ncnn llama

AMD Ryzen Threadripper 7980X 64-Cores testing with a System76 Thelio Major (FA Z5 BIOS) and AMD Radeon RX 6700 XT 12GB on Ubuntu 24.04 via the Phoronix Test Suite.

HTML result view exported from: https://openbenchmarking.org/result/2412293-PTS-NCNNLLAM81.

ncnn llamaProcessorMotherboardChipsetMemoryDiskGraphicsAudioMonitorNetworkOSKernelDesktopDisplay ServerOpenGLCompilerFile-SystemScreen ResolutionabcdAMD Ryzen Threadripper 7980X 64-Cores @ 5.37GHz (64 Cores / 128 Threads)System76 Thelio Major (FA Z5 BIOS)AMD Device 14a44 x 32GB DDR5-4800MT/s Micron MTC20F1045S1RC48BA21000GB CT1000T700SSD5AMD Radeon RX 6700 XT 12GBAMD Device 14ccDELL P2415QAquantia AQC113C NBase-T/IEEE + Realtek RTL8125 2.5GbE + Intel Wi-Fi 6EUbuntu 24.046.12.3-061203-generic (x86_64)GNOME Shell 46.0X Server + Wayland4.6 Mesa 24.0.9-0ubuntu0.2 (LLVM 17.0.6 DRM 3.59)GCC 13.3.0ext41920x1200OpenBenchmarking.orgKernel Details- Transparent Huge Pages: madviseCompiler Details- --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-backtrace --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-fG75Ri/gcc-13-13.3.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-13-fG75Ri/gcc-13-13.3.0/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 Processor Details- Scaling Governor: amd-pstate-epp powersave (Boost: Enabled EPP: balance_performance) - CPU Microcode: 0xa108105 Security Details- gather_data_sampling: Not affected + itlb_multihit: Not affected + l1tf: Not affected + mds: Not affected + meltdown: Not affected + mmio_stale_data: Not affected + reg_file_data_sampling: 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; BHI: Not affected + srbds: Not affected + tsx_async_abort: Not affected

