svt deepsparse

AMD Ryzen Threadripper 7980X 64-Cores testing with a System76 Thelio Major (FA Z5 BIOS) and AMD Radeon Pro W7900 45GB on Ubuntu 23.10 via the Phoronix Test Suite.

Compare your own system(s) to this result file with the Phoronix Test Suite by running the command: phoronix-test-suite benchmark 2403159-PTS-SVTDEEPS65
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
Show Result Confidence Charts

Statistics

Show Overall Harmonic Mean(s)
Show Overall Geometric Mean
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
a
March 15
  1 Hour, 32 Minutes
b
March 15
  1 Hour, 33 Minutes
c
March 15
  1 Hour, 32 Minutes
d
March 15
  1 Hour, 32 Minutes
Invert Hiding All Results Option
  1 Hour, 32 Minutes

Only show results where is faster than
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):


svt deepsparseOpenBenchmarking.orgPhoronix Test SuiteAMD Ryzen Threadripper 7980X 64-Cores @ 7.79GHz (64 Cores / 128 Threads)System76 Thelio Major (FA Z5 BIOS)AMD Device 14a44 x 32GB DRAM-4800MT/s Micron MTC20F1045S1RC48BA21000GB CT1000T700SSD5AMD Radeon Pro W7900 45GB (1760/1124MHz)AMD Device 14ccDELL P2415QAquantia AQC113C NBase-T/IEEE + Realtek RTL8125 2.5GbE + Intel Wi-Fi 6 AX210/AX211/AX411Ubuntu 23.106.5.0-25-generic (x86_64)GNOME Shell 45.2X Server + Wayland4.6 Mesa 23.2.1-1ubuntu3.1 (LLVM 15.0.7 DRM 3.54)GCC 13.2.0ext41920x1080ProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerOpenGLCompilerFile-SystemScreen ResolutionSvt Deepsparse BenchmarksSystem Logs- Transparent Huge Pages: madvise- --build=x86_64-linux-gnu --disable-vtable-verify --disable-werror --enable-bootstrap --enable-cet --enable-checking=release --enable-clocale=gnu --enable-default-pie --enable-gnu-unique-object --enable-languages=c,ada,c++,go,d,fortran,objc,obj-c++,m2 --enable-libphobos-checking=release --enable-libstdcxx-debug --enable-libstdcxx-time=yes --enable-link-serialization=2 --enable-multiarch --enable-multilib --enable-nls --enable-objc-gc=auto --enable-offload-defaulted --enable-offload-targets=nvptx-none=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-XYspKM/gcc-13-13.2.0/debian/tmp-gcn/usr --enable-plugin --enable-shared --enable-threads=posix --host=x86_64-linux-gnu --program-prefix=x86_64-linux-gnu- --target=x86_64-linux-gnu --with-abi=m64 --with-arch-32=i686 --with-build-config=bootstrap-lto-lean --with-default-libstdcxx-abi=new --with-gcc-major-version-only --with-multilib-list=m32,m64,mx32 --with-target-system-zlib=auto --with-tune=generic --without-cuda-driver -v - Scaling Governor: amd-pstate-epp powersave (EPP: balance_performance) - CPU Microcode: 0xa108105- Python 3.11.6- gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + retbleed: Not affected + spec_rstack_overflow: Mitigation of Safe RET + spec_store_bypass: Mitigation of SSB disabled via prctl + spectre_v1: Mitigation of usercopy/swapgs barriers and __user pointer sanitization + spectre_v2: Mitigation of Enhanced / Automatic IBRS IBPB: conditional STIBP: always-on RSB filling PBRSB-eIBRS: Not affected + srbds: Not affected + tsx_async_abort: Not affected

