new ai AMD Ryzen 9 7950X 16-Core testing with a ASUS ROG CROSSHAIR X670E HERO (1101 BIOS) and AMD Radeon RX 7900 XTX 24GB on Ubuntu 22.04 via the Phoronix Test Suite. a: Processor: AMD Ryzen 9 7950X 16-Core @ 4.50GHz (16 Cores / 32 Threads), Motherboard: ASUS ROG CROSSHAIR X670E HERO (1101 BIOS), Chipset: AMD Device 14d8, Memory: 32GB, Disk: 2048GB SOLIDIGM SSDPFKKW020X7 + 2000GB, Graphics: AMD Radeon RX 7900 XTX 24GB (2304/1249MHz), Audio: AMD Device ab30, Monitor: ASUS MG28U, Network: Intel I225-V + Intel Wi-Fi 6 AX210/AX211/AX411 OS: Ubuntu 22.04, Kernel: 6.3.0-060300rc7daily20230417-generic (x86_64), Desktop: GNOME Shell 42.5, Display Server: X Server 1.21.1.3 + Wayland, OpenGL: 4.6 Mesa 23.2.0-devel (git-f6fb189 2023-04-18 jammy-oibaf-ppa) (LLVM 15.0.7 DRM 3.52), Vulkan: 1.3.246, Compiler: GCC 11.3.0, File-System: ext4, Screen Resolution: 3840x2160 b: Processor: AMD Ryzen 9 7950X 16-Core @ 4.50GHz (16 Cores / 32 Threads), Motherboard: ASUS ROG CROSSHAIR X670E HERO (1101 BIOS), Chipset: AMD Device 14d8, Memory: 32GB, Disk: 2048GB SOLIDIGM SSDPFKKW020X7 + 2000GB, Graphics: AMD Radeon RX 7900 XTX 24GB (2304/1249MHz), Audio: AMD Device ab30, Monitor: ASUS MG28U, Network: Intel I225-V + Intel Wi-Fi 6 AX210/AX211/AX411 OS: Ubuntu 22.04, Kernel: 6.3.0-060300rc7daily20230417-generic (x86_64), Desktop: GNOME Shell 42.5, Display Server: X Server 1.21.1.3 + Wayland, OpenGL: 4.6 Mesa 23.2.0-devel (git-f6fb189 2023-04-18 jammy-oibaf-ppa) (LLVM 15.0.7 DRM 3.52), Vulkan: 1.3.246, Compiler: GCC 11.3.0, File-System: ext4, Screen Resolution: 3840x2160 c: Processor: AMD Ryzen 9 7950X 16-Core @ 4.50GHz (16 Cores / 32 Threads), Motherboard: ASUS ROG CROSSHAIR X670E HERO (1101 BIOS), Chipset: AMD Device 14d8, Memory: 32GB, Disk: 2048GB SOLIDIGM SSDPFKKW020X7 + 2000GB, Graphics: AMD Radeon RX 7900 XTX 24GB (2304/1249MHz), Audio: AMD Device ab30, Monitor: ASUS MG28U, Network: Intel I225-V + Intel Wi-Fi 6 AX210/AX211/AX411 OS: Ubuntu 22.04, Kernel: 6.3.0-060300rc7daily20230417-generic (x86_64), Desktop: GNOME Shell 42.5, Display Server: X Server 1.21.1.3 + Wayland, OpenGL: 4.6 Mesa 23.2.0-devel (git-f6fb189 2023-04-18 jammy-oibaf-ppa) (LLVM 15.0.7 DRM 3.52), Vulkan: 1.3.246, Compiler: GCC 11.3.0, File-System: ext4, Screen Resolution: 3840x2160 SVT-AV1 1.5 Encoder Mode: Preset 4 - Input: Bosphorus 4K Frames Per Second > Higher Is Better a . 6.111 |=================================================================== b . 6.165 |==================================================================== c . 6.111 |=================================================================== SVT-AV1 1.5 Encoder Mode: Preset 8 - Input: Bosphorus 4K Frames Per Second > Higher Is Better a . 75.89 |==================================================================== b . 75.81 |==================================================================== c . 75.61 |==================================================================== SVT-AV1 1.5 Encoder Mode: Preset 12 - Input: Bosphorus 4K Frames Per Second > Higher Is Better a . 210.91 |================================================================== b . 213.40 |=================================================================== c . 210.66 |================================================================== SVT-AV1 1.5 Encoder Mode: Preset 13 - Input: Bosphorus 4K Frames Per Second > Higher Is Better a . 206.52 |=================================================================== b . 207.37 |=================================================================== c . 206.94 |=================================================================== SVT-AV1 1.5 Encoder Mode: Preset 4 - Input: Bosphorus 1080p Frames Per Second > Higher Is Better a . 14.17 |==================================================================== b . 14.09 |=================================================================== c . 14.23 |==================================================================== SVT-AV1 1.5 Encoder Mode: Preset 8 - Input: Bosphorus 1080p Frames Per Second > Higher Is Better a . 120.00 |=================================================================== b . 120.31 |=================================================================== c . 118.00 |================================================================== SVT-AV1 1.5 Encoder Mode: Preset 12 - Input: Bosphorus 1080p Frames Per Second > Higher Is Better a . 716.46 |=================================================================== b . 711.29 |=================================================================== c . 710.85 |================================================================== SVT-AV1 1.5 Encoder Mode: Preset 13 - Input: Bosphorus 1080p Frames Per Second > Higher Is Better a . 698.57 |=================================================================== b . 699.65 |=================================================================== c . 698.01 |=================================================================== Faiss 1.7.4 Test: demo_sift1M Seconds < Lower Is Better a . 64.01 |==================================================================== b . 63.90 |==================================================================== c . 64.06 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - PQ baseline ms per query < Lower Is Better a . 2.624 |==================================================================== b . 2.613 |==================================================================== c . 2.627 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 64 ms per query < Lower Is Better a . 4.127 |==================================================================== b . 4.091 |=================================================================== c . 4.117 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 62 ms per query < Lower Is Better a . 3.466 |==================================================================== b . 3.449 |==================================================================== c . 3.468 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 58 ms per query < Lower Is Better a . 2.132 |==================================================================== b . 2.126 |==================================================================== c . 