ncnn llamancnn: CPU - mobilenetncnn: CPU-v2-v2 - mobilenet-v2ncnn: CPU-v3-v3 - mobilenet-v3ncnn: CPU - shufflenet-v2ncnn: CPU - mnasnetncnn: CPU - efficientnet-b0ncnn: CPU - blazefacencnn: CPU - googlenetncnn: CPU - vgg16ncnn: CPU - resnet18ncnn: CPU - alexnetncnn: CPU - resnet50ncnn: CPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3ncnn: CPU - yolov4-tinyncnn: CPU - squeezenet_ssdncnn: CPU - regnety_400mncnn: CPU - vision_transformerncnn: CPU - FastestDetncnn: Vulkan GPU - mobilenetncnn: Vulkan GPU-v2-v2 - mobilenet-v2ncnn: Vulkan GPU-v3-v3 - mobilenet-v3ncnn: Vulkan GPU - shufflenet-v2ncnn: Vulkan GPU - mnasnetncnn: Vulkan GPU - efficientnet-b0ncnn: Vulkan GPU - blazefacencnn: Vulkan GPU - googlenetncnn: Vulkan GPU - vgg16ncnn: Vulkan GPU - resnet18ncnn: Vulkan GPU - alexnetncnn: Vulkan GPU - resnet50ncnn: Vulkan GPUv2-yolov3v2-yolov3 - mobilenetv2-yolov3ncnn: Vulkan GPU - yolov4-tinyncnn: Vulkan GPU - squeezenet_ssdncnn: Vulkan GPU - regnety_400mncnn: Vulkan GPU - vision_transformerncnn: Vulkan GPU - FastestDetllama-cpp: CPU BLAS - Llama-3.1-Tulu-3-8B-Q8_0 - Text Generation 128llama-cpp: CPU BLAS - Llama-3.1-Tulu-3-8B-Q8_0 - Prompt Processing 512llama-cpp: CPU BLAS - Llama-3.1-Tulu-3-8B-Q8_0 - Prompt Processing 1024llama-cpp: CPU BLAS - Llama-3.1-Tulu-3-8B-Q8_0 - Prompt Processing 2048llama-cpp: CPU BLAS - Mistral-7B-Instruct-v0.3-Q8_0 - Text Generation 128llama-cpp: CPU BLAS - Mistral-7B-Instruct-v0.3-Q8_0 - Prompt Processing 512llama-cpp: CPU BLAS - Mistral-7B-Instruct-v0.3-Q8_0 - Prompt Processing 1024llama-cpp: CPU BLAS - Mistral-7B-Instruct-v0.3-Q8_0 - Prompt Processing 2048llama-cpp: CPU BLAS - granite-3.0-3b-a800m-instruct-Q8_0 - Text Generation 128llama-cpp: CPU BLAS - granite-3.0-3b-a800m-instruct-Q8_0 - Prompt Processing 512llama-cpp: CPU BLAS - granite-3.0-3b-a800m-instruct-Q8_0 - Prompt Processing 1024llama-cpp: CPU BLAS - granite-3.0-3b-a800m-instruct-Q8_0 - Prompt Processing 2048abcd16.277.948.5710.627.8511.364.3215.3931.598.514.7115.2616.2725.0518.3133.0743.4912.7215.537.658.4710.647.8411.124.2515.1330.778.414.3514.6715.5324.2017.6232.4642.3712.7415.7868.7070.1569.7716.6070.7669.7969.3975.57268.35275.79250.9016.047.88.3210.577.811.084.3115.0130.248.44.3414.8616.0424.5917.5433.2440.7612.9216.017.848.0810.317.2110.714.1614.7930.248.494.315.0516.0123.7316.8432.6441.6112.5115.6570.2770.0268.5816.6668.6570.369.9975.4272.33278242.9515.887.848.5510.517.7711.274.2815.2731.398.444.3714.7815.8824.9218.0432.8243.0212.7815.97.868.5510.697.8311.274.2915.3131.588.44.3814.715.924.8918.0533.342.2111.1915.8169.0970.1569.2216.5268.6569.6269.5475.88283.43268.94250.6717.057.838.610.717.8711.324.6215.3232.298.594.4416.4517.0525.3119.1833.242.1412.2316.277.918.5310.757.8111.244.2815.3230.648.424.3415.1816.2724.517.6932.8341.1412.6315.7970.5469.1169.9616.6769.6668.7669.7475.13283.5277.51260.63OpenBenchmarking.org

NCNN

Target: CPU - Model: mobilenet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: mobilenetabcd48121620SE +/- 0.20, N = 416.2716.0415.8817.05MIN: 15.82 / MAX: 17.16MIN: 15.95 / MAX: 16.58MIN: 15.76 / MAX: 24.56MIN: 15.91 / MAX: 274.261. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU-v2-v2 - Model: mobilenet-v2

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU-v2-v2 - Model: mobilenet-v2abcd246810SE +/- 0.04, N = 47.947.807.847.83MIN: 7.74 / MAX: 13.85MIN: 7.7 / MAX: 9.02MIN: 7.72 / MAX: 8.87MIN: 7.73 / MAX: 11.331. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU-v3-v3 - Model: mobilenet-v3

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU-v3-v3 - Model: mobilenet-v3abcd246810SE +/- 0.03, N = 48.578.328.558.60MIN: 8.42 / MAX: 11.02MIN: 8.24 / MAX: 9.54MIN: 8.45 / MAX: 10.15MIN: 8.49 / MAX: 12.581. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: shufflenet-v2

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: shufflenet-v2abcd3691215SE +/- 0.05, N = 410.6210.5710.5110.71MIN: 10.47 / MAX: 11.33MIN: 10.52 / MAX: 11.08MIN: 10.46 / MAX: 11.05MIN: 10.66 / MAX: 11.251. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: mnasnet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: mnasnetabcd246810SE +/- 0.01, N = 47.857.807.777.87MIN: 7.75 / MAX: 8.45MIN: 7.71 / MAX: 8.32MIN: 7.7 / MAX: 8.35MIN: 7.79 / MAX: 8.51. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: efficientnet-b0

