vulkan tests AMD Ryzen Threadripper 3990X 64-Core testing with a Gigabyte TRX40 AORUS PRO WIFI (F6 BIOS) and AMD Radeon RX 5700 8GB on Ubuntu 23.04 via the Phoronix Test Suite. a: Processor: AMD Ryzen Threadripper 3990X 64-Core @ 2.90GHz (64 Cores / 128 Threads), Motherboard: Gigabyte TRX40 AORUS PRO WIFI (F6 BIOS), Chipset: AMD Starship/Matisse, Memory: 128GB, Disk: Samsung SSD 970 EVO Plus 500GB, Graphics: AMD Radeon RX 5700 8GB (1750/875MHz), Audio: AMD Navi 10 HDMI Audio, Monitor: DELL P2415Q, Network: Intel I211 + Intel Wi-Fi 6 AX200 OS: Ubuntu 23.04, Kernel: 6.2.0-26-generic (x86_64), Desktop: GNOME Shell 44.0, Display Server: X Server + Wayland, OpenGL: 4.6 Mesa 23.0.2 (LLVM 15.0.7 DRM 3.49), Compiler: GCC 12.2.0, File-System: ext4, Screen Resolution: 3840x2160 b: Processor: AMD Ryzen Threadripper 3990X 64-Core @ 2.90GHz (64 Cores / 128 Threads), Motherboard: Gigabyte TRX40 AORUS PRO WIFI (F6 BIOS), Chipset: AMD Starship/Matisse, Memory: 128GB, Disk: Samsung SSD 970 EVO Plus 500GB, Graphics: AMD Radeon RX 5700 8GB (1750/875MHz), Audio: AMD Navi 10 HDMI Audio, Monitor: DELL P2415Q, Network: Intel I211 + Intel Wi-Fi 6 AX200 OS: Ubuntu 23.04, Kernel: 6.2.0-26-generic (x86_64), Desktop: GNOME Shell 44.0, Display Server: X Server + Wayland, OpenGL: 4.6 Mesa 23.0.2 (LLVM 15.0.7 DRM 3.49), Compiler: GCC 12.2.0, File-System: ext4, Screen Resolution: 3840x2160 c: Processor: AMD Ryzen Threadripper 3990X 64-Core @ 2.90GHz (64 Cores / 128 Threads), Motherboard: Gigabyte TRX40 AORUS PRO WIFI (F6 BIOS), Chipset: AMD Starship/Matisse, Memory: 128GB, Disk: Samsung SSD 970 EVO Plus 500GB, Graphics: AMD Radeon RX 5700 8GB (1750/875MHz), Audio: AMD Navi 10 HDMI Audio, Monitor: DELL P2415Q, Network: Intel I211 + Intel Wi-Fi 6 AX200 OS: Ubuntu 23.04, Kernel: 6.2.0-26-generic (x86_64), Desktop: GNOME Shell 44.0, Display Server: X Server + Wayland, OpenGL: 4.6 Mesa 23.0.2 (LLVM 15.0.7 DRM 3.49), Compiler: GCC 12.2.0, File-System: ext4, Screen Resolution: 3840x2160 vkpeak 20230730 fp32-scalar GFLOPS > Higher Is Better a . 7836.61 |================================================================== b . 7856.19 |================================================================== c . 7866.83 |================================================================== vkpeak 20230730 fp32-vec4 GFLOPS > Higher Is Better a . 7781.34 |================================================================== b . 7808.56 |================================================================== c . 7808.48 |================================================================== vkpeak 20230730 fp16-scalar GFLOPS > Higher Is Better a . 7853.66 |================================================================== b . 7860.29 |================================================================== c . 7866.33 |================================================================== vkpeak 20230730 fp16-vec4 GFLOPS > Higher Is Better a . 12516.94 |================================================================= b . 12550.91 |================================================================= c . 12559.31 |================================================================= vkpeak 20230730 fp64-scalar GFLOPS > Higher Is Better a . 493.36 |=================================================================== b . 493.01 |=================================================================== c . 493.51 |=================================================================== vkpeak 20230730 fp64-vec4 GFLOPS > Higher Is Better a . 493.28 |=================================================================== b . 492.96 |=================================================================== c . 493.57 |=================================================================== vkpeak 20230730 int32-scalar GIOPS > Higher Is Better a . 1214.41 |================================================================== b . 1213.48 |================================================================== c . 1214.40 |================================================================== vkpeak 20230730 int32-vec4 GIOPS > Higher Is Better a . 1574.82 |================================================================== b . 1574.60 |================================================================== c . 1575.91 |================================================================== vkpeak 20230730 int16-scalar GIOPS > Higher Is Better a . 7856.21 |================================================================== b . 7852.03 |================================================================== c . 7857.01 |================================================================== vkpeak 20230730 int16-vec4 GIOPS > Higher Is Better a . 12531.35 |================================================================= b . 12550.38 |================================================================= c . 12563.14 |================================================================= VkFFT 1.2.31 Test: FFT + iFFT R2C / C2R Benchmark Score > Higher Is Better a . 25525 |==================================================================== b . 25447 |==================================================================== c . 25451 |==================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C 1D batched in half precision Benchmark Score > Higher Is Better a . 100903 |=================================================================== b . 101063 |=================================================================== c . 100839 |=================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C Bluestein in single precision Benchmark Score > Higher Is Better a . 7268 |===================================================================== b . 7258 |===================================================================== c . 7259 |===================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C 1D batched in double precision Benchmark Score > Higher Is Better a . 18431 |==================================================================== b . 18428 |==================================================================== c . 