abcdResult OverviewPhoronix Test Suite100%101%102%103%104%SVT-AV1SVT-AV1SVT-AV1SVT-AV1SVT-AV1Neural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseSVT-AV1SVT-AV1Neural Magic DeepSparseNeural Magic DeepSparseSVT-AV1Neural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparseNeural Magic DeepSparsePreset 12 - Bosphorus 4KPreset 8 - Bosphorus 1080pPreset 13 - Bosphorus 4KPreset 13 - Bosphorus 1080pPreset 12 - Bosphorus 1080pR.5.S.I - A.M.SR.5.S.I - A.M.SN.T.C.D.m - S.S.SN.T.C.D.m - S.S.SPreset 4 - Bosphorus 1080pPreset 4 - Bosphorus 4KL.C.7.Q - S.S.SL.C.7.Q - S.S.SPreset 8 - Bosphorus 4KB.L.N.Q.A.S.I - S.S.SB.L.N.Q.A.S.I - S.S.SN.T.C.B.b.u.S.S.I - S.S.SN.T.C.B.b.u.S.S.I - S.S.SR.5.B - S.S.SR.5.B - S.S.SC.D.Y.C.S.I - A.M.SC.D.Y.C.S.I - A.M.SN.D.C.o.b.u.o.I - A.M.SL.C.7.Q - A.M.SL.C.7.Q - A.M.SN.T.C.B.b.u.S.S.I - A.M.SN.T.C.B.b.u.S.S.I - A.M.SC.D.Y.C.S.I - S.S.SC.D.Y.C.S.I - S.S.SC.S.9.P.Y.P - A.M.SN.D.C.o.b.u.o.I - A.M.SB.L.N.Q.A.S.I - A.M.SB.L.N.Q.A.S.I - A.M.SR.5.S.I - S.S.SR.5.S.I - S.S.SN.T.C.B.b.u.c - A.M.SC.C.R.5.I - S.S.SC.S.9.P.Y.P - A.M.SC.C.R.5.I - S.S.SN.T.C.B.b.u.c - A.M.SN.T.C.B.b.u.c - S.S.SN.T.C.B.b.u.c - S.S.SN.T.C.D.m - A.M.SN.T.C.D.m - A.M.SC.S.9.P.Y.P - S.S.SC.S.9.P.Y.P - S.S.SC.C.R.5.I - A.M.SN.D.C.o.b.u.o.I - S.S.SN.D.C.o.b.u.o.I - S.S.SC.C.R.5.I - A.M.SR.5.B - A.M.SR.5.B - A.M.S

svt deepsparsesvt-av1: Preset 4 - Bosphorus 4Ksvt-av1: Preset 8 - Bosphorus 4Ksvt-av1: Preset 12 - Bosphorus 4Ksvt-av1: Preset 13 - Bosphorus 4Ksvt-av1: Preset 4 - Bosphorus 1080psvt-av1: Preset 8 - Bosphorus 1080psvt-av1: Preset 12 - Bosphorus 1080psvt-av1: Preset 13 - Bosphorus 1080pdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Streamdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Synchronous Single-Streamdeepsparse: ResNet-50, Baseline - Asynchronous Multi-Streamdeepsparse: ResNet-50, Baseline - Synchronous Single-Streamdeepsparse: ResNet-50, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: ResNet-50, Sparse INT8 - Synchronous Single-Streamdeepsparse: Llama2 Chat 7b Quantized - Asynchronous Multi-Streamdeepsparse: Llama2 Chat 7b Quantized - Synchronous Single-Streamdeepsparse: CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Streamdeepsparse: CV Classification, ResNet-50 ImageNet - Synchronous Single-Streamdeepsparse: CV Detection, YOLOv5s COCO, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: CV Detection, YOLOv5s COCO, Sparse INT8 - Synchronous Single-Streamdeepsparse: NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Streamdeepsparse: NLP Text Classification, DistilBERT mnli - Synchronous Single-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Streamdeepsparse: BERT-Large, NLP Question Answering, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: BERT-Large, NLP Question Answering, Sparse INT8 - Synchronous Single-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Streamdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Streamdeepsparse: NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Synchronous Single-Streamdeepsparse: ResNet-50, Baseline - Asynchronous Multi-Streamdeepsparse: ResNet-50, Baseline - Synchronous Single-Streamdeepsparse: ResNet-50, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: ResNet-50, Sparse INT8 - Synchronous Single-Streamdeepsparse: Llama2 Chat 7b Quantized - Asynchronous Multi-Streamdeepsparse: Llama2 Chat 7b Quantized - Synchronous Single-Streamdeepsparse: CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Streamdeepsparse: CV Classification, ResNet-50 ImageNet - Synchronous Single-Streamdeepsparse: CV Detection, YOLOv5s COCO, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: CV Detection, YOLOv5s COCO, Sparse INT8 - Synchronous Single-Streamdeepsparse: NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Streamdeepsparse: NLP Text Classification, DistilBERT mnli - Synchronous Single-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Streamdeepsparse: CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Streamdeepsparse: BERT-Large, NLP Question Answering, Sparse INT8 - Asynchronous Multi-Streamdeepsparse: BERT-Large, NLP Question Answering, Sparse INT8 - Synchronous Single-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Streamdeepsparse: NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Streamabcd9.89294.509188.558192.56727.476193.759616.583662.49321.847417.2332887.4442252.3577274.2576170.30902253.86021134.77043.439111.1249274.5853170.9594129.440887.3635188.3255113.470139.572929.7744396.576889.709121.922417.2184365.335558.01918.99703.959429.14535.86483.53320.87882252.462789.870429.11085.842361.769811.440642.45098.8070202.015333.568120.151611.1350363.533658.06899.79393.734195.907193.28627.282194.769615.540653.68821.822517.2332886.1847251.6102274.4692171.49342269.50681131.63733.440611.0159274.3872171.3099128.956087.5496188.1505114.407239.468529.8179396.444690.498321.902717.1960365.983258.01889.00943.970729.12155.82433.50870.88102251.612590.752129.13245.830261.997811.415942.48538.7347202.505633.519420.158511.0383364.428358.14429.88993.930194.065192.63827.518189.608625.817671.83221.860617.2329884.6567251.2477274.4696171.47812254.22281133.49563.429811.0814274.2789171.2808129.101287.6317188.2128113.876139.466429.7891397.587089.805321.904117.1825365.577158.02079.02523.976629.12305.82453.53240.87962258.140790.225329.14355.831261.937111.405542.47568.7752202.367733.552520.100711.1223363.762358.19169.86694.559195.858187.62527.241195.612621.442670.13821.795017.2130886.7758253.0264274.4604171.35612236.91691132.31513.428411.0145274.5981171.3647129.455287.6213188.4415114.825239.452529.7882397.565290.387521.952717.1981366.682358.08799.00403.949029.12325.82853.56040.88062259.264890.766729.10835.828161.757911.406742.42178.7026202.463033.553220.101411.0513363.585758.1383OpenBenchmarking.org