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: efficientnet-b0abcd3691215SE +/- 0.03, N = 411.3611.0811.2711.32MIN: 11.2 / MAX: 14.79MIN: 10.99 / MAX: 11.6MIN: 11.22 / MAX: 11.82MIN: 11.24 / MAX: 11.981. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: blazeface

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: blazefaceabcd1.03952.0793.11854.1585.1975SE +/- 0.02, N = 44.324.314.284.62MIN: 4.25 / MAX: 11.58MIN: 4.24 / MAX: 10.24MIN: 4.25 / MAX: 4.63MIN: 4.28 / MAX: 71.021. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: googlenet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: googlenetabcd48121620SE +/- 0.04, N = 415.3915.0115.2715.32MIN: 15.16 / MAX: 16.08MIN: 14.89 / MAX: 15.67MIN: 15.13 / MAX: 16.55MIN: 15.21 / MAX: 15.931. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: vgg16

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: vgg16abcd714212835SE +/- 0.25, N = 431.5930.2431.3932.29MIN: 30.46 / MAX: 154.3MIN: 29.7 / MAX: 30.87MIN: 30.66 / MAX: 39.08MIN: 30.58 / MAX: 64.171. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: resnet18

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: resnet18abcd246810SE +/- 0.09, N = 48.518.408.448.59MIN: 8.33 / MAX: 9.52MIN: 8.35 / MAX: 9.02MIN: 8.37 / MAX: 13.03MIN: 8.55 / MAX: 9.181. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: alexnet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: alexnetabcd1.05982.11963.17944.23925.299SE +/- 0.35, N = 44.714.344.374.44MIN: 4.18 / MAX: 227.25MIN: 4.18 / MAX: 4.78MIN: 4.21 / MAX: 4.92MIN: 4.27 / MAX: 5.771. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: resnet50

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: resnet50abcd48121620SE +/- 0.23, N = 415.2614.8614.7816.45MIN: 14.7 / MAX: 23.32MIN: 14.69 / MAX: 23.77MIN: 14.7 / MAX: 15.35MIN: 15.23 / MAX: 37.141. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3abcd48121620SE +/- 0.20, N = 416.2716.0415.8817.05MIN: 15.82 / MAX: 17.16MIN: 15.95 / MAX: 16.58MIN: 15.76 / MAX: 24.56MIN: 15.91 / MAX: 274.261. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: yolov4-tiny

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: yolov4-tinyabcd612182430SE +/- 0.08, N = 425.0524.5924.9225.31MIN: 24.63 / MAX: 44.44MIN: 24.35 / MAX: 25.01MIN: 24.55 / MAX: 25.65MIN: 25.06 / MAX: 26.031. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: squeezenet_ssd

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: squeezenet_ssdabcd510152025SE +/- 0.15, N = 418.3117.5418.0419.18MIN: 17.83 / MAX: 26.62MIN: 17.31 / MAX: 18.18MIN: 17.83 / MAX: 18.56MIN: 18.04 / MAX: 165.281. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: regnety_400m

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: regnety_400mabcd816243240SE +/- 0.17, N = 433.0733.2432.8233.20MIN: 32.56 / MAX: 188.49MIN: 32.54 / MAX: 166.34MIN: 32.65 / MAX: 48.88MIN: 33.01 / MAX: 35.051. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: vision_transformer