18426 |==================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C 1D batched in single precision Benchmark Score > Higher Is Better a . 58944 |==================================================================== b . 58947 |==================================================================== c . 58938 |==================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C multidimensional in single precision Benchmark Score > Higher Is Better a . 16993 |==================================================================== b . 16976 |==================================================================== c . 16967 |==================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C Bluestein benchmark in double precision Benchmark Score > Higher Is Better a . 3066 |===================================================================== b . 3064 |===================================================================== c . 3064 |===================================================================== VkFFT 1.2.31 Test: FFT + iFFT C2C 1D batched in single precision, no reshuffling Benchmark Score > Higher Is Better a . 64140 |==================================================================== b . 64139 |==================================================================== c . 64131 |==================================================================== VkResample 1.0 Upscale: 2x - Precision: Single ms < Lower Is Better a . 12.77 |==================================================================== b . 12.77 |==================================================================== c . 12.78 |==================================================================== NCNN 20230517 Target: CPU - Model: mobilenet ms < Lower Is Better a . 25.21 |================================================================== b . 25.78 |==================================================================== c . 25.96 |==================================================================== NCNN 20230517 Target: CPU-v2-v2 - Model: mobilenet-v2 ms < Lower Is Better a . 13.83 |=========================================================== b . 14.49 |============================================================== c . 16.01 |==================================================================== NCNN 20230517 Target: CPU-v3-v3 - Model: mobilenet-v3 ms < Lower Is Better a . 12.73 |================================================================ b . 13.45 |=================================================================== c . 13.60 |==================================================================== NCNN 20230517 Target: CPU - Model: shufflenet-v2 ms < Lower Is Better a . 14.08 |=================================================================== b . 14.38 |==================================================================== c . 14.12 |=================================================================== NCNN 20230517 Target: CPU - Model: mnasnet ms < Lower Is Better a . 13.13 |========================================================== b . 12.90 |========================================================= c . 15.27 |==================================================================== NCNN 20230517 Target: CPU - Model: efficientnet-b0 ms < Lower Is Better a . 18.00 |============================================================== b . 19.69 |==================================================================== c . 19.06 |================================================================== NCNN 20230517 Target: CPU - Model: blazeface ms < Lower Is Better a . 5.89 |========================================================= b . 7.12 |===================================================================== c . 6.95 |=================================================================== NCNN 20230517 Target: CPU - Model: googlenet ms < Lower Is Better a . 30.87 |================================================================== b . 31.55 |==================================================================== c . 31.64 |==================================================================== NCNN 20230517 Target: CPU - Model: vgg16 ms < Lower Is Better a . 45.64 |================================================================= b . 47.53 |==================================================================== c . 46.88 |=================================================================== NCNN 20230517 Target: CPU - Model: resnet18 ms < Lower Is Better a . 19.07 |=================================================================== b . 19.40 |==================================================================== c . 18.91 |================================================================== NCNN 20230517 Target: CPU - Model: alexnet ms < Lower Is Better a . 11.76 |========================================================= b . 13.97 |==================================================================== c . 12.37 |============================================================ NCNN 20230517 Target: CPU - Model: resnet50 ms < Lower Is Better a . 35.20 |============================================================= b . 39.03 |==================================================================== c . 38.18 |=================================================================== NCNN 20230517 Target: CPU - Model: yolov4-tiny ms < Lower Is Better a . 37.72 |==================================================================== b . 37.38 |=================================================================== c . 37.14 |=================================================================== NCNN 20230517 Target: CPU - Model: squeezenet_ssd ms < Lower Is Better a . 22.30 |================================================================== b . 22.27 |================================================================== c . 22.89 |==================================================================== NCNN 20230517 Target: CPU - Model: regnety_400m ms < Lower Is Better a . 45.43 |================================================================ b . 