Xeon & EPYC Tests

Various Server CPU benchmarks

HTML result view exported from: https://openbenchmarking.org/result/2105236-VE-2105231VE89&grs.

ArrayFire

Test: BLAS CPU

NAS Parallel Benchmarks

Test / Class: SP.B

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_1d - Data Type: u8s8f32

MKL-DNN DNNL

Harness: IP Batch 1D - Data Type: f32

NAS Parallel Benchmarks

Test / Class: BT.C

NAS Parallel Benchmarks

Test / Class: EP.D

NAMD

ATPase Simulation - 327,506 Atoms

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32

NAS Parallel Benchmarks

Test / Class: MG.C

OSPray

Demo: NASA Streamlines - Renderer: Path Tracer

LuxCoreRender

Scene: DLSC

OSPray

Demo: XFrog Forest - Renderer: Path Tracer

OSPray

Demo: XFrog Forest - Renderer: SciVis

LuxCoreRender

Scene: Rainbow Colors and Prism

MKL-DNN DNNL

Harness: IP Batch 1D - Data Type: bf16bf16bf16

Blender

Blend File: Classroom - Compute: CPU-Only

Embree

Binary: Pathtracer - Model: Crown

OSPray

Demo: NASA Streamlines - Renderer: SciVis

Embree

Binary: Pathtracer ISPC - Model: Crown

OSPray

Demo: San Miguel - Renderer: Path Tracer

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_all - Data Type: f32

MKL-DNN DNNL

Harness: Convolution Batch conv_alexnet - Data Type: bf16bf16bf16

John The Ripper

Test: MD5

John The Ripper

Test: Blowfish

MKL-DNN DNNL

Harness: Convolution Batch conv_all - Data Type: bf16bf16bf16

Blender

Blend File: Pabellon Barcelona - Compute: CPU-Only

OSPray

Demo: San Miguel - Renderer: SciVis

Embree

Binary: Pathtracer - Model: Asian Dragon

Intel MPI Benchmarks

Test: IMB-P2P PingPong

Chaos Group V-RAY

Mode: CPU

MKL-DNN DNNL

Harness: Convolution Batch conv_3d - Data Type: bf16bf16bf16

Embree

Binary: Pathtracer - Model: Asian Dragon Obj

MKL-DNN DNNL

Harness: Convolution Batch conv_googlenet_v3 - Data Type: bf16bf16bf16

OSPray

Demo: Magnetic Reconnection - Renderer: SciVis

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon Obj

High Performance Conjugate Gradient

Stockfish

Total Time

Blender

Blend File: BMW27 - Compute: CPU-Only

MKL-DNN DNNL

Harness: Convolution Batch conv_3d - Data Type: u8s8f32

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_all - Data Type: bf16bf16bf16

asmFish

1024 Hash Memory, 26 Depth

NAS Parallel Benchmarks

Test / Class: LU.C

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_3d - Data Type: f32

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_1d - Data Type: bf16bf16bf16

Blender

Blend File: Fishy Cat - Compute: CPU-Only

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_3d - Data Type: bf16bf16bf16

MKL-DNN DNNL

Harness: Convolution Batch conv_alexnet - Data Type: f32

Tungsten Renderer

Scene: Hair

NAS Parallel Benchmarks

Test / Class: EP.C

MKL-DNN DNNL

Harness: Convolution Batch conv_googlenet_v3 - Data Type: f32

PyPerformance

Benchmark: python_startup

dav1d

Video Input: Chimera 1080p 10-bit

MKL-DNN DNNL

Harness: Convolution Batch conv_all - Data Type: f32

MKL-DNN DNNL

Harness: Deconvolution Batch deconv_1d - Data Type: f32

Appleseed

Scene: Material Tester

MKL-DNN DNNL

Harness: IP Batch All - Data Type: f32

Blender

Blend File: Barbershop - Compute: CPU-Only

dav1d

Video Input: Summer Nature 1080p

Basis Universal

Settings: UASTC Level 3

PostgreSQL pgbench

Scaling: Buffer Test - Test: Normal Load - Mode: Read Only

MKL-DNN DNNL

Harness: Convolution Batch conv_googlenet_v3 - Data Type: u8s8f32

ArrayFire

Test: Conjugate Gradient CPU

MKL-DNN DNNL

Harness: Convolution Batch conv_alexnet - Data Type: u8s8f32