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: vision_transformerabcd1020304050SE +/- 0.56, N = 443.4940.7643.0242.14MIN: 39.99 / MAX: 695.96MIN: 40.33 / MAX: 48.85MIN: 40.17 / MAX: 328MIN: 39.94 / MAX: 176.011. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: CPU - Model: FastestDet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: CPU - Model: FastestDetabcd3691215SE +/- 0.09, N = 412.7212.9212.7812.23MIN: 12.35 / MAX: 21.71MIN: 12.82 / MAX: 13.42MIN: 12.65 / MAX: 21.72MIN: 12.15 / MAX: 12.931. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: mobilenet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: mobilenetabcd48121620SE +/- 0.06, N = 315.5316.0115.9016.27MIN: 14.75 / MAX: 166.8MIN: 14.83 / MAX: 287.48MIN: 15.82 / MAX: 16.54MIN: 15.66 / MAX: 147.921. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2abcd246810SE +/- 0.20, N = 37.657.847.867.91MIN: 7.14 / MAX: 10.2MIN: 7.73 / MAX: 9.01MIN: 7.75 / MAX: 8.98MIN: 7.81 / MAX: 9.111. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3abcd246810SE +/- 0.09, N = 38.478.088.558.53MIN: 8.06 / MAX: 43.04MIN: 7.98 / MAX: 9.07MIN: 8.45 / MAX: 9.56MIN: 8.44 / MAX: 9.691. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: shufflenet-v2

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: shufflenet-v2abcd3691215SE +/- 0.12, N = 310.6410.3110.6910.75MIN: 10.32 / MAX: 18.32MIN: 10.25 / MAX: 10.71MIN: 10.63 / MAX: 11.3MIN: 10.69 / MAX: 11.291. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: mnasnet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: mnasnetabcd246810SE +/- 0.32, N = 37.847.217.837.81MIN: 7.19 / MAX: 145.21MIN: 7.15 / MAX: 7.74MIN: 7.76 / MAX: 8.2MIN: 7.73 / MAX: 8.341. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: efficientnet-b0

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: efficientnet-b0abcd3691215SE +/- 0.17, N = 311.1210.7111.2711.24MIN: 10.7 / MAX: 11.97MIN: 10.64 / MAX: 11.31MIN: 11.21 / MAX: 11.91MIN: 11.18 / MAX: 11.811. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: blazeface

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: blazefaceabcd0.96531.93062.89593.86124.8265SE +/- 0.05, N = 34.254.164.294.28MIN: 4.11 / MAX: 4.76MIN: 4.13 / MAX: 4.7MIN: 4.26 / MAX: 4.61MIN: 4.26 / MAX: 4.921. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: googlenet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: googlenetabcd48121620SE +/- 0.19, N = 315.1314.7915.3115.32MIN: 14.67 / MAX: 15.95MIN: 14.64 / MAX: 23.47MIN: 15.17 / MAX: 15.97MIN: 15.15 / MAX: 29.151. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: vgg16

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: vgg16abcd714212835SE +/- 0.53, N = 330.7730.2431.5830.64MIN: 29.22 / MAX: 55.24MIN: 29.83 / MAX: 34.6MIN: 30.78 / MAX: 32.25MIN: 29.89 / MAX: 31.451. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: resnet18

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: resnet18abcd246810SE +/- 0.01, N = 38.418.498.408.42MIN: 8.33 / MAX: 9.68MIN: 8.44 / MAX: 8.86MIN: 8.35 / MAX: 9.01MIN: 8.37 / MAX: 8.641. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: alexnet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: alexnetabcd0.98551.9712.95653.9424.9275SE +/- 0.00, N = 34.354.304.384.34MIN: 4.16 / MAX: 4.98MIN: 4.15 / MAX: 4.78MIN: 4.19 / MAX: 12.97MIN: 4.18 / MAX: 4.751. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: resnet50

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: resnet50abcd48121620SE +/- 0.04, N = 314.6715.0514.7015.18MIN: 14.53 / MAX: 15.27MIN: 14.98 / MAX: 15.58MIN: 14.62 / MAX: 15.29MIN: 15.09 / MAX: 15.941. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPUv2-yolov3v2-yolov3 - Model: mobilenetv2-yolov3abcd48121620SE +/- 0.06, N = 315.5316.0115.9016.27MIN: 14.75 / MAX: 166.8MIN: 14.83 / MAX: 287.48MIN: 15.82 / MAX: 16.54MIN: 15.66 / MAX: 147.921. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: yolov4-tiny

