System76 Thelio Major 3990X Pop OS 21.10

AMD Ryzen Threadripper 3990X benchmarking of Pop OS 21.10 versus 20.04 LTS by Michael Larabel for a future article.

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2112296-NE-2010258FI98
Jump To Table - Results

View

Do Not Show Noisy Results
Do Not Show Results With Incomplete Data
Do Not Show Results With Little Change/Spread
List Notable Results

Limit displaying results to tests within:

Audio Encoding 2 Tests
Timed Code Compilation 4 Tests
C/C++ Compiler Tests 15 Tests
CPU Massive 17 Tests
Creator Workloads 23 Tests
Database Test Suite 5 Tests
Desktop Graphics 4 Tests
Encoding 5 Tests
Fortran Tests 4 Tests
Game Development 4 Tests
HPC - High Performance Computing 13 Tests
Imaging 7 Tests
Common Kernel Benchmarks 3 Tests
Machine Learning 7 Tests
Molecular Dynamics 2 Tests
MPI Benchmarks 2 Tests
Multi-Core 19 Tests
NVIDIA GPU Compute 5 Tests
Intel oneAPI 4 Tests
OpenMPI Tests 2 Tests
Productivity 2 Tests
Programmer / Developer System Benchmarks 9 Tests
Python 3 Tests
Raytracing 2 Tests
Renderers 4 Tests
Scientific Computing 6 Tests
Server 6 Tests
Server CPU Tests 13 Tests
Single-Threaded 11 Tests
Speech 2 Tests
Telephony 2 Tests
Video Encoding 3 Tests
Vulkan Compute 3 Tests
Common Workstation Benchmarks 4 Tests

Statistics

Show Overall Harmonic Mean(s)
Show Overall Geometric Mean
Show Geometric Means Per-Suite/Category
Show Wins / Losses Counts (Pie Chart)
Normalize Results
Remove Outliers Before Calculating Averages

Graph Settings

Force Line Graphs Where Applicable
Convert To Scalar Where Applicable
Prefer Vertical Bar Graphs

Multi-Way Comparison

Condense Multi-Option Tests Into Single Result Graphs

Table

Show Detailed System Result Table

Run Management

Highlight
Result
Hide
Result
Result
Identifier
Performance Per
Dollar
Date
Run
  Test
  Duration
Pop OS 20.04
October 23 2020
  11 Hours, 45 Minutes
Pop OS 21.10
December 28 2021
  9 Hours, 58 Minutes
Invert Hiding All Results Option
  10 Hours, 51 Minutes
Only show results matching title/arguments (delimit multiple options with a comma):
Do not show results matching title/arguments (delimit multiple options with a comma):


