Ryzen 9 3900XT Ubuntu 20.04 LTS vs. Ubuntu 20.10
Benchmarks by Michael Larabel.
HTML result view exported from: https://openbenchmarking.org/result/2010126-FI-2010058FI68&grs.
oneDNN
Harness: IP Batch 1D - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPU
oneDNN
Harness: Deconvolution Batch deconv_1d - Data Type: f32 - Engine: CPU
Hugin
Panorama Photo Assistant + Stitching Time
John The Ripper
Test: MD5
oneDNN
Harness: Deconvolution Batch deconv_1d - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: IP Batch 1D - Data Type: f32 - Engine: CPU
TTSIOD 3D Renderer
Phong Rendering With Soft-Shadow Mapping
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: u8s8f32 - Engine: CPU
LevelDB
Benchmark: Fill Sync
TensorFlow Lite
Model: NASNet Mobile
Systemd Total Boot Time
Test: Kernel
oneDNN
Harness: IP Batch All - Data Type: u8s8f32 - Engine: CPU
LevelDB
Benchmark: Fill Sync
RealSR-NCNN
Scale: 4x - TAA: No
SQLite
Threads / Copies: 1
Polyhedron Fortran Benchmarks
Benchmark: tfft2
Mobile Neural Network
Model: mobilenet-v1-1.0
LuaJIT
Test: Fast Fourier Transform
John The Ripper
Test: Blowfish
oneDNN
Harness: IP Batch All - Data Type: f32 - Engine: CPU
ebizzy
G'MIC
Test: 3D Elevated Function In Random Colors, 100 Times
Mobile Neural Network
Model: SqueezeNetV1.0
CP2K Molecular Dynamics
Fayalite-FIST Data
PolyBench-C
Test: Covariance Computation
PolyBench-C
Test: Correlation Computation
Mlpack Benchmark
Benchmark: scikit_linearridgeregression
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: u8s8f32 - Engine: CPU
Hackbench
Count: 8 - Type: Process
Hackbench
Count: 8 - Type: Thread
VkFFT
LevelDB
Benchmark: Seek Random
PyPerformance
Benchmark: python_startup
ParaView
Test: Many Spheres - Resolution: 1920 x 1080
ParaView
Test: Many Spheres - Resolution: 1920 x 1080
LevelDB
Benchmark: Sequential Fill
LevelDB
Benchmark: Random Fill
LevelDB
Benchmark: Overwrite
CLOMP
Static OMP Speedup
Perl Benchmarks
Test: Interpreter
Mobile Neural Network
Model: MobileNetV2_224
GLmark2
Resolution: 3840 x 2160
OSPray
Demo: Magnetic Reconnection - Renderer: SciVis
MPV
Video Input: Big Buck Bunny Sunflower 1080p - Decode: Software Only
TensorFlow Lite
Model: Mobilenet Float
LevelDB
Benchmark: Random Delete
MPV
Video Input: Big Buck Bunny Sunflower 4K - Decode: Software Only
Basemark GPU
Renderer: Vulkan - Resolution: 3840 x 2160 - Graphics Preset: High
ParaView
Test: Wavelet Contour - Resolution: 1920 x 1080
ParaView
Test: Wavelet Contour - Resolution: 1920 x 1080
Hackbench
Count: 1 - Type: Process
TensorFlow Lite
Model: Mobilenet Quant
Polyhedron Fortran Benchmarks
Benchmark: capacita
Basemark GPU
Renderer: Vulkan - Resolution: 2560 x 1440 - Graphics Preset: High
GLmark2
Resolution: 2560 x 1440
Mobile Neural Network
Model: resnet-v2-50
Caffe
Model: AlexNet - Acceleration: CPU - Iterations: 100
Caffe
Model: AlexNet - Acceleration: CPU - Iterations: 200
DaCapo Benchmark
Java Test: Tradebeans
Smallpt
Global Illumination Renderer; 128 Samples
Zstd Compression
Compression Level: 3
Hackbench
Count: 16 - Type: Process
TNN
Target: CPU - Model: MobileNet v2
Basemark GPU
Renderer: Vulkan - Resolution: 1920 x 1080 - Graphics Preset: High
Hackbench
Count: 1 - Type: Thread
Unigine Superposition
Resolution: 1920 x 1080 - Mode: Fullscreen - Quality: Medium - Renderer: OpenGL
LevelDB
Benchmark: Random Read
Basemark GPU
Renderer: Vulkan - Resolution: 2560 x 1440 - Graphics Preset: Medium