2.134 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 54 ms per query < Lower Is Better a . 1.284 |==================================================================== b . 1.278 |==================================================================== c . 1.285 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 50 ms per query < Lower Is Better a . 0.830 |==================================================================== b . 0.825 |==================================================================== c . 0.831 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 46 ms per query < Lower Is Better a . 0.612 |==================================================================== b . 0.610 |==================================================================== c . 0.612 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 42 ms per query < Lower Is Better a . 0.525 |==================================================================== b . 0.522 |==================================================================== c . 0.525 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 38 ms per query < Lower Is Better a . 0.495 |==================================================================== b . 0.494 |==================================================================== c . 0.496 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 34 ms per query < Lower Is Better a . 0.485 |==================================================================== b . 0.485 |==================================================================== c . 0.486 |==================================================================== Faiss 1.7.4 Test: bench_polysemous_sift1m - Polysemous 30 ms per query < Lower Is Better a . 0.482 |==================================================================== b . 0.481 |==================================================================== c . 0.483 |==================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 1 images/sec > Higher Is Better a . 139.05 |=================================================================== b . 139.52 |=================================================================== c . 135.89 |================================================================= Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 1 ms < Lower Is Better a . 7.192 |================================================================== b . 7.168 |================================================================== c . 7.360 |==================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 1 images/sec > Higher Is Better a . 547.78 |=================================================================== b . 503.50 |============================================================== c . 523.17 |================================================================ Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 1 ms < Lower Is Better a . 1.826 |=============================================================== b . 1.986 |==================================================================== c . 1.924 |================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 16 images/sec > Higher Is Better a . 264.85 |================================================================== b . 267.53 |=================================================================== c . 267.66 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 32 images/sec > Higher Is Better a . 252.89 |=================================================================== b . 253.60 |=================================================================== c . 252.68 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 64 images/sec > Higher Is Better a . 247.81 |=================================================================== b . 247.96 |=================================================================== c . 248.70 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 96 images/sec > Higher Is Better a . 245.03 |=================================================================== b . 246.19 |=================================================================== c . 245.61 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 16 images/sec > Higher Is Better a . 800.10 |=================================================================== b . 798.92 |=================================================================== c . 793.97 |================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 32 images/sec > Higher Is Better a . 824.03 |=================================================================== b . 825.34 |=================================================================== c . 830.17 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 64 images/sec > Higher Is Better a . 830.21 |================================================================== b . 837.08 |=================================================================== c . 833.73 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 96 images/sec > Higher Is Better a . 803.35 |=================================================================== b . 803.69 |=================================================================== c . 805.52 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 256 images/sec > Higher Is Better a . 247.46 |=================================================================== b . 247.09 |=================================================================== c . 247.45 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_fp32_pretrained_model - Batch Size: 512 images/sec > Higher Is Better a . 249.62 |=================================================================== b . 249.92 |=================================================================== c . 249.76 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 256 images/sec > Higher Is Better a . 779.33 |=================================================================== b . 779.24 |=================================================================== c . 780.45 |=================================================================== Intel TensorFlow 2.12 Model: resnet50_int8_pretrained_model - Batch Size: 512 images/sec > Higher Is Better a . 