SVT-AV1

This is a benchmark of the SVT-AV1 open-source video encoder/decoder. SVT-AV1 was originally developed by Intel as part of their Open Visual Cloud / Scalable Video Technology (SVT). Development of SVT-AV1 has since moved to the Alliance for Open Media as part of upstream AV1 development. SVT-AV1 is a CPU-based multi-threaded video encoder for the AV1 video format with a sample YUV video file. Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 4 - Input: Bosphorus 4Kbdca3691215SE +/- 0.014, N = 3SE +/- 0.034, N = 3SE +/- 0.028, N = 3SE +/- 0.009, N = 39.7939.8669.8899.8921. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 8 - Input: Bosphorus 4Kbcad20406080100SE +/- 0.31, N = 3SE +/- 0.20, N = 3SE +/- 0.33, N = 3SE +/- 0.60, N = 393.7393.9394.5194.561. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 12 - Input: Bosphorus 4Kacdb4080120160200SE +/- 2.09, N = 5SE +/- 1.67, N = 3SE +/- 2.35, N = 3SE +/- 2.54, N = 3188.56194.07195.86195.911. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 13 - Input: Bosphorus 4Kdacb4080120160200SE +/- 1.36, N = 3SE +/- 2.20, N = 3SE +/- 2.22, N = 3SE +/- 2.02, N = 15187.63192.57192.64193.291. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 4 - Input: Bosphorus 1080pdbac612182430SE +/- 0.12, N = 3SE +/- 0.10, N = 3SE +/- 0.22, N = 3SE +/- 0.16, N = 327.2427.2827.4827.521. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 8 - Input: Bosphorus 1080pcabd4080120160200SE +/- 0.78, N = 3SE +/- 2.31, N = 3SE +/- 1.86, N = 3SE +/- 1.92, N = 3189.61193.76194.77195.611. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 12 - Input: Bosphorus 1080pbadc140280420560700SE +/- 5.36, N = 3SE +/- 8.04, N = 3SE +/- 4.99, N = 3SE +/- 4.75, N = 3615.54616.58621.44625.821. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

OpenBenchmarking.orgFrames Per Second, More Is BetterSVT-AV1 2.0Encoder Mode: Preset 13 - Input: Bosphorus 1080pbadc140280420560700SE +/- 1.76, N = 3SE +/- 3.32, N = 3SE +/- 6.00, N = 15SE +/- 5.18, N = 15653.69662.49670.14671.831. (CXX) g++ options: -march=native -mno-avx -mavx2 -mavx512f -mavx512bw -mavx512dq

Neural Magic DeepSparse

This is a benchmark of Neural Magic's DeepSparse using its built-in deepsparse.benchmark utility and various models from their SparseZoo (https://sparsezoo.neuralmagic.com/). Learn more via the OpenBenchmarking.org test page.