Java 2D Microbenchmark
Rendering Test: Vector Graphics Rendering
LevelDB
Benchmark: Sequential Fill
LevelDB
Benchmark: Random Fill
LevelDB
Benchmark: Hot Read
Xonotic
Resolution: 3840 x 2160 - Effects Quality: Ultra
Systemd Total Boot Time
Test: Firmware
Basemark GPU
Renderer: Vulkan - Resolution: 1920 x 1080 - Graphics Preset: Medium
Sudokut
Total Time
Go Benchmarks
Test: http
Basemark GPU
Renderer: Vulkan - Resolution: 3840 x 2160 - Graphics Preset: Medium
LevelDB
Benchmark: Overwrite
GraphicsMagick
Operation: Swirl
Xonotic
Resolution: 3840 x 2160 - Effects Quality: Ultimate
Bork File Encrypter
File Encryption Time
GLmark2
Resolution: 1920 x 1080
glibc bench
Benchmark: sqrt
Polyhedron Fortran Benchmarks
Benchmark: air
Polyhedron Fortran Benchmarks
Benchmark: induct2
CloverLeaf
Lagrangian-Eulerian Hydrodynamics
Hackbench
Count: 32 - Type: Process
oneDNN
Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPU
Tesseract
Resolution: 3840 x 2160
PlaidML
FP16: No - Mode: Inference - Network: IMDB LSTM - Device: CPU
Apache CouchDB
Bulk Size: 100 - Inserts: 1000 - Rounds: 24
GraphicsMagick
Operation: Resizing
FFTW
Build: Float + SSE - Size: 2D FFT Size 4096
FFTE
N=256, 3D Complex FFT Routine
Mobile Neural Network
Model: inception-v3
CacheBench
Test: Read
Algebraic Multi-Grid Benchmark
Java SciMark
Computational Test: Monte Carlo
Caffe
Model: GoogleNet - Acceleration: CPU - Iterations: 200
KeyDB
Caffe
Model: GoogleNet - Acceleration: CPU - Iterations: 100
Xonotic
Resolution: 3840 x 2160 - Effects Quality: High
OSPray
Demo: San Miguel - Renderer: Path Tracer
Unigine Superposition
Resolution: 1920 x 1080 - Mode: Fullscreen - Quality: Low - Renderer: OpenGL
Polyhedron Fortran Benchmarks
Benchmark: linpk
TensorFlow Lite
Model: SqueezeNet
GraphicsMagick
Operation: Rotate
dav1d
Video Input: Summer Nature 1080p
GPAW
Input: Carbon Nanotube
dav1d
Video Input: Chimera 1080p
Nettle
Test: poly1305-aes
glibc bench
Benchmark: ffsll
glibc bench
Benchmark: pthread_once
Hierarchical INTegration
Test: FLOAT
NeatBench
Acceleration: CPU
GnuPG
2GB File Encryption
RawTherapee
Total Benchmark Time
PHPBench
PHP Benchmark Suite
Nettle
Test: chacha
glibc bench
Benchmark: ffs
AOM AV1
Encoder Mode: Speed 8 Realtime
AOM AV1
Encoder Mode: Speed 6 Realtime
glibc bench
Benchmark: modf
Selenium
Benchmark: StyleBench - Browser: Firefox
PyPerformance
Benchmark: crypto_pyaes
Intel Open Image Denoise
Scene: Memorial
InfluxDB
Concurrent Streams: 4 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
Embree
Binary: Pathtracer ISPC - Model: Asian Dragon
glibc bench
Benchmark: exp
Nettle
Test: aes256
PyPerformance
Benchmark: json_loads
m-queens
Time To Solve
Polyhedron Fortran Benchmarks
Benchmark: fatigue2
Zstd Compression
Compression Level: 19
ASKAP
Test: tConvolve MT - Gridding
librsvg
Operation: SVG Files To PNG
SVT-AV1
Encoder Mode: Enc Mode 0 - Input: 1080p
Gzip Compression
Linux Source Tree Archiving To .tar.gz
PyPerformance
Benchmark: pickle_pure_python
Java Gradle Build
Gradle Build: Reactor
Embree
Binary: Pathtracer ISPC - Model: Asian Dragon Obj
Go Benchmarks
Test: build
Redis
Test: LPUSH
Polyhedron Fortran Benchmarks
Benchmark: ac
TensorFlow Lite
Model: Inception ResNet V2
InfluxDB
Concurrent Streams: 64 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
ASKAP
Test: tConvolve MPI - Gridding
Selenium
Benchmark: Maze Solver - Browser: Firefox
PyPerformance
Benchmark: chaos
PyBench
Total For Average Test Times