48.03 |==================================================================== c . 47.02 |=================================================================== NCNN 20230517 Target: CPU - Model: vision_transformer ms < Lower Is Better a . 73.47 |================================================================== b . 74.11 |================================================================== c . 75.96 |==================================================================== NCNN 20230517 Target: CPU - Model: FastestDet ms < Lower Is Better a . 14.66 |============================================================== b . 13.70 |========================================================== c . 16.03 |==================================================================== NCNN 20230517 Target: Vulkan GPU - Model: mobilenet ms < Lower Is Better a . 24.66 |================================================================== b . 25.25 |==================================================================== c . 25.37 |==================================================================== NCNN 20230517 Target: Vulkan GPU-v2-v2 - Model: mobilenet-v2 ms < Lower Is Better a . 15.71 |==================================================================== b . 14.78 |================================================================ c . 13.63 |=========================================================== NCNN 20230517 Target: Vulkan GPU-v3-v3 - Model: mobilenet-v3 ms < Lower Is Better a . 12.97 |=================================================================== b . 13.25 |==================================================================== c . 12.57 |================================================================= NCNN 20230517 Target: Vulkan GPU - Model: shufflenet-v2 ms < Lower Is Better a . 14.75 |==================================================================== b . 14.27 |================================================================== c . 13.29 |============================================================= NCNN 20230517 Target: Vulkan GPU - Model: mnasnet ms < Lower Is Better a . 14.18 |==================================================================== b . 13.72 |================================================================== c . 12.93 |============================================================== NCNN 20230517 Target: Vulkan GPU - Model: efficientnet-b0 ms < Lower Is Better a . 18.42 |================================================================ b . 19.47 |==================================================================== c . 18.02 |=============================================================== NCNN 20230517 Target: Vulkan GPU - Model: blazeface ms < Lower Is Better a . 7.22 |===================================================================== b . 7.09 |==================================================================== c . 6.08 |========================================================== NCNN 20230517 Target: Vulkan GPU - Model: googlenet ms < Lower Is Better a . 31.43 |==================================================================== b . 31.24 |==================================================================== c . 31.46 |==================================================================== NCNN 20230517 Target: Vulkan GPU - Model: vgg16 ms < Lower Is Better a . 47.52 |=================================================================== b . 48.13 |==================================================================== c . 47.50 |=================================================================== NCNN 20230517 Target: Vulkan GPU - Model: resnet18 ms < Lower Is Better a . 19.20 |=================================================================== b . 19.42 |==================================================================== c . 19.04 |=================================================================== NCNN 20230517 Target: Vulkan GPU - Model: alexnet ms < Lower Is Better a . 15.22 |=================================================================== b . 15.34 |==================================================================== c . 13.07 |========================================================== NCNN 20230517 Target: Vulkan GPU - Model: resnet50 ms < Lower Is Better a . 38.91 |================================================================== b . 39.84 |==================================================================== c . 38.43 |================================================================== NCNN 20230517 Target: Vulkan GPU - Model: yolov4-tiny ms < Lower Is Better a . 36.72 |================================================================== b . 37.79 |==================================================================== c . 37.36 |=================================================================== NCNN 20230517 Target: Vulkan GPU - Model: squeezenet_ssd ms < Lower Is Better a . 21.81 |=================================================================== b . 22.01 |==================================================================== c . 21.78 |=================================================================== NCNN 20230517 Target: Vulkan GPU - Model: regnety_400m ms < Lower Is Better a . 46.73 |==================================================================== b . 46.22 |=================================================================== c . 46.10 |=================================================================== NCNN 20230517 Target: Vulkan GPU - Model: vision_transformer ms < Lower Is Better a . 75.17 |==================================================================== b . 72.26 |================================================================= c . 72.70 |================================================================== NCNN 20230517 Target: Vulkan GPU - Model: FastestDet ms < Lower Is Better a . 14.25 |=================================================================== b . 14.53 |==================================================================== c . 13.88 |=================================================================