EPYC 4004 Benchmarks

Benchmarks preparing for a future article upon launch by Michael Larabel.

HTML result view exported from: https://openbenchmarking.org/result/2405239-NE-4124P914394&grs&sor.

SVT-AV1

Encoder Mode: Preset 13 - Input: Bosphorus 1080p

Liquid-DSP

Threads: 2 - Buffer Length: 256 - Filter Length: 512

ONNX Runtime

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

SVT-AV1

Encoder Mode: Preset 13 - Input: Bosphorus 4K

ONNX Runtime

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

Llama.cpp

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

LuxCoreRender

Scene: Danish Mood - Acceleration: CPU

Apache HTTP Server

Concurrent Requests: 500

OpenVINO

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

OpenVINO

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

Apache IoTDB

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

ASKAP

Test: tConvolve OpenMP - Gridding

ASKAP

Test: tConvolve MT - Degridding

Apache IoTDB

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

OpenVINO

Model: Handwritten English Recognition FP16 - Device: CPU

OpenVINO

Model: Handwritten English Recognition FP16 - Device: CPU

Apache IoTDB

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

Apache IoTDB

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

Apache IoTDB

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

Apache IoTDB

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

Apache IoTDB

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

Apache IoTDB

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

Apache IoTDB

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

OpenVINO

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

SPECFEM3D

Model: Homogeneous Halfspace

srsRAN Project

Test: PUSCH Processor Benchmark, Throughput Total

oneDNN

Harness: Deconvolution Batch shapes_1d - Engine: CPU

Speedb

Test: Read While Writing

RocksDB

Test: Read Random Write Random

PostgreSQL

Scaling Factor: 100 - Clients: 500 - Mode: Read Only - Average Latency

Apache IoTDB

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

PostgreSQL

Scaling Factor: 100 - Clients: 500 - Mode: Read Only

Liquid-DSP

Threads: 4 - Buffer Length: 256 - Filter Length: 512

Speedb

Test: Random Fill Sync

HeFFTe - Highly Efficient FFT for Exascale

Test: r2c - Backend: FFTW - Precision: double - X Y Z: 128

Node.js V8 Web Tooling Benchmark

SVT-VP9

Tuning: Visual Quality Optimized - Input: Bosphorus 4K

SVT-VP9

Tuning: Visual Quality Optimized - Input: Bosphorus 1080p

nginx

Connections: 1000

GraphicsMagick

Operation: HWB Color Space

Apache IoTDB

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

RocksDB

Test: Read While Writing

OpenFOAM

Input: drivaerFastback, Small Mesh Size - Execution Time

OpenVINO

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

Llama.cpp

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

QMCPACK

Input: simple-H2O

Apache IoTDB

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

ASKAP

Test: tConvolve MPI - Degridding

Apache IoTDB

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

Apache IoTDB

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

Liquid-DSP

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

RocksDB

Test: Random Fill

OpenFOAM

Input: drivaerFastback, Medium Mesh Size - Execution Time

ASKAP

Test: tConvolve OpenMP - Degridding

OpenRadioss

Model: Bumper Beam

ONNX Runtime

Model: yolov4 - Device: CPU - Executor: Parallel

Apache Hadoop

Operation: Create - Threads: 100 - Files: 100000

Apache IoTDB

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

nginx

Connections: 500

SPECFEM3D

Model: Mount St. Helens

Apache IoTDB

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

OSPRay

Benchmark: particle_volume/pathtracer/real_time

m-queens

Time To Solve

MariaDB

Test: oltp_update_non_index - Threads: 64

Timed Linux Kernel Compilation

Build: allmodconfig

nginx

Connections: 200

ClickHouse

100M Rows Hits Dataset, Second Run

ACES DGEMM

Sustained Floating-Point Rate

srsRAN Project

Test: PDSCH Processor Benchmark, Throughput Thread

ONNX Runtime

Model: bertsquad-12 - Device: CPU - Executor: Parallel

Timed Linux Kernel Compilation

Build: defconfig

SPECFEM3D

Model: Tomographic Model

ONNX Runtime

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

Zstd Compression

Compression Level: 19, Long Mode - Decompression Speed

GraphicsMagick

Operation: Swirl

PostgreSQL

Scaling Factor: 100 - Clients: 500 - Mode: Read Write

PostgreSQL

Scaling Factor: 100 - Clients: 500 - Mode: Read Write - Average Latency

PyTorch

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

SPECFEM3D

Model: Layered Halfspace

PETSc

Test: Streams

Apache HTTP Server

Concurrent Requests: 1000

GraphicsMagick

Operation: Rotate

OpenVINO

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

CockroachDB

Workload: KV, 95% Reads - Concurrency: 128

Speedb

Test: Update Random

OpenVINO

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

QuantLib

Configuration: Single-Threaded

ONNX Runtime

Model: ResNet50 v1-12-int8 - Device: CPU - Executor: Parallel

x265

Video Input: Bosphorus 4K

Coremark

CoreMark Size 666 - Iterations Per Second

OpenVINO

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

Dragonflydb

Clients Per Thread: 10 - Set To Get Ratio: 1:10

OpenVINO

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

Timed Mesa Compilation

Time To Compile

ASKAP

Test: tConvolve MT - Gridding

BRL-CAD

VGR Performance Metric

MariaDB

Test: oltp_update_non_index - Threads: 32

OpenRadioss

Model: Rubber O-Ring Seal Installation

Zstd Compression

Compression Level: 19, Long Mode - Compression Speed

SPECFEM3D

Model: Water-layered Halfspace

LAMMPS Molecular Dynamics Simulator

Model: 20k Atoms

PyBench

Total For Average Test Times

Timed Wasmer Compilation

Time To Compile

Zstd Compression

Compression Level: 8 - Decompression Speed

High Performance Conjugate Gradient

X Y Z: 104 104 104 - RT: 60

Blender

Blend File: BMW27 - Compute: CPU-Only

Llamafile

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

NAMD

Input: ATPase with 327,506 Atoms

Timed CPython Compilation

Build Configuration: Default

MariaDB

Test: oltp_read_write - Threads: 32

LuxCoreRender

Scene: Rainbow Colors and Prism - Acceleration: CPU

LuxCoreRender

Scene: LuxCore Benchmark - Acceleration: CPU

OSPRay Studio

Camera: 1 - Resolution: 1080p - Samples Per Pixel: 16 - Renderer: Path Tracer - Acceleration: CPU

Zstd Compression

Compression Level: 12 - Decompression Speed

JPEG-XL libjxl

Input: JPEG - Quality: 90

OpenFOAM

Input: drivaerFastback, Small Mesh Size - Mesh Time

Timed FFmpeg Compilation

Time To Compile

Kvazaar

Video Input: Bosphorus 4K - Video Preset: Medium

ClickHouse

100M Rows Hits Dataset, First Run / Cold Cache

ONNX Runtime

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

uvg266

Video Input: Bosphorus 1080p - Video Preset: Ultra Fast

Parallel BZIP2 Compression

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

Blender

Blend File: Fishy Cat - Compute: CPU-Only

OpenSSL

Algorithm: RSA4096

Zstd Compression

Compression Level: 12 - Compression Speed

TensorFlow

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

Z3 Theorem Prover

SMT File: 2.smt2

MariaDB

Test: oltp_update_index - Threads: 64

Timed PHP Compilation

Time To Compile

ASKAP

Test: tConvolve MPI - Gridding

Zstd Compression

Compression Level: 8 - Compression Speed