OSPray
Demo: Magnetic Reconnection - Renderer: Path Tracer
Perl Benchmarks
Test: Pod2html
C-Ray
Total Time - 4K, 16 Rays Per Pixel
PyPerformance
Benchmark: float
DeepSpeech
Acceleration: CPU
Monte Carlo Simulations of Ionised Nebulae
Input: Dust 2D tau100.0
Java 2D Microbenchmark
Rendering Test: All Rendering Tests
OSPray
Demo: San Miguel - Renderer: SciVis
ASTC Encoder
Preset: Exhaustive
InfluxDB
Concurrent Streams: 1024 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000
PyPerformance
Benchmark: raytrace
Mlpack Benchmark
Benchmark: scikit_qda
Botan
Test: Blowfish
Darktable
Test: Boat - Acceleration: CPU-only
LuaJIT
Test: Monte Carlo
TSCP
AI Chess Performance
Apache Siege
Concurrent Users: 10
SQLite Speedtest
Timed Time - Size 1,000
Numpy Benchmark
NAMD
ATPase Simulation - 327,506 Atoms
Inkscape
Operation: SVG Files To PNG
ASTC Encoder
Preset: Thorough
Selenium
Benchmark: WASM collisionDetection - Browser: Firefox
LuaJIT
Test: Sparse Matrix Multiply
TensorFlow Lite
Model: Inception V4
RNNoise
Embree
Binary: Pathtracer ISPC - Model: Crown
Crafty
Elapsed Time
Mlpack Benchmark
Benchmark: scikit_ica
7-Zip Compression
Compress Speed Test
PyPerformance
Benchmark: go
Botan
Test: CAST-256
LuaJIT
Test: Composite
dav1d
Video Input: Chimera 1080p 10-bit
x265
H.265 1080p Video Encoding
DaCapo Benchmark
Java Test: Jython
Botan
Test: KASUMI
G'MIC
Test: Plotting Isosurface Of A 3D Volume, 1000 Times
PyPerformance
Benchmark: django_template
Tesseract OCR
Time To OCR 7 Images
AOBench
Size: 2048 x 2048 - Total Time
AOM AV1
Encoder Mode: Speed 6 Two-Pass
Timed FFmpeg Compilation
Time To Compile
Botan
Test: Twofish
Timed Apache Compilation
Time To Compile
GROMACS
Water Benchmark
GIMP
Test: unsharp-mask
Tesseract
Resolution: 1920 x 1080
High Performance Conjugate Gradient
SciMark
Computational Test: Dense LU Matrix Factorization
Polyhedron Fortran Benchmarks
Benchmark: rnflow
Botan
Test: AES-256
PlaidML
FP16: No - Mode: Inference - Network: Inception V3 - Device: CPU
OSPray
Demo: XFrog Forest - Renderer: Path Tracer
DaCapo Benchmark
Java Test: Tradesoap
Himeno Benchmark
Poisson Pressure Solver
Timed HMMer Search
Pfam Database Search
LuaJIT
Test: Jacobi Successive Over-Relaxation
Memtier_benchmark
Protocol: Redis
Selenium
Benchmark: Octane - Browser: Firefox
PyPerformance
Benchmark: regex_compile
GraphicsMagick
Operation: HWB Color Space
GNU Octave Benchmark
LuxCoreRender
Scene: DLSC
GraphicsMagick
Operation: Noise-Gaussian
Basis Universal
Settings: UASTC Level 3
Systemd Total Boot Time
Test: Userspace
TNN
Target: CPU - Model: SqueezeNet v1.1
GIMP
Test: auto-levels
Selenium
Benchmark: ARES-6 - Browser: Firefox
SciMark
Computational Test: Sparse Matrix Multiply
Appleseed
Scene: Emily
Systemd Total Boot Time
Test: Loader
N-Queens
Elapsed Time
Appleseed
Scene: Disney Material
Selenium
Benchmark: PSPDFKit WASM - Browser: Firefox
PyPerformance
Benchmark: nbody
Nebular Empirical Analysis Tool
CacheBench
Test: Write
SciMark
Computational Test: Composite
Xonotic
Resolution: 3840 x 2160 - Effects Quality: Low
XZ Compression
Compressing ubuntu-16.04.3-server-i386.img, Compression Level 9
OpenVKL
Benchmark: vklBenchmark
LuxCoreRender
Scene: Rainbow Colors and Prism
PlaidML
FP16: No - Mode: Inference - Network: DenseNet 201 - Device: CPU
Hackbench
Count: 2 - Type: Process
AOM AV1
Encoder Mode: Speed 4 Two-Pass
Basis Universal
Settings: UASTC Level 2
LAMMPS Molecular Dynamics Simulator