System76 Thelio Major 3990X Pop OS 21.10 Suite 1.0.0 System Test suite extracted from System76 Thelio Major 3990X Pop OS 21.10. pts/polyhedron-1.0.1 tfft2 Benchmark: tfft2 pts/polyhedron-1.0.1 gas_dyn2 Benchmark: gas_dyn2 pts/j2dbench-1.1.2 TEST_ALL Rendering Test: All Rendering Tests pts/polyhedron-1.0.1 test_fpu2 Benchmark: test_fpu2 pts/openvkl-1.0.0 vklBenchmark Benchmark: vklBenchmark pts/j2dbench-1.1.2 TEST_GRAPHICS Rendering Test: Vector Graphics Rendering pts/lammps-1.2.1 benchmark_20k_atoms.in Model: 20k Atoms pts/byte-1.2.2 TEST_DHRY2 Computational Test: Dhrystone 2 pts/leveldb-1.0.0 --benchmarks=fillseq --num=500000 Benchmark: Sequential Fill pts/leveldb-1.0.0 --benchmarks=deleterandom --num=500000 Benchmark: Random Delete pts/hint-1.0.3 FLOAT Test: FLOAT pts/mocassin-1.0.0 Input: Dust 2D tau100.0 pts/polyhedron-1.0.1 channel2 Benchmark: channel2 pts/polyhedron-1.0.1 mp_prop_design Benchmark: mp_prop_design pts/tensorflow-lite-1.0.0 --graph=inception_v4.tflite Model: Inception V4 pts/kripke-1.0.0 pts/leveldb-1.0.0 --benchmarks=seekrandom --num=1000000 Benchmark: Seek Random pts/polyhedron-1.0.1 fatigue2 Benchmark: fatigue2 pts/hmmer-1.2.0 Pfam Database Search system/gmic-1.0.0 repeat 1000 1000,1,1,2 fill "X=3*(x-500)/500;X^2*sin(3*X^2)+if(c==0,u(0,-1),cos(X*10))" output output.png done Test: 2D Function Plotting, 1000 Times pts/caffe-1.5.0 --model=../models/bvlc_googlenet/deploy.prototxt -iterations 100 Model: GoogleNet - Acceleration: CPU - Iterations: 100 pts/j2dbench-1.1.2 TEST_IMAGES Rendering Test: Image Rendering pts/tensorflow-lite-1.0.0 --graph=mobilenet_v1_1.0_224.tflite Model: Mobilenet Float pts/blender-1.8.0 -b ../barbershop_interior_gpu.blend -o output.test -x 1 -F JPEG -f 1 NONE Blend File: Barbershop - Compute: CPU-Only pts/mlpack-1.0.2 SCIKIT_LINEARRIDGEREGRESSION Benchmark: scikit_linearridgeregression pts/glmark2-1.2.0 -s 1920x1080 Resolution: 1920 x 1080 pts/glmark2-1.2.0 -s 3840x2160 Resolution: 3840 x 2160 pts/leveldb-1.0.0 --benchmarks=readrandom --num=1000000 Benchmark: Random Read pts/ncnn-1.0.3 -1 Target: CPU - Model: yolov4-tiny pts/ncnn-1.0.3 -1 Target: CPU - Model: alexnet pts/ncnn-1.0.3 -1 Target: CPU - Model: googlenet pts/ncnn-1.0.3 -1 Target: CPU - Model: mnasnet pts/ncnn-1.0.3 -1 Target: CPU - Model: shufflenet-v2 pts/ncnn-1.0.3 -1 Target: CPU - Model: squeezenet pts/sqlite-2.1.0 8 Threads / Copies: 8 pts/blender-1.8.0 -b ../pavillon_barcelone_gpu.blend -o output.test -x 1 -F JPEG -f 1 NONE Blend File: Pabellon Barcelona - Compute: CPU-Only pts/appleseed-1.0.1 material_tester_ambient_occlusion.appleseed Scene: Material Tester pts/couchdb-1.0.1 100 1000 24 Bulk Size: 100 - Inserts: 1000 - Rounds: 24 pts/tesseract-1.1.0 -w1920 -h1080 Resolution: 1920 x 1080 pts/mlpack-1.0.2 SCIKIT_QDA Benchmark: scikit_qda system/gegl-1.0.0 cartoon Operation: Cartoon pts/appleseed-1.0.1 emily.appleseed Scene: Emily pts/realsr-ncnn-1.0.0 -s 4 -x Scale: 4x - TAA: Yes pts/blender-1.8.0 -b ../classroom_gpu.blend -o output.test -x 1 -F JPEG -f 1 NONE Blend File: Classroom - Compute: CPU-Only system/gmic-1.0.0 repeat 100 128,128,1,3,"u(0,255)" plasma 10,3 blur 4 sharpen 10000 elevation3d[-1] "'X=(x-64)/6;Y=(y-64)/6;100*exp(-(X^2+Y^2)/30)*abs(cos(X)*sin(Y))'" done output output.png Test: 3D Elevated Function In Random Colors, 100 Times pts/pyperformance-1.0.2 python_startup Benchmark: python_startup