Ryzen 9 3900XT Ubuntu 20.04 LTS vs. Ubuntu 20.10

Benchmarks by Michael Larabel.

HTML result view exported from: https://openbenchmarking.org/result/2010058-FI-2010031PT68&grs&sor.

oneDNN

Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPU

oneDNN

Harness: Deconvolution Batch deconv_1d - Data Type: f32 - Engine: CPU

John The Ripper

Test: MD5

oneDNN

Harness: IP Batch 1D - Data Type: u8s8f32 - Engine: CPU

TTSIOD 3D Renderer

Phong Rendering With Soft-Shadow Mapping

Hugin

Panorama Photo Assistant + Stitching Time

oneDNN

Harness: Deconvolution Batch deconv_1d - Data Type: u8s8f32 - Engine: CPU

oneDNN

Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPU

TensorFlow Lite

Model: NASNet Mobile

oneDNN

Harness: IP Batch 1D - Data Type: f32 - Engine: CPU

RealSR-NCNN

Scale: 4x - TAA: No

oneDNN

Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPU

Polyhedron Fortran Benchmarks

Benchmark: tfft2

oneDNN

Harness: IP Batch All - Data Type: f32 - Engine: CPU

John The Ripper

Test: Blowfish

Perl Benchmarks

Test: Interpreter

PyPerformance

Benchmark: python_startup

TensorFlow Lite

Model: Mobilenet Float

oneDNN

Harness: Convolution Batch Shapes Auto - Data Type: u8s8f32 - Engine: CPU

TensorFlow Lite

Model: Mobilenet Quant

Smallpt

Global Illumination Renderer; 128 Samples

G'MIC

Test: 3D Elevated Function In Random Colors, 100 Times

TNN

Target: CPU - Model: MobileNet v2

GraphicsMagick

Operation: Resizing

TensorFlow Lite

Model: SqueezeNet

oneDNN

Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPU

RawTherapee

Total Benchmark Time

Mobile Neural Network

Model: inception-v3

Mobile Neural Network

Model: SqueezeNetV1.0

TensorFlow Lite

Model: Inception ResNet V2

Polyhedron Fortran Benchmarks

Benchmark: air

Intel Open Image Denoise

Scene: Memorial

GraphicsMagick

Operation: Swirl

Mobile Neural Network

Model: MobileNetV2_224

Hackbench

Count: 4 - Type: Process

TensorFlow Lite

Model: Inception V4

Mobile Neural Network

Model: resnet-v2-50

Hackbench

Count: 32 - Type: Process

Mlpack Benchmark

Benchmark: scikit_qda

Timed HMMer Search

Pfam Database Search

GraphicsMagick

Operation: HWB Color Space

Hackbench

Count: 16 - Type: Process

PyPerformance

Benchmark: json_loads

Hackbench

Count: 8 - Type: Thread

Gzip Compression

Linux Source Tree Archiving To .tar.gz

Hackbench

Count: 16 - Type: Thread

Hackbench

Count: 8 - Type: Process

Mobile Neural Network

Model: mobilenet-v1-1.0

Hackbench

Count: 4 - Type: Thread

Caffe

Model: AlexNet - Acceleration: CPU - Iterations: 100

Caffe

Model: AlexNet - Acceleration: CPU - Iterations: 200

Go Benchmarks

Test: http

GraphicsMagick

Operation: Sharpen

Scikit-Learn

Mlpack Benchmark

Benchmark: scikit_ica

CacheBench

Test: Read

Numpy Benchmark

Hackbench

Count: 2 - Type: Process

Go Benchmarks

Test: garbage

glibc bench

Benchmark: sin

NGINX Benchmark

Static Web Page Serving

GraphicsMagick

Operation: Enhanced

Hackbench

Count: 1 - Type: Process

librsvg

Operation: SVG Files To PNG

OpenVKL

Benchmark: vklBenchmark

GraphicsMagick

Operation: Noise-Gaussian

Hackbench

Count: 1 - Type: Thread

FFTW

Build: Float + SSE - Size: 2D FFT Size 4096

PyPerformance

Benchmark: nbody

glibc bench

Benchmark: cos

LevelDB

Benchmark: Overwrite

LevelDB

Benchmark: Random Fill

LevelDB

Benchmark: Overwrite

LevelDB

Benchmark: Random Fill

Caffe

Model: GoogleNet - Acceleration: CPU - Iterations: 100

LevelDB

Benchmark: Fill Sync

Polyhedron Fortran Benchmarks

Benchmark: induct2

PyPerformance

Benchmark: crypto_pyaes

Polyhedron Fortran Benchmarks

Benchmark: rnflow

LAME MP3 Encoding

WAV To MP3

GPAW

Input: Carbon Nanotube

LuaJIT

Test: Fast Fourier Transform

PyPerformance

Benchmark: pickle_pure_python

LevelDB

Benchmark: Sequential Fill

LevelDB

Benchmark: Sequential Fill

Caffe

Model: GoogleNet - Acceleration: CPU - Iterations: 200

Go Benchmarks

Test: build

POV-Ray

Trace Time

LuaJIT

Test: Jacobi Successive Over-Relaxation

glibc bench

Benchmark: ffs

Java Gradle Build

Gradle Build: Reactor

Java SciMark

Computational Test: Jacobi Successive Over-Relaxation

Timed MAFFT Alignment

Multiple Sequence Alignment - LSU RNA

Basis Universal

Settings: UASTC Level 2

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon

SVT-HEVC

1080p 8-bit YUV To HEVC Video Encode

GIMP

Test: resize

LevelDB

Benchmark: Random Delete

Embree

Binary: Pathtracer - Model: Crown

ASKAP

Test: tConvolve MPI - Degridding

AOM AV1

Encoder Mode: Speed 6 Two-Pass

Embree

Binary: Pathtracer - Model: Asian Dragon

Embree

Binary: Pathtracer ISPC - Model: Asian Dragon Obj

SVT-VP9

Tuning: Visual Quality Optimized - Input: Bosphorus 1080p

PHPBench

PHP Benchmark Suite

ASKAP

Test: tConvolve MT - Degridding

glibc bench

Benchmark: pthread_once

Botan

Test: KASUMI

libjpeg-turbo tjbench

Test: Decompression Throughput

Embree

Binary: Pathtracer ISPC - Model: Crown

Selenium

Benchmark: WASM collisionDetection - Browser: Firefox

Polyhedron Fortran Benchmarks

Benchmark: doduc

m-queens

Time To Solve

Ogg Encoding

WAV To Ogg

LuaJIT

Test: Monte Carlo

PyPerformance

Benchmark: django_template

LuaJIT

Test: Composite

Java SciMark

Computational Test: Dense LU Matrix Factorization

Crafty

Elapsed Time

Basis Universal

Settings: UASTC Level 3

Apache CouchDB

Bulk Size: 100 - Inserts: 1000 - Rounds: 24

Selenium

Benchmark: Basemark - Browser: Firefox

Embree

Binary: Pathtracer - Model: Asian Dragon Obj

Coremark

CoreMark Size 666 - Iterations Per Second

glibc bench

Benchmark: modf

OSPray

Demo: XFrog Forest - Renderer: SciVis

Appleseed

Scene: Disney Material

OSPray

Demo: NASA Streamlines - Renderer: Path Tracer

Java SciMark

Computational Test: Composite

Basis Universal

Settings: UASTC Level 2 + RDO Post-Processing

LuxCoreRender

Scene: DLSC

glibc bench

Benchmark: ffsll