774.18 |=================================================================== b . 775.26 |=================================================================== c . 775.56 |=================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 1 images/sec > Higher Is Better a . 61.65 |==================================================================== b . 61.00 |=================================================================== c . 57.42 |=============================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 1 ms < Lower Is Better a . 17.07 |=================================================================== b . 17.30 |==================================================================== c . 17.20 |==================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 1 images/sec > Higher Is Better a . 104.96 |=================================================================== b . 105.50 |=================================================================== c . 99.00 |=============================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 1 ms < Lower Is Better a . 10.56 |==================================================================== b . 10.35 |=================================================================== c . 10.45 |=================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 1 images/sec > Higher Is Better a . 1591.73 |================================================================= b . 1573.54 |================================================================= c . 1606.41 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 1 images/sec > Higher Is Better a . 4525.90 |================================================================== b . 4539.90 |================================================================== c . 4528.32 |================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 16 images/sec > Higher Is Better a . 83.63 |==================================================================== b . 83.25 |==================================================================== c . 82.65 |=================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 32 images/sec > Higher Is Better a . 80.42 |==================================================================== b . 80.56 |==================================================================== c . 80.28 |==================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 64 images/sec > Higher Is Better a . 79.15 |==================================================================== b . 79.50 |==================================================================== c . 79.00 |==================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 96 images/sec > Higher Is Better a . 79.05 |==================================================================== b . 79.44 |==================================================================== c . 78.91 |==================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 16 images/sec > Higher Is Better a . 311.78 |================================================================= b . 319.25 |=================================================================== c . 312.28 |================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 32 images/sec > Higher Is Better a . 312.25 |================================================================== b . 315.79 |=================================================================== c . 311.68 |================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 64 images/sec > Higher Is Better a . 309.86 |================================================================== b . 309.26 |================================================================== c . 314.14 |=================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 96 images/sec > Higher Is Better a . 300.91 |=================================================================== b . 303.13 |=================================================================== c . 297.39 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 16 images/sec > Higher Is Better a . 1107.68 |================================================================== b . 1108.83 |================================================================== c . 1107.45 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 32 images/sec > Higher Is Better a . 1047.32 |================================================================== b . 1047.98 |================================================================== c . 1048.20 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 64 images/sec > Higher Is Better a . 952.20 |=================================================================== b . 951.88 |=================================================================== c . 951.85 |=================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 96 images/sec > Higher Is Better a . 914.91 |=================================================================== b . 915.83 |=================================================================== c . 915.29 |=================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 16 images/sec > Higher Is Better a . 4671.25 |================================================================== b . 4689.94 |================================================================== c . 4667.75 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 32 images/sec > Higher Is Better a . 5094.81 |================================================================== b . 5078.87 |================================================================== c . 5092.76 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 64 images/sec > Higher Is Better a . 4327.78 |================================================================== b . 4330.68 |================================================================== c . 4334.43 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 96 images/sec > Higher Is Better a . 4293.10 |================================================================== b . 4285.76 |================================================================== c . 4278.58 |================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 256 images/sec > Higher Is Better a . 