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Streamdbac510152025SE +/- 0.03, N = 3SE +/- 0.02, N = 3SE +/- 0.08, N = 3SE +/- 0.04, N = 321.8021.8221.8521.86

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-Streamdcab48121620SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 317.2117.2317.2317.23

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Streamcbda2004006008001000SE +/- 0.15, N = 3SE +/- 0.69, N = 3SE +/- 0.73, N = 3SE +/- 0.68, N = 3884.66886.18886.78887.44

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Synchronous Single-Streamcbad60120180240300SE +/- 0.20, N = 3SE +/- 0.19, N = 3SE +/- 1.31, N = 3SE +/- 1.88, N = 3251.25251.61252.36253.03

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Streamadbc60120180240300SE +/- 0.04, N = 3SE +/- 0.20, N = 3SE +/- 0.15, N = 3SE +/- 0.11, N = 3274.26274.46274.47274.47

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Baseline - Scenario: Synchronous Single-Streamadcb4080120160200SE +/- 0.32, N = 3SE +/- 0.19, N = 3SE +/- 0.16, N = 3SE +/- 0.01, N = 3170.31171.36171.48171.49

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Streamdacb5001000150020002500SE +/- 15.99, N = 3SE +/- 6.75, N = 3SE +/- 9.51, N = 3SE +/- 8.66, N = 32236.922253.862254.222269.51

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Sparse INT8 - Scenario: Synchronous Single-Streambdca2004006008001000SE +/- 7.00, N = 3SE +/- 5.41, N = 3SE +/- 3.06, N = 3SE +/- 2.80, N = 31131.641132.321133.501134.77

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: Llama2 Chat 7b Quantized - Scenario: Asynchronous Multi-Streamdcab0.77411.54822.32233.09643.8705SE +/- 0.0032, N = 3SE +/- 0.0031, N = 3SE +/- 0.0053, N = 3SE +/- 0.0006, N = 33.42843.42983.43913.4406

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: Llama2 Chat 7b Quantized - Scenario: Synchronous Single-Streamdbca3691215SE +/- 0.00, N = 3SE +/- 0.00, N = 3SE +/- 0.06, N = 3SE +/- 0.07, N = 311.0111.0211.0811.12

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Streamcbad60120180240300SE +/- 0.12, N = 3SE +/- 0.04, N = 3SE +/- 0.26, N = 3SE +/- 0.36, N = 3274.28274.39274.59274.60

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-Streamacbd4080120160200SE +/- 0.03, N = 3SE +/- 0.12, N = 3SE +/- 0.40, N = 3SE +/- 0.12, N = 3170.96171.28171.31171.36

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Asynchronous Multi-Streambcad306090120150SE +/- 0.15, N = 3SE +/- 0.42, N = 3SE +/- 0.18, N = 3SE +/- 0.18, N = 3128.96129.10129.44129.46

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Synchronous Single-Streamabdc20406080100SE +/- 0.16, N = 3SE +/- 0.07, N = 3SE +/- 0.08, N = 3SE +/- 0.06, N = 387.3687.5587.6287.63

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Streambcad4080120160200SE +/- 0.02, N = 3SE +/- 0.05, N = 3SE +/- 0.07, N = 3SE +/- 0.00, N = 3188.15188.21188.33188.44

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-Streamacbd306090120150SE +/- 0.37, N = 3SE +/- 0.17, N = 3SE +/- 0.52, N = 3SE +/- 0.34, N = 3113.47113.88114.41114.83

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Streamdcba918273645SE +/- 0.04, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 3SE +/- 0.02, N = 339.4539.4739.4739.57

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Synchronous Single-Streamadcb714212835SE +/- 0.02, N = 3SE +/- 0.02, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 329.7729.7929.7929.82

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Streambadc90180270360450SE +/- 0.87, N = 3SE +/- 0.23, N = 3SE +/- 0.27, N = 3SE +/- 0.37, N = 3396.44396.58397.57397.59

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Synchronous Single-Streamacdb20406080100SE +/- 0.23, N = 3SE +/- 0.14, N = 3SE +/- 0.05, N = 3SE +/- 0.29, N = 389.7189.8190.3990.50

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Streambcad510152025SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.04, N = 321.9021.9021.9221.95

OpenBenchmarking.orgitems/sec, More Is BetterNeural Magic DeepSparse 1.7Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-Streamcbda48121620SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 317.1817.2017.2017.22

