AMD EPYC Zen 4C Power Efficiency

AMD EPYC Zen 4C power efficiency benchmarks by Michael Larabel for a future article.

HTML result view exported from: https://openbenchmarking.org/result/2401147-NE-AMDEPYCZE24&grs&rdt.

OpenVINO

Model: Vehicle Detection FP16-INT8 - Device: CPU

OpenVINO

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

OpenVINO

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

OpenVINO

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

Llama.cpp

Model: llama-2-70b-chat.Q5_0.gguf

Llama.cpp

Model: llama-2-13b.Q4_0.gguf

miniBUDE

Implementation: OpenMP - Input Deck: BM2

miniBUDE

Implementation: OpenMP - Input Deck: BM2

miniBUDE

Implementation: OpenMP - Input Deck: BM1

miniBUDE

Implementation: OpenMP - Input Deck: BM1

Neural Magic DeepSparse

Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Stream

OpenSSL

Algorithm: ChaCha20-Poly1305

OpenVINO

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

OSPRay

Benchmark: gravity_spheres_volume/dim_512/scivis/real_time

OpenSSL

Algorithm: ChaCha20

TensorFlow

Device: CPU - Batch Size: 16 - Model: ResNet-50

OSPRay

Benchmark: gravity_spheres_volume/dim_512/ao/real_time

OpenVINO

Model: Handwritten English Recognition FP16-INT8 - Device: CPU

OpenVINO

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

OpenVINO

Model: Person Vehicle Bike Detection FP16 - Device: CPU

ACES DGEMM

Sustained Floating-Point Rate

Neural Magic DeepSparse

Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Stream

OpenVINO

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

OpenSSL

Algorithm: AES-128-GCM

OpenVINO

Model: Person Detection FP16 - Device: CPU

Neural Magic DeepSparse

Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Stream

OpenSSL

Algorithm: AES-256-GCM

OpenSSL

Algorithm: RSA4096

Neural Magic DeepSparse

Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Stream

Xmrig

Variant: GhostRider - Hash Count: 1M

OpenVINO

Model: Face Detection FP16-INT8 - Device: CPU

OSPRay

Benchmark: gravity_spheres_volume/dim_512/pathtracer/real_time

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

easyWave

Input: e2Asean Grid + BengkuluSept2007 Source - Time: 1200

Xmrig

Variant: Wownero - Hash Count: 1M

OSPRay Studio

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

OSPRay Studio

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

OpenVINO

Model: Handwritten English Recognition FP16-INT8 - Device: CPU

OSPRay Studio

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

OSPRay Studio

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

Intel Open Image Denoise

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

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon

Neural Magic DeepSparse

Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: BERT-Large, NLP Question Answering - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Stream

easyWave

Input: e2Asean Grid + BengkuluSept2007 Source - Time: 2400

Y-Cruncher

Pi Digits To Calculate: 1B

Embree

Binary: Pathtracer ISPC - Model: Crown

OpenVINO

Model: Vehicle Detection FP16-INT8 - Device: CPU

Neural Magic DeepSparse

Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Stream

Y-Cruncher

Pi Digits To Calculate: 500M

RocksDB

Test: Random Read

Speedb

Test: Random Read

Blender

Blend File: Barbershop - Compute: CPU-Only

OSPRay

Benchmark: particle_volume/scivis/real_time

OpenSSL

Algorithm: RSA4096

OSPRay

Benchmark: particle_volume/ao/real_time

Neural Magic DeepSparse

Model: CV Detection, YOLOv5s COCO - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Asynchronous Multi-Stream

Chaos Group V-RAY

Mode: CPU

OpenSSL

Algorithm: SHA512

SVT-AV1

Encoder Mode: Preset 12 - Input: Bosphorus 4K

7-Zip Compression

Test: Compression Rating

Blender

Blend File: Fishy Cat - Compute: CPU-Only

OpenSSL

Algorithm: SHA256

Blender

Blend File: Pabellon Barcelona - Compute: CPU-Only

7-Zip Compression

Test: Decompression Rating

nginx

Connections: 1000

IndigoBench

Acceleration: CPU - Scene: Supercar

OpenVINO

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

Blender

Blend File: Classroom - Compute: CPU-Only

IndigoBench

Acceleration: CPU - Scene: Bedroom

nginx

Connections: 500

Redis 7.0.12 + memtier_benchmark

Protocol: Redis - Clients: 100 - Set To Get Ratio: 1:5

SVT-AV1

Encoder Mode: Preset 8 - Input: Bosphorus 4K

QuantLib

Configuration: Multi-Threaded

Blender

Blend File: BMW27 - Compute: CPU-Only

Timed LLVM Compilation

Build System: Ninja

NAMD

ATPase Simulation - 327,506 Atoms

FFmpeg

Encoder: libx265 - Scenario: Video On Demand

FFmpeg

Encoder: libx265 - Scenario: Upload

Timed Node.js Compilation

Time To Compile

FFmpeg

Encoder: libx265 - Scenario: Platform

Redis 7.0.12 + memtier_benchmark

Protocol: Redis - Clients: 100 - Set To Get Ratio: 1:10

Timed Linux Kernel Compilation

Build: allmodconfig

GPAW

Input: Carbon Nanotube

Neural Magic DeepSparse

Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Stream

uvg266

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

SVT-AV1

Encoder Mode: Preset 4 - Input: Bosphorus 4K

OpenVINO

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

PyTorch

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

Speedb

Test: Read While Writing

Neural Magic DeepSparse

Model: BERT-Large, NLP Question Answering - Scenario: Asynchronous Multi-Stream

uvg266

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

OpenFOAM

Input: drivaerFastback, Small Mesh Size - Execution Time

Timed FFmpeg Compilation

Time To Compile

Neural Magic DeepSparse

Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Stream

Neural Magic DeepSparse

Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Stream

OpenVINO

Model: Face Detection FP16-INT8 - Device: CPU

OpenVINO

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

Neural Magic DeepSparse

Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Stream

PyTorch

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

FFmpeg

Encoder: libx265 - Scenario: Live

CloverLeaf

Input: clover_bm16

Xcompact3d Incompact3d

Input: X3D-benchmarking input.i3d

Neural Magic DeepSparse

Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Stream

RocksDB

Test: Read While Writing

Neural Magic DeepSparse

Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Stream

x265

Video Input: Bosphorus 4K

Timed LLVM Compilation

Build System: Unix Makefiles