AMD EPYC 9755 DDR5 Turin Memory Performance

AMD EPYC 9755 with varying DDR5-6000 default versus DDR5-4800 memory performance. Benchmarks by Michael Larabel for a future article.

HTML result view exported from: https://openbenchmarking.org/result/2410130-NE-TURINDDR565&grs&sor.

ASKAP

Test: tConvolve MT - Degridding

High Performance Conjugate Gradient

X Y Z: 144 144 144 - RT: 60

Algebraic Multi-Grid Benchmark

CloverLeaf

Input: clover_bm16

miniFE

Problem Size: Small

CloverLeaf

Input: clover_bm64_short

ASKAP

Test: tConvolve MT - Gridding

NAS Parallel Benchmarks

Test / Class: SP.C

NAS Parallel Benchmarks

Test / Class: MG.C

Xcompact3d Incompact3d

Input: X3D-benchmarking input.i3d

WRF

Input: conus 2.5km

OpenRadioss

Model: Chrysler Neon 1M

OpenFOAM

Input: drivaerFastback, Medium Mesh Size - Execution Time

Llamafile

Test: wizardcoder-python-34b-v1.0.Q6_K - Acceleration: CPU

OpenVINO

Model: Person Detection FP16 - Device: CPU

OpenVINO

Model: Person Detection FP16 - Device: CPU

ASKAP

Test: tConvolve MPI - Gridding

NAS Parallel Benchmarks

Test / Class: IS.D

OpenVINO

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

OpenFOAM

Input: drivaerFastback, Small Mesh Size - Mesh Time

LULESH

OpenVINO

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

OpenVINO

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

TensorFlow

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

ASKAP

Test: tConvolve MPI - Degridding

ClickHouse

100M Rows Hits Dataset, Second Run

TensorFlow

Device: CPU - Batch Size: 512 - Model: GoogLeNet

Xcompact3d Incompact3d

Input: input.i3d 193 Cells Per Direction

TensorFlow

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

VVenC

Video Input: Bosphorus 4K - Video Preset: Faster

OpenVINO

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

Intel Open Image Denoise

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

ClickHouse

100M Rows Hits Dataset, Third Run

7-Zip Compression

Test: Compression Rating

TensorFlow

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

SVT-AV1

Encoder Mode: Preset 12 - Input: Bosphorus 4K

ClickHouse

100M Rows Hits Dataset, First Run / Cold Cache

uvg266

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

TensorFlow

Device: CPU - Batch Size: 64 - Model: ResNet-50

PyTorch

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

Kvazaar

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

TensorFlow

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

TensorFlow

Device: CPU - Batch Size: 1 - Model: ResNet-50

Blender

Blend File: Junkshop - Compute: 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