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Asynchronous Multi-Streamdbca80160240320400SE +/- 0.31, N = 3SE +/- 0.38, N = 3SE +/- 0.63, N = 3SE +/- 0.99, N = 3366.68365.98365.58365.34

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Document Classification, oBERT base uncased on IMDB - Scenario: Synchronous Single-Streamdcab1326395265SE +/- 0.07, N = 3SE +/- 0.03, N = 3SE +/- 0.03, N = 3SE +/- 0.05, N = 358.0958.0258.0258.02

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Asynchronous Multi-Streamcbda3691215SE +/- 0.0015, N = 3SE +/- 0.0071, N = 3SE +/- 0.0073, N = 3SE +/- 0.0068, N = 39.02529.00949.00408.9970

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Scenario: Synchronous Single-Streamcbad0.89471.78942.68413.57884.4735SE +/- 0.0030, N = 3SE +/- 0.0031, N = 3SE +/- 0.0204, N = 3SE +/- 0.0293, N = 33.97663.97073.95943.9490

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Baseline - Scenario: Asynchronous Multi-Streamadcb714212835SE +/- 0.00, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 329.1529.1229.1229.12

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Baseline - Scenario: Synchronous Single-Streamadcb1.31962.63923.95885.27846.598SE +/- 0.0111, N = 3SE +/- 0.0066, N = 3SE +/- 0.0053, N = 3SE +/- 0.0004, N = 35.86485.82855.82455.8243

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Sparse INT8 - Scenario: Asynchronous Multi-Streamdacb0.80111.60222.40333.20444.0055SE +/- 0.0258, N = 3SE +/- 0.0104, N = 3SE +/- 0.0151, N = 3SE +/- 0.0132, N = 33.56043.53323.53243.5087

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: ResNet-50, Sparse INT8 - Scenario: Synchronous Single-Streambdca0.19820.39640.59460.79280.991SE +/- 0.0054, N = 3SE +/- 0.0042, N = 3SE +/- 0.0025, N = 3SE +/- 0.0021, N = 30.88100.88060.87960.8788

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: Llama2 Chat 7b Quantized - Scenario: Asynchronous Multi-Streamdcab5001000150020002500SE +/- 1.96, N = 3SE +/- 1.96, N = 3SE +/- 3.18, N = 3SE +/- 0.34, N = 32259.262258.142252.462251.61

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: Llama2 Chat 7b Quantized - Scenario: Synchronous Single-Streamdbca20406080100SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.49, N = 3SE +/- 0.58, N = 390.7790.7590.2389.87

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: CV Classification, ResNet-50 ImageNet - Scenario: Asynchronous Multi-Streamcbad714212835SE +/- 0.01, N = 3SE +/- 0.00, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 329.1429.1329.1129.11

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: CV Classification, ResNet-50 ImageNet - Scenario: Synchronous Single-Streamacbd1.31452.6293.94355.2586.5725SE +/- 0.0009, N = 3SE +/- 0.0039, N = 3SE +/- 0.0134, N = 3SE +/- 0.0043, N = 35.84235.83125.83025.8281

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Asynchronous Multi-Streambcad1428425670SE +/- 0.07, N = 3SE +/- 0.21, N = 3SE +/- 0.09, N = 3SE +/- 0.07, N = 362.0061.9461.7761.76

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: CV Detection, YOLOv5s COCO, Sparse INT8 - Scenario: Synchronous Single-Streamabdc3691215SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 311.4411.4211.4111.41

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, DistilBERT mnli - Scenario: Asynchronous Multi-Streambcad1020304050SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 3SE +/- 0.00, N = 342.4942.4842.4542.42

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Text Classification, DistilBERT mnli - Scenario: Synchronous Single-Streamacbd246810SE +/- 0.0288, N = 3SE +/- 0.0131, N = 3SE +/- 0.0397, N = 3SE +/- 0.0260, N = 38.80708.77528.73478.7026

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Asynchronous Multi-Streambdca4080120160200SE +/- 0.18, N = 3SE +/- 0.10, N = 3SE +/- 0.08, N = 3SE +/- 0.12, N = 3202.51202.46202.37202.02

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: CV Segmentation, 90% Pruned YOLACT Pruned - Scenario: Synchronous Single-Streamadcb816243240SE +/- 0.02, N = 3SE +/- 0.03, N = 3SE +/- 0.00, N = 3SE +/- 0.01, N = 333.5733.5533.5533.52

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Asynchronous Multi-Streambadc510152025SE +/- 0.04, N = 3SE +/- 0.01, N = 3SE +/- 0.01, N = 3SE +/- 0.02, N = 320.1620.1520.1020.10