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: yolov4-tinyabcd612182430SE +/- 0.42, N = 324.2023.7324.8924.50MIN: 23.13 / MAX: 25.24MIN: 23.48 / MAX: 24.3MIN: 24.57 / MAX: 25.81MIN: 24.15 / MAX: 25.191. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: squeezenet_ssd

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: squeezenet_ssdabcd48121620SE +/- 0.39, N = 317.6216.8418.0517.69MIN: 16.8 / MAX: 168.75MIN: 16.71 / MAX: 25.09MIN: 17.95 / MAX: 18.63MIN: 17.57 / MAX: 18.431. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: regnety_400m

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: regnety_400mabcd816243240SE +/- 0.38, N = 332.4632.6433.3032.83MIN: 31.48 / MAX: 39.76MIN: 31.4 / MAX: 315.52MIN: 32.59 / MAX: 177.33MIN: 32.53 / MAX: 42.141. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: vision_transformer

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: vision_transformerabcd1020304050SE +/- 0.63, N = 342.3741.6142.2141.14MIN: 40.41 / MAX: 257.66MIN: 40.26 / MAX: 214.45MIN: 39.99 / MAX: 306MIN: 40.24 / MAX: 165.91. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

NCNN

Target: Vulkan GPU - Model: FastestDet

OpenBenchmarking.orgms, Fewer Is BetterNCNN 20241226Target: Vulkan GPU - Model: FastestDetabcd3691215SE +/- 0.04, N = 312.7412.5111.1912.63MIN: 12.58 / MAX: 14.39MIN: 12.42 / MAX: 13.37MIN: 11.09 / MAX: 11.81MIN: 12.53 / MAX: 13.141. (CXX) g++ options: -O3 -rdynamic -lgomp -lpthread

Llama.cpp

Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Text Generation 128

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Text Generation 128abcd48121620SE +/- 0.00, N = 315.7815.6515.8115.791. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Prompt Processing 512

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Prompt Processing 512abcd1632486480SE +/- 0.36, N = 368.7070.2769.0970.541. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Prompt Processing 1024

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Prompt Processing 1024abcd1632486480SE +/- 0.11, N = 370.1570.0270.1569.111. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Prompt Processing 2048

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Llama-3.1-Tulu-3-8B-Q8_0 - Test: Prompt Processing 2048abcd1632486480SE +/- 0.24, N = 369.7768.5869.2269.961. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Text Generation 128

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Text Generation 128abcd48121620SE +/- 0.07, N = 316.6016.6616.5216.671. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Prompt Processing 512

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Prompt Processing 512abcd1632486480SE +/- 0.22, N = 370.7668.6568.6569.661. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Prompt Processing 1024

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Prompt Processing 1024abcd1632486480SE +/- 0.33, N = 369.7970.3069.6268.761. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Prompt Processing 2048

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: Mistral-7B-Instruct-v0.3-Q8_0 - Test: Prompt Processing 2048abcd1632486480SE +/- 0.26, N = 369.3969.9969.5469.741. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Text Generation 128

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Text Generation 128abcd20406080100SE +/- 0.21, N = 375.5775.4075.8875.131. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Prompt Processing 512

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Prompt Processing 512abcd60120180240300SE +/- 3.83, N = 3268.35272.33283.43283.501. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Prompt Processing 1024

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Prompt Processing 1024abcd60120180240300SE +/- 1.45, N = 3275.79278.00268.94277.511. (CXX) g++ options: -O3

Llama.cpp

Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Prompt Processing 2048

OpenBenchmarking.orgTokens Per Second, More Is BetterLlama.cpp b4397Backend: CPU BLAS - Model: granite-3.0-3b-a800m-instruct-Q8_0 - Test: Prompt Processing 2048abcd60120180240300SE +/- 1.47, N = 3250.90242.95250.67260.631. (CXX) g++ options: -O3


Phoronix Test Suite v10.8.5