pts/pyperformance-1.0.2 raytrace Benchmark: raytrace pts/polyhedron-1.0.1 induct2 Benchmark: induct2 pts/polyhedron-1.0.1 capacita Benchmark: capacita pts/influxdb-1.0.0 -c 1024 -b 10000 -t 2,5000,1 -p 10000 Concurrent Streams: 1024 - Batch Size: 10000 - Tags: 2,5000,1 - Points Per Series: 10000 pts/onednn-1.5.0 --ip --batch=inputs/ip/ip_all --cfg=f32 --engine=cpu Harness: IP Batch All - Data Type: f32 - Engine: CPU pts/sqlite-speedtest-1.0.0 Timed Time - Size 1,000 pts/leveldb-1.0.0 --benchmarks=fillrandom --num=100000 Benchmark: Random Fill pts/leveldb-1.0.0 --benchmarks=overwrite --num=100000 Benchmark: Overwrite pts/ospray-1.0.3 sanm/sanm.obj -vp 22.958788 3.204613 2.712676 -vu 0.000000 1.000000 0.000000 -vi 12.364944 0.176316 4.009342 -sg:sun:intensity=4.0 -sg:sun:direction=0,-1,0 -sg:bounce:intensity=0.0 --hdri-light rnl_probe.pfm -sg:hdri:intensity=1.25 --renderer pathtracer Demo: San Miguel - Renderer: Path Tracer pts/tensorflow-lite-1.0.0 --graph=inception_resnet_v2.tflite Model: Inception ResNet V2 pts/tensorflow-lite-1.0.0 --graph=nasnet_mobile.tflite Model: NASNet Mobile pts/tensorflow-lite-1.0.0 --graph=squeezenet.tflite Model: SqueezeNet system/gegl-1.0.0 wavelet-blur Operation: Wavelet Blur pts/tensorflow-lite-1.0.0 --graph=mobilenet_v1_1.0_224_quant.tflite Model: Mobilenet Quant pts/pyperformance-1.0.2 2to3 Benchmark: 2to3 pts/stockfish-1.1.1 Total Time system/gegl-1.0.0 color-enhance Operation: Color Enhance pts/caffe-1.5.0 --model=../models/bvlc_alexnet/deploy.prototxt -iterations 100 Model: AlexNet - Acceleration: CPU - Iterations: 100 pts/j2dbench-1.1.2 TEST_TEXT Rendering Test: Text Rendering pts/polyhedron-1.0.1 rnflow Benchmark: rnflow pts/mlpack-1.0.2 SCIKIT_ICA Benchmark: scikit_ica system/rawtherapee-1.0.1 Total Benchmark Time pts/pyperformance-1.0.2 go Benchmark: go pts/blender-1.8.0 -b ../fishy_cat_gpu.blend -o output.test -x 1 -F JPEG -f 1 NONE Blend File: Fishy Cat - Compute: CPU-Only pts/git-1.1.0 Time To Complete Common Git Commands system/hugin-1.0.0 Panorama Photo Assistant + Stitching Time pts/xonotic-1.5.1 +vid_width 3840 +vid_height 2160 +exec effects-ultimate.cfg Resolution: 3840 x 2160 - Effects Quality: Ultimate pts/polyhedron-1.0.1 protein Benchmark: protein pts/compress-zstd-1.2.1 -b19 Compression Level: 19 pts/ospray-1.0.3 sanm/sanm.obj -vp 22.958788 3.204613 2.712676 -vu 0.000000 1.000000 0.000000 -vi 12.364944 0.176316 4.009342 -sg:sun:intensity=4.0 -sg:sun:direction=0,-1,0 -sg:bounce:intensity=0.0 --hdri-light rnl_probe.pfm -sg:hdri:intensity=1.25 --renderer scivis Demo: San Miguel - Renderer: SciVis pts/ncnn-1.0.3 Target: Vulkan GPU - Model: resnet50 pts/ncnn-1.0.3 Target: Vulkan GPU - Model: yolov4-tiny pts/ncnn-1.0.3 Target: Vulkan GPU - Model: alexnet pts/ncnn-1.0.3 Target: Vulkan GPU - Model: resnet18 pts/ncnn-1.0.3 Target: Vulkan GPU - Model: vgg16 pts/ncnn-1.0.3 Target: Vulkan GPU - Model: googlenet pts/ncnn-1.0.3 Target: Vulkan GPU - Model: blazeface pts/ncnn-1.0.3 Target: Vulkan GPU - Model: efficientnet-b0 pts/ncnn-1.0.3 Target: Vulkan GPU - Model: mnasnet pts/ncnn-1.0.3 Target: Vulkan GPU - Model: shufflenet-v2 pts/ncnn-1.0.3 Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3 pts/ncnn-1.0.3 Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2 pts/ncnn-1.0.3 Target: Vulkan GPU - Model: mobilenet pts/ncnn-1.0.3 Target: Vulkan GPU - Model: squeezenet