Model: Rhodopsin Protein
LuaJIT
Test: Dense LU Matrix Factorization
CacheBench
Test: Read / Modify / Write
Primesieve
1e12 Prime Number Generation
OSPray
Demo: XFrog Forest - Renderer: SciVis
Timed Linux Kernel Compilation
Time To Compile
SciMark
Computational Test: Monte Carlo
Java SciMark
Computational Test: Fast Fourier Transform
GraphicsMagick
Operation: Enhanced
Darktable
Test: Server Rack - Acceleration: CPU-only
SciMark
Computational Test: Jacobi Successive Over-Relaxation
POV-Ray
Trace Time
Git
Time To Complete Common Git Commands
Stockfish
Total Time
PyPerformance
Benchmark: 2to3
FFTW
Build: Stock - Size: 2D FFT Size 4096
SVT-HEVC
1080p 8-bit YUV To HEVC Video Encode
BYTE Unix Benchmark
Computational Test: Dhrystone 2
Timed MrBayes Analysis
Primate Phylogeny Analysis
GraphicsMagick
Operation: Sharpen
Polyhedron Fortran Benchmarks
Benchmark: aermod
glibc bench
Benchmark: sinh
glibc bench
Benchmark: sin
Minion
Benchmark: Graceful
glibc bench
Benchmark: asinh
Scikit-Learn
WireGuard + Linux Networking Stack Stress Test
Go Benchmarks
Test: garbage
Bullet Physics Engine
Test: 1000 Stack
Build2
Time To Compile
GIMP
Test: resize
Compile Bench
Test: Initial Create
Selenium
Benchmark: Jetstream - Browser: Firefox
ET: Legacy
Renderer: Renderer2 - Resolution: 3840 x 2160
Polyhedron Fortran Benchmarks
Benchmark: gas_dyn2
ASTC Encoder
Preset: Fast
Polyhedron Fortran Benchmarks
Benchmark: mdbx
Mlpack Benchmark
Benchmark: scikit_svm
Embree
Binary: Pathtracer - Model: Asian Dragon
Timed MAFFT Alignment
Multiple Sequence Alignment - LSU RNA
NGINX Benchmark
Static Web Page Serving
Ogg Encoding
WAV To Ogg
OSPray
Demo: NASA Streamlines - Renderer: Path Tracer
SciMark
Computational Test: Fast Fourier Transform
Selenium
Benchmark: Basemark - Browser: Firefox
Timed ImageMagick Compilation
Time To Compile
PyPerformance
Benchmark: pathlib
Chaos Group V-RAY
Mode: CPU
glibc bench
Benchmark: cos
Selenium
Benchmark: WebXPRT - Browser: Firefox
PolyBench-C
Test: 3 Matrix Multiplications
ASKAP
Test: tConvolve MPI - Degridding
Radiance Benchmark
Test: SMP Parallel
Polyhedron Fortran Benchmarks
Benchmark: test_fpu2
Montage Astronomical Image Mosaic Engine
Mosaic of M17, K band, 1.5 deg x 1.5 deg
Bullet Physics Engine
Test: 1000 Convex
YafaRay
Total Time For Sample Scene
LibreOffice
Test: 20 Documents To PDF
Dolfyn
Computational Fluid Dynamics
Java SciMark
Computational Test: Dense LU Matrix Factorization
Embree
Binary: Pathtracer - Model: Asian Dragon Obj
Cython benchmark
Polyhedron Fortran Benchmarks
Benchmark: mp_prop_design
Selenium
Benchmark: CanvasMark - Browser: Firefox
OpenSSL
RSA 4096-bit Performance
libavif avifenc
Encoder Speed: 0
GIMP
Test: rotate
libjpeg-turbo tjbench
Test: Decompression Throughput
Incompact3D
Input: Cylinder
Selenium
Benchmark: Jetstream 2 - Browser: Firefox
Bullet Physics Engine
Test: 3000 Fall
dav1d
Video Input: Summer Nature 4K
LAME MP3 Encoding
WAV To MP3
Basis Universal
Settings: UASTC Level 0
Minion
Benchmark: Quasigroup
Hackbench
Count: 2 - Type: Thread
SVT-AV1
Encoder Mode: Enc Mode 8 - Input: 1080p
Radiance Benchmark
Test: Serial
SVT-VP9
Tuning: VMAF Optimized - Input: Bosphorus 1080p
Polyhedron Fortran Benchmarks
Benchmark: doduc
glibc bench
Benchmark: atanh
miniFE
Problem Size: Small
ASKAP
Test: tConvolve MT - Degridding
AOM AV1
Encoder Mode: Speed 0 Two-Pass
Basis Universal
Settings: UASTC Level 2 + RDO Post-Processing
Java SciMark