GPAW

Input: Carbon Nanotube

OpenVINO

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

OpenVINO

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

Timed Gem5 Compilation

Time To Compile

Zstd Compression

Compression Level: 19 - Compression Speed

Blender

Blend File: BMW27 - Compute: CPU-Only

PyTorch

Device: CPU - Batch Size: 64 - Model: ResNet-152

TensorFlow

Device: CPU - Batch Size: 64 - Model: GoogLeNet

Timed Godot Game Engine Compilation

Time To Compile

NAMD

Input: STMV with 1,066,628 Atoms

GraphicsMagick

Operation: Swirl

Llamafile

Test: llava-v1.5-7b-q4 - Acceleration: CPU

WebP Image Encode

Encode Settings: Quality 100, Lossless, Highest Compression

NAMD

Input: ATPase with 327,506 Atoms

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon Obj

RocksDB

Test: Read While Writing

Timed ImageMagick Compilation

Time To Compile

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon

Timed Linux Kernel Compilation

Build: allmodconfig

PyTorch

Device: CPU - Batch Size: 64 - Model: ResNet-50

TensorFlow

Device: CPU - Batch Size: 64 - Model: AlexNet

ONNX Runtime

Model: ArcFace ResNet-100 - Device: CPU - Executor: Standard

srsRAN Project

Test: PUSCH Processor Benchmark, Throughput Total

ONNX Runtime

Model: CaffeNet 12-int8 - Device: CPU - Executor: Parallel

LAMMPS Molecular Dynamics Simulator

Model: Rhodopsin Protein

GNU Octave Benchmark

PyTorch

Device: CPU - Batch Size: 1 - Model: ResNet-152

OpenVKL

Benchmark: vklBenchmarkCPU ISPC

TensorFlow

Device: CPU - Batch Size: 1 - Model: GoogLeNet

oneDNN

Harness: Deconvolution Batch shapes_3d - Engine: CPU

PostgreSQL

Scaling Factor: 100 - Clients: 1000 - Mode: Read Only - Average Latency

Memcached

Set To Get Ratio: 1:100

LAMMPS Molecular Dynamics Simulator

Model: 20k Atoms

PostgreSQL

Scaling Factor: 100 - Clients: 1000 - Mode: Read Only

ONNX Runtime

Model: GPT-2 - Device: CPU - Executor: Standard

ONNX Runtime

Model: T5 Encoder - Device: CPU - Executor: Standard

GROMACS

Implementation: MPI CPU - Input: water_GMX50_bare

ACES DGEMM

Sustained Floating-Point Rate

VVenC

Video Input: Bosphorus 4K - Video Preset: Fast

ONNX Runtime

Model: yolov4 - Device: CPU - Executor: Parallel

PyTorch

Device: CPU - Batch Size: 256 - Model: ResNet-152

Timed Node.js Compilation

Time To Compile

Helsing

Digit Range: 14 digit

Blender

Blend File: Classroom - Compute: CPU-Only

BRL-CAD

VGR Performance Metric

PyTorch

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

Timed LLVM Compilation

Build System: Ninja

libavif avifenc

Encoder Speed: 6, Lossless

RocksDB

Test: Random Read

Apache IoTDB

Device Count: 800 - Batch Size Per Write: 100 - Sensor Count: 800 - Client Number: 100

FFmpeg

Encoder: libx265 - Scenario: Video On Demand

FFmpeg

Encoder: libx265 - Scenario: Platform

TensorFlow

Device: CPU - Batch Size: 512 - Model: AlexNet

PyTorch

Device: CPU - Batch Size: 1 - Model: ResNet-50

OpenSSL

Algorithm: SHA512

PostgreSQL

Scaling Factor: 100 - Clients: 1000 - Mode: Read Write - Average Latency

PostgreSQL

Scaling Factor: 100 - Clients: 1000 - Mode: Read Write

Blender

Blend File: Barbershop - Compute: CPU-Only

GraphicsMagick

Operation: Sharpen

Y-Cruncher

Pi Digits To Calculate: 1B

OpenRadioss

Model: Cell Phone Drop Test

TensorFlow

Device: CPU - Batch Size: 1 - Model: AlexNet

Liquid-DSP

Threads: 128 - Buffer Length: 256 - Filter Length: 57

srsRAN Project

Test: PDSCH Processor Benchmark, Throughput Total

FFmpeg

Encoder: libx265 - Scenario: Upload

Y-Cruncher

Pi Digits To Calculate: 500M

OpenVINO

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

m-queens

Time To Solve

OpenVINO

Model: Person Vehicle Bike Detection FP16 - Device: CPU

OSPRay Studio

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

Parallel BZIP2 Compression

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

OSPRay Studio

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

ASTC Encoder

Preset: Medium

Apache IoTDB

Device Count: 500 - Batch Size Per Write: 100 - Sensor Count: 800 - Client Number: 400

Appleseed

Scene: Disney Material

OSPRay Studio

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

NAS Parallel Benchmarks

Test / Class: CG.C

SecureMark

Benchmark: SecureMark-TLS

OpenVINO

Model: Person Vehicle Bike Detection FP16 - Device: CPU

Liquid-DSP

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

Zstd Compression

Compression Level: 19, Long Mode - Decompression Speed

Timed Linux Kernel Compilation

Build: defconfig

Embree

Binary: Pathtracer ISPC - Model: Crown

miniBUDE

Implementation: OpenMP - Input Deck: BM1

miniBUDE

Implementation: OpenMP - Input Deck: BM1

WebP Image Encode

Encode Settings: Quality 100, Highest Compression

NAS Parallel Benchmarks

Test / Class: LU.C

Blender

Blend File: Pabellon Barcelona - Compute: CPU-Only

Apache IoTDB

Device Count: 800 - Batch Size Per Write: 100 - Sensor Count: 800 - Client Number: 100

OSPRay Studio

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

ONNX Runtime

Model: super-resolution-10 - Device: CPU - Executor: Parallel

OSPRay Studio

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

GraphicsMagick

Operation: Enhanced

Stockfish

Chess Benchmark

Speedb

Test: Random Read

Timed FFmpeg Compilation

Time To Compile

OSPRay

Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_time

ONNX Runtime

Model: ArcFace ResNet-100 - Device: CPU - Executor: Parallel

Primesieve

Length: 1e12

x265

Video Input: Bosphorus 4K