system/gegl-1.0.0 antialias Operation: Antialias system/gegl-1.0.0 rotate-on-center degrees=90 Operation: Rotate 90 Degrees pts/onednn-1.5.0 --deconv --batch=inputs/deconv/deconv_1d --cfg=f32 --engine=cpu Harness: Deconvolution Batch deconv_1d - Data Type: f32 - Engine: CPU pts/appleseed-1.0.1 disney_material_1.appleseed Scene: Disney Material pts/espeak-1.6.0 Text-To-Speech Synthesis pts/onednn-1.5.0 --rnn --batch=inputs/rnn/rnn_training --cfg=f32 --engine=cpu Harness: Recurrent Neural Network Training - Data Type: f32 - Engine: CPU pts/namd-1.2.1 ATPase Simulation - 327,506 Atoms pts/onednn-1.5.0 --rnn --batch=inputs/rnn/rnn_inference --cfg=f32 --engine=cpu Harness: Recurrent Neural Network Inference - Data Type: f32 - Engine: CPU pts/pyperformance-1.0.2 django_template Benchmark: django_template pts/xonotic-1.5.1 +vid_width 3840 +vid_height 2160 +exec effects-ultra.cfg Resolution: 3840 x 2160 - Effects Quality: Ultra pts/build-php-1.5.1 Time To Compile pts/webp-1.0.0 -q 100 -lossless -m 6 Encode Settings: Quality 100, Lossless, Highest Compression pts/blender-1.8.0 -b ../bmw27_gpu.blend -o output.test -x 1 -F JPEG -f 1 NONE Blend File: BMW27 - Compute: CPU-Only pts/ospray-1.0.3 xfrog-forest.xml --renderer pathtracer Demo: XFrog Forest - Renderer: Path Tracer pts/sqlite-2.1.0 1 Threads / Copies: 1 pts/pyperformance-1.0.2 regex_compile Benchmark: regex_compile pts/ospray-1.0.3 NASA-B-field-sun.osx --renderer pathtracer Demo: NASA Streamlines - Renderer: Path Tracer pts/tesseract-1.1.0 -w3840 -h2160 Resolution: 3840 x 2160 pts/xonotic-1.5.1 +vid_width 3840 +vid_height 2160 +exec effects-high.cfg Resolution: 3840 x 2160 - Effects Quality: High system/gegl-1.0.0 tile-glass tile-width=20 tile-height=20 Operation: Tile Glass pts/polyhedron-1.0.1 doduc Benchmark: doduc system/gegl-1.0.0 reflect Operation: Reflect pts/libraw-1.0.0 Post-Processing Benchmark pts/compress-zstd-1.2.1 -b3 Compression Level: 3 pts/pyperformance-1.0.2 pathlib Benchmark: pathlib pts/luajit-1.1.0 TEST_COMPOSITE Test: Composite pts/pyperformance-1.0.2 pickle_pure_python Benchmark: pickle_pure_python pts/xonotic-1.5.1 +vid_width 3840 +vid_height 2160 +exec effects-low.cfg Resolution: 3840 x 2160 - Effects Quality: Low system/rsvg-1.0.0 Operation: SVG Files To PNG pts/coremark-1.0.0 CoreMark Size 666 - Iterations Per Second pts/etlegacy-1.0.0 +set cl_renderer opengl2 +set r_customwidth 3840 +set r_customheight 2160 Renderer: Renderer2 - Resolution: 3840 x 2160 system/tesseract-ocr-1.0.1 Time To OCR 7 Images pts/mlpack-1.0.2 SCIKIT_SVM Benchmark: scikit_svm pts/x265-1.3.0 Bosphorus_3840x2160.y4m Video Input: Bosphorus 4K pts/polyhedron-1.0.1 linpk Benchmark: linpk pts/pyperformance-1.0.2 json_loads Benchmark: json_loads pts/pybench-1.1.3 Total For Average Test Times pts/pyperformance-1.0.2 nbody Benchmark: nbody pts/pyperformance-1.0.2 chaos Benchmark: chaos pts/polyhedron-1.0.1 ac Benchmark: ac pts/pyperformance-1.0.2 float Benchmark: float pts/pyperformance-1.0.2 crypto_pyaes Benchmark: crypto_pyaes pts/openssl-1.11.0 RSA 4096-bit Performance pts/tnn-1.0.0 -dt NAIVE -mp ../benchmark/benchmark-model/mobilenet_v2.tnnproto Target: CPU - Model: MobileNet v2 pts/build-apache-1.6.1 Time To Compile pts/ospray-1.0.3 xfrog-forest.xml --renderer scivis Demo: XFrog Forest - Renderer: SciVis pts/rnnoise-1.0.2 pts/build-ffmpeg-1.0.1 Time To Compile pts/onednn-1.5.0 --matmul --batch=inputs/matmul/shapes_transformer --cfg=f32 --engine=cpu Harness: Matrix Multiply Batch Shapes Transformer - Data Type: f32 - Engine: CPU pts/onednn-1.5.0 --ip --batch=inputs/ip/ip_1d --cfg=f32 --engine=cpu Harness: IP Batch 1D - Data Type: f32 - Engine: CPU pts/tnn-1.0.0 -dt NAIVE -mp ../benchmark/benchmark-model/squeezenet_v1.1.tnnproto Target: CPU - Model: SqueezeNet v1.1 system/gmic-1.0.0 repeat 1000 mode3d 5 moded3d 5 double3d 0 isosurface3d "'x^2+y^2+abs(z)^abs(4*cos(x*y*z*3))'",3 done output output.png Test: Plotting Isosurface Of A 3D Volume, 1000 Times pts/webp-1.0.0 -q 100 -lossless Encode Settings: Quality 100, Lossless pts/polyhedron-1.0.1 air Benchmark: air pts/ospray-1.0.3 magnetic-512-volume/magnetic-512-volume.osp --renderer scivis Demo: Magnetic Reconnection - Renderer: SciVis pts/realsr-ncnn-1.0.0 -s 4 Scale: 4x - TAA: No pts/encode-flac-1.6.0 WAV To FLAC pts/ospray-1.0.3 NASA-B-field-sun.osx --renderer scivis Demo: NASA Streamlines - Renderer: SciVis pts/x265-1.3.0 Bosphorus_1920x1080_120fps_420_8bit_YUV.y4m Video Input: Bosphorus 1080p pts/build-mplayer-1.4.0 Time To Compile pts/onednn-1.5.0 --conv --batch=inputs/conv/shapes_auto --cfg=f32 --engine=cpu Harness: Convolution Batch Shapes Auto - Data Type: f32 - Engine: CPU system/gegl-1.0.0 crop x=100 y=100 width=1920 height=1080 Operation: Crop pts/paraview-1.0.3 waveletcontour.py -d 256 -f 30 -v 1920,1080 Test: Wavelet Contour - Resolution: 1920 x 1080 pts/waifu2x-ncnn-1.0.0 -s 2 -n 3 -x Scale: 2x - Denoise: 3 - TAA: Yes pts/webp-1.0.0 -q 100 -m 6 Encode Settings: Quality 100, Highest Compression pts/encode-mp3-1.7.4 WAV To MP3 pts/oidn-1.2.0 -hdr memorial.pfm Scene: Memorial system/gegl-1.0.0 scale-size x=400 y=400 Operation: Scale pts/leveldb-1.0.0 --benchmarks=fillsync --num=1000000 Benchmark: Fill Sync pts/paraview-1.0.3 /waveletvolume.py -d 256 -f 30 -v 3840,2160 Test: Wavelet Volume - Resolution: 3840 x 2160 pts/paraview-1.0.3 waveletcontour.py -d 256 -f 30 -v 3840,2160 Test: Wavelet Contour - Resolution: 3840 x 2160 pts/lammps-1.2.1 in.rhodo Model: Rhodopsin Protein pts/rays1bench-1.0.0 Large Scene pts/ospray-1.0.3 magnetic-512-volume/magnetic-512-volume.osp --renderer pathtracer Demo: Magnetic Reconnection - Renderer: Path Tracer pts/ffte-1.2.1 N=256, 3D Complex FFT Routine pts/svt-hevc-1.1.0 1080p 8-bit YUV To HEVC Video Encode pts/onednn-1.5.0 --deconv --batch=inputs/deconv/deconv_3d --cfg=f32 --engine=cpu Harness: Deconvolution Batch deconv_3d - Data Type: f32 - Engine: CPU pts/svt-vp9-1.2.2 -tune 0 -i Bosphorus_1920x1080_120fps_420_8bit_YUV.yuv -w 1920 -h 1080 Tuning: Visual Quality Optimized - Input: Bosphorus 1080p pts/optcarrot-1.0.1 Optimized Benchmark pts/svt-vp9-1.2.2 -tune 1 -i Bosphorus_1920x1080_120fps_420_8bit_YUV.yuv -w 1920 -h 1080 Tuning: PSNR/SSIM Optimized - Input: Bosphorus 1080p pts/webp-1.0.0 -q 100 Encode Settings: Quality 100 pts/paraview-1.0.3 /waveletvolume.py -d 256 -f 30 -v 1920,1080 Test: Wavelet Volume - Resolution: 1920 x 1080 pts/webp-1.0.0 Encode Settings: Default pts/ior-1.0.0 Read Test pts/ior-1.0.0 Write Test pts/luajit-1.1.0 TEST_SOR Test: Jacobi Successive Over-Relaxation pts/luajit-1.1.0 TEST_DENSE Test: Dense LU Matrix Factorization pts/luajit-1.1.0 TEST_SPARSE Test: Sparse Matrix Multiply pts/luajit-1.1.0 TEST_FFT Test: Fast Fourier Transform pts/luajit-1.1.0 TEST_MONTE Test: Monte Carlo