Computational Test: Jacobi Successive Over-Relaxation
Bullet Physics Engine
Test: Prim Trimesh
SVT-VP9
Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p
OSPray
Demo: NASA Streamlines - Renderer: SciVis
libavif avifenc
Encoder Speed: 2
Polyhedron Fortran Benchmarks
Benchmark: protein
Go Benchmarks
Test: json
FLAC Audio Encoding
WAV To FLAC
C-Blosc
Compressor: blosclz
ASKAP
Test: tConvolve OpenMP - Degridding
Appleseed
Scene: Material Tester
Embree
Binary: Pathtracer - Model: Crown
SVT-VP9
Tuning: Visual Quality Optimized - Input: Bosphorus 1080p
Tesseract
Resolution: 2560 x 1440
PlaidML
FP16: No - Mode: Inference - Network: Mobilenet - Device: CPU
Darktable
Test: Server Room - Acceleration: CPU-only
Selenium
Benchmark: Speedometer - Browser: Firefox
SVT-AV1
Encoder Mode: Enc Mode 4 - Input: 1080p
Bullet Physics Engine
Test: Convex Trimesh
Java SciMark
Computational Test: Composite
Coremark
CoreMark Size 666 - Iterations Per Second
glibc bench
Benchmark: log2
asmFish
1024 Hash Memory, 26 Depth
Compile Bench
Test: Read Compiled Tree
libavif avifenc
Encoder Speed: 10
libavif avifenc
Encoder Speed: 8
Polyhedron Fortran Benchmarks
Benchmark: channel2
Bullet Physics Engine
Test: Raytests
Java 2D Microbenchmark
Rendering Test: Image Rendering
Systemd Total Boot Time
Test: Total
PlaidML
FP16: No - Mode: Inference - Network: ResNet 50 - Device: CPU
Selenium
Benchmark: WASM imageConvolute - Browser: Firefox
OCRMyPDF
Processing 60 Page PDF Document
Java SciMark
Computational Test: Sparse Matrix Multiply
Basis Universal
Settings: ETC1S
Selenium
Benchmark: Kraken - Browser: Firefox
PlaidML
FP16: No - Mode: Inference - Network: VGG16 - Device: CPU
Darktable
Test: Masskrug - Acceleration: CPU-only
glibc bench
Benchmark: sincos
Apache Benchmark
Static Web Page Serving
PlaidML
FP16: No - Mode: Inference - Network: VGG19 - Device: CPU
ASTC Encoder
Preset: Medium
Minion
Benchmark: Solitaire
glibc bench
Benchmark: tanh
Bullet Physics Engine
Test: 136 Ragdolls
BRL-CAD
VGR Performance Metric
Nettle
Test: sha512
OCRMyPDF
Processing 60 Page PDF Document
Darktable
Test: Server Room - Acceleration: CPU-only
Darktable
Test: Server Rack - Acceleration: CPU-only
Darktable
Test: Masskrug - Acceleration: CPU-only
Darktable
Test: Boat - Acceleration: CPU-only
LibRaw
Post-Processing Benchmark
Selenium
Benchmark: MotionMark - Browser: Firefox
Apache Siege
Concurrent Users: 250
Apache Siege
Concurrent Users: 200
Apache Siege
Concurrent Users: 100
Apache Siege
Concurrent Users: 50
ctx_clock
Context Switch Time
Redis
Test: SET
Redis
Test: GET
Redis
Test: SADD
Redis
Test: LPOP
ASKAP
Test: tConvolve OpenMP - Gridding
Hackbench
Count: 4 - Type: Process
Hackbench
Count: 16 - Type: Thread
Hackbench
Count: 4 - Type: Thread
oneDNN
Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPU
oneDNN
Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPU
oneDNN
Harness: Deconvolution Batch deconv_3d - Data Type: u8s8f32 - Engine: CPU
oneDNN
Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPU
DaCapo Benchmark
Java Test: H2
Java 2D Microbenchmark
Rendering Test: Text Rendering
ParaView
Test: Wavelet Volume - Resolution: 1920 x 1080
ParaView
Test: Wavelet Volume - Resolution: 1920 x 1080
Compile Bench
Test: Compile
IOR
Read Test
IOR
Write Test
FS-Mark
Test: 4000 Files, 32 Sub Dirs, 1MB Size
FS-Mark
Test: 5000 Files, 1MB Size, 4 Threads
FS-Mark
Test: 1000 Files, 1MB Size
Phoronix Test Suite v10.8.5