

OpenFOAM

Input: drivaerFastback, Medium Mesh Size - Mesh Time

Primesieve

Length: 1e12

OSPRay Studio

Camera: 3 - Resolution: 4K - Samples Per Pixel: 16 - Renderer: Path Tracer - Acceleration: CPU

OSPRay Studio

Camera: 3 - Resolution: 4K - Samples Per Pixel: 1 - Renderer: Path Tracer - Acceleration: CPU

OSPRay Studio

Camera: 1 - Resolution: 4K - Samples Per Pixel: 1 - Renderer: Path Tracer - Acceleration: CPU

OSPRay Studio

Camera: 1 - Resolution: 4K - Samples Per Pixel: 16 - Renderer: Path Tracer - Acceleration: CPU

RocksDB

Test: Random Read

Embree

Binary: Pathtracer ISPC - Model: Crown

Parallel BZIP2 Compression

FreeBSD-13.0-RELEASE-amd64-memstick.img Compression

Blender

Blend File: Classroom - Compute: CPU-Only

Blender

Blend File: Barbershop - Compute: CPU-Only

Speedb

Test: Random Read

SecureMark

Benchmark: SecureMark-TLS

OSPRay

Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_time

Liquid-DSP

Threads: 256 - Buffer Length: 256 - Filter Length: 32

Blender

Blend File: BMW27 - Compute: CPU-Only

Blender

Blend File: Pabellon Barcelona - Compute: CPU-Only

OpenVKL

Benchmark: vklBenchmarkCPU ISPC

Coremark

CoreMark Size 666 - Iterations Per Second

uvg266

Video Input: Bosphorus 4K - Video Preset: Slow

Blender

Blend File: Junkshop - Compute: CPU-Only

Blender

Blend File: Fishy Cat - Compute: CPU-Only

Timed Node.js Compilation

Time To Compile

uvg266

Video Input: Bosphorus 4K - Video Preset: Medium

Kvazaar

Video Input: Bosphorus 4K - Video Preset: Slow

Timed LLVM Compilation

Build System: Ninja

OSPRay

Benchmark: particle_volume/scivis/real_time

OSPRay

Benchmark: particle_volume/ao/real_time

Pennant

Test: sedovbig

Zstd Compression

Compression Level: 19 - Compression Speed

Kvazaar

Video Input: Bosphorus 4K - Video Preset: Medium

Zstd Compression

Compression Level: 19 - Decompression Speed

Helsing

Digit Range: 14 digit

WebP Image Encode

Encode Settings: Quality 100, Lossless, Highest Compression

Liquid-DSP

Threads: 1 - Buffer Length: 256 - Filter Length: 32

High Performance Conjugate Gradient

X Y Z: 144 144 144 - RT: 60

Zstd Compression

Compression Level: 19, Long Mode - Decompression Speed

PyBench

Total For Average Test Times

NAMD

Input: ATPase with 327,506 Atoms

Timed Godot Game Engine Compilation

Time To Compile

GraphicsMagick

Operation: Sharpen

miniFE

Problem Size: Small

OpenFOAM

Input: drivaerFastback, Small Mesh Size - Execution Time

GraphicsMagick

Operation: Enhanced

Algebraic Multi-Grid Benchmark

LAMMPS Molecular Dynamics Simulator

Model: Rhodopsin Protein

ClickHouse

100M Rows Hits Dataset, Second Run

Numpy Benchmark

Zstd Compression

Compression Level: 19, Long Mode - Compression Speed

GraphicsMagick

Operation: Noise-Gaussian

QMCPACK

Input: Li2_STO_ae

ClickHouse

100M Rows Hits Dataset, Third Run

ClickHouse

100M Rows Hits Dataset, First Run / Cold Cache

CloverLeaf

Input: clover_bm64_short

Intel Open Image Denoise

Run: RTLightmap.hdr.4096x4096 - Device: CPU-Only

Intel Open Image Denoise

Run: RT.hdr_alb_nrm.3840x2160 - Device: CPU-Only

Intel Open Image Denoise

Run: RT.ldr_alb_nrm.3840x2160 - Device: CPU-Only

NAMD

Input: STMV with 1,066,628 Atoms

FFmpeg

Encoder: libx265 - Scenario: Platform

NWChem

Input: C240 Buckyball

Timed Gem5 Compilation

Time To Compile

OpenRadioss

Model: Cell Phone Drop Test

SVT-AV1

Encoder Mode: Preset 4 - Input: Bosphorus 4K

Xcompact3d Incompact3d

Input: X3D-benchmarking input.i3d

LAMMPS Molecular Dynamics Simulator

Model: 20k Atoms

Timed Linux Kernel Compilation

Build: allmodconfig

OpenRadioss

Model: Chrysler Neon 1M

SVT-AV1

Encoder Mode: Preset 12 - Input: Bosphorus 4K

uvg266

Video Input: Bosphorus 4K - Video Preset: Very Fast

TensorFlow

Device: CPU - Batch Size: 256 - Model: ResNet-50

SVT-AV1

Encoder Mode: Preset 8 - Input: Bosphorus 4K

OpenRadioss

Model: INIVOL and Fluid Structure Interaction Drop Container

GPAW

Input: Carbon Nanotube

TensorFlow

Device: CPU - Batch Size: 512 - Model: ResNet-50

TensorFlow

Device: CPU - Batch Size: 256 - Model: AlexNet

Meta Performance Per Watts

Performance Per Watts

OpenFOAM

Input: drivaerFastback, Small Mesh Size - Mesh Time

CPU Power Consumption Monitor

Phoronix Test Suite System Monitoring