80.36 |==================================================================== b . 80.40 |==================================================================== c . 80.20 |==================================================================== Intel TensorFlow 2.12 Model: inceptionv4_fp32_pretrained_model - Batch Size: 512 images/sec > Higher Is Better a . 81.04 |==================================================================== b . 81.07 |==================================================================== c . 80.95 |==================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 256 images/sec > Higher Is Better a . 288.45 |=================================================================== b . 289.43 |=================================================================== c . 287.48 |=================================================================== Intel TensorFlow 2.12 Model: inceptionv4_int8_pretrained_model - Batch Size: 512 images/sec > Higher Is Better a . 290.66 |=================================================================== b . 288.25 |================================================================== c . 288.82 |=================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 256 images/sec > Higher Is Better a . 867.02 |=================================================================== b . 867.13 |=================================================================== c . 866.92 |=================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_fp32_pretrained_model - Batch Size: 512 images/sec > Higher Is Better a . 856.50 |=================================================================== c . 855.61 |=================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 256 images/sec > Higher Is Better a . 3482.01 |================================================================== c . 3482.91 |================================================================== Intel TensorFlow 2.12 Model: mobilenetv1_int8_pretrained_model - Batch Size: 512 images/sec > Higher Is Better a . 3263.94 |================================================================== c . 3272.66 |================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 1 ms < Lower Is Better a . 8.070 |==================================================================== c . 8.122 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 1 ms < Lower Is Better a . 23.16 |==================================================================== c . 23.13 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 16 images/sec > Higher Is Better a . 241.34 |=================================================================== c . 240.72 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 32 images/sec > Higher Is Better a . 228.35 |=================================================================== c . 226.86 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 64 images/sec > Higher Is Better a . 225.81 |=================================================================== c . 226.03 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 96 images/sec > Higher Is Better a . 228.11 |================================================================== c . 230.14 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 16 images/sec > Higher Is Better a . 62.78 |==================================================================== c . 62.79 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 32 images/sec > Higher Is Better a . 62.82 |==================================================================== c . 62.91 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 64 images/sec > Higher Is Better a . 64.63 |==================================================================== c . 64.61 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 96 images/sec > Higher Is Better a . 65.86 |==================================================================== c . 65.59 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 256 images/sec > Higher Is Better a . 238.11 |=================================================================== c . 239.23 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_resnetv1_50_imagenet_224_224_6.97G_1.1_Z4.0 - Batch Size: 512 images/sec > Higher Is Better a . 242.65 |=================================================================== c . 243.29 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 256 images/sec > Higher Is Better a . 68.24 |==================================================================== c . 68.42 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_inceptionv4_imagenet_299_299_24.55G_1.1_Z4.0 - Batch Size: 512 images/sec > Higher Is Better a . 70.36 |==================================================================== c . 70.06 |==================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 1 ms < Lower Is Better a . 1.586 |==================================================================== c . 1.573 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 16 images/sec > Higher Is Better a . 953.78 |=================================================================== c . 953.58 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 32 images/sec > Higher Is Better a . 770.81 |=================================================================== c . 768.31 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 64 images/sec > Higher Is Better a . 715.78 |=================================================================== c . 717.29 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 96 images/sec > Higher Is Better a . 691.01 |=================================================================== c . 690.86 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 256 images/sec > Higher Is Better a . 650.17 |=================================================================== c . 650.04 |=================================================================== AMD ZenDNN TensorFlow 2.10 ZenDNN 4.0 Model: tf_mobilenetv1_1.0_imagenet_224_224_1.14G_1.1_Z4.0 - Batch Size: 512 images/sec > Higher Is Better a . 643.44 |=================================================================== c . 642.22 |===================================================================