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: BERT-Large, NLP Question Answering, Sparse INT8 - Scenario: Synchronous Single-Streamacdb3691215SE +/- 0.03, N = 3SE +/- 0.02, N = 3SE +/- 0.01, N = 3SE +/- 0.04, N = 311.1411.1211.0511.04

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Asynchronous Multi-Streambcda80160240320400SE +/- 0.43, N = 3SE +/- 0.33, N = 3SE +/- 1.03, N = 3SE +/- 0.17, N = 3364.43363.76363.59363.53

OpenBenchmarking.orgms/batch, Fewer Is BetterNeural Magic DeepSparse 1.7Model: NLP Token Classification, BERT base uncased conll2003 - Scenario: Synchronous Single-Streamcbda1326395265SE +/- 0.05, N = 3SE +/- 0.03, N = 3SE +/- 0.04, N = 3SE +/- 0.06, N = 358.1958.1458.1458.07

52 Results Shown

SVT-AV1:
  Preset 4 - Bosphorus 4K
  Preset 8 - Bosphorus 4K
  Preset 12 - Bosphorus 4K
  Preset 13 - Bosphorus 4K
  Preset 4 - Bosphorus 1080p
  Preset 8 - Bosphorus 1080p
  Preset 12 - Bosphorus 1080p
  Preset 13 - Bosphorus 1080p
Neural Magic DeepSparse:
  NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Stream
  NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Stream
  NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Asynchronous Multi-Stream
  NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Synchronous Single-Stream
  ResNet-50, Baseline - Asynchronous Multi-Stream
  ResNet-50, Baseline - Synchronous Single-Stream
  ResNet-50, Sparse INT8 - Asynchronous Multi-Stream
  ResNet-50, Sparse INT8 - Synchronous Single-Stream
  Llama2 Chat 7b Quantized - Asynchronous Multi-Stream
  Llama2 Chat 7b Quantized - Synchronous Single-Stream
  CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Stream
  CV Classification, ResNet-50 ImageNet - Synchronous Single-Stream
  CV Detection, YOLOv5s COCO, Sparse INT8 - Asynchronous Multi-Stream
  CV Detection, YOLOv5s COCO, Sparse INT8 - Synchronous Single-Stream
  NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Stream
  NLP Text Classification, DistilBERT mnli - Synchronous Single-Stream
  CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Stream
  CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Stream
  BERT-Large, NLP Question Answering, Sparse INT8 - Asynchronous Multi-Stream
  BERT-Large, NLP Question Answering, Sparse INT8 - Synchronous Single-Stream
  NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Stream
  NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Stream
  NLP Document Classification, oBERT base uncased on IMDB - Asynchronous Multi-Stream
  NLP Document Classification, oBERT base uncased on IMDB - Synchronous Single-Stream
  NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Asynchronous Multi-Stream
  NLP Text Classification, BERT base uncased SST2, Sparse INT8 - Synchronous Single-Stream
  ResNet-50, Baseline - Asynchronous Multi-Stream
  ResNet-50, Baseline - Synchronous Single-Stream
  ResNet-50, Sparse INT8 - Asynchronous Multi-Stream
  ResNet-50, Sparse INT8 - Synchronous Single-Stream
  Llama2 Chat 7b Quantized - Asynchronous Multi-Stream
  Llama2 Chat 7b Quantized - Synchronous Single-Stream
  CV Classification, ResNet-50 ImageNet - Asynchronous Multi-Stream
  CV Classification, ResNet-50 ImageNet - Synchronous Single-Stream
  CV Detection, YOLOv5s COCO, Sparse INT8 - Asynchronous Multi-Stream
  CV Detection, YOLOv5s COCO, Sparse INT8 - Synchronous Single-Stream
  NLP Text Classification, DistilBERT mnli - Asynchronous Multi-Stream
  NLP Text Classification, DistilBERT mnli - Synchronous Single-Stream
  CV Segmentation, 90% Pruned YOLACT Pruned - Asynchronous Multi-Stream
  CV Segmentation, 90% Pruned YOLACT Pruned - Synchronous Single-Stream
  BERT-Large, NLP Question Answering, Sparse INT8 - Asynchronous Multi-Stream
  BERT-Large, NLP Question Answering, Sparse INT8 - Synchronous Single-Stream
  NLP Token Classification, BERT base uncased conll2003 - Asynchronous Multi-Stream
  NLP Token Classification, BERT base uncased conll2003 - Synchronous Single-Stream