Skip to content

Instantly share code, notes, and snippets.

@antiagainst

antiagainst/1v2.md

Last active November 19, 2021 15:47
Show Gist options
  • Save antiagainst/58875b3ea7243170bb30e8a71056d80e to your computer and use it in GitHub Desktop.
Save antiagainst/58875b3ea7243170bb30e8a71056d80e to your computer and use it in GitHub Desktop.
Download ZIP
Raw
1v2.md

Full Benchmark Summary

Similar Benchmarks

Benchmark Name Average Latency (ms) Median Latency (ms) Latency Standard Deviation (ms)
MobileNetV2 [fp32,imagenet] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 12 (vs. 13, 7.69%↓) 12 0
DeepLabV3 [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 16 (vs. 15, 6.67%↑) 16 0
DeepLabV3 [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 37 (vs. 35, 5.71%↑) 37 2
PoseNet [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 37 (vs. 39, 5.13%↓) 37 3
MobileNetV2 [fp32,imagenet] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 33 (vs. 32, 3.12%↑) 33 0
MobileBertSquad [fp32] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 239 (vs. 237, 0.84%↑) 239 3
MobileBertSquad [fp16] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 138 (vs. 137, 0.73%↑) 138 1
MobileBertSquad [fp16] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 239 (vs. 238, 0.42%↑) 239 1
PoseNet [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 38 (vs. 38, 0.00%) 38 1
PoseNet [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 20 (vs. 20, 0.00%) 20 1
MobileSSD [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 95 (vs. 95, 0.00%) 96 3
MobileSSD [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 94 (vs. 94, 0.00%) 94 3
MobileSSD [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 52 (vs. 52, 0.00%) 52 1
DeepLabV3 [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 35 (vs. 35, 0.00%) 35 1
MobileNetV3Small [fp32,imagenet] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 25 (vs. 25, 0.00%) 25 0
MobileNetV3Small [fp32,imagenet] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 9 (vs. 9, 0.00%) 9 0
Raw
1v3.md

Full Benchmark Summary

Similar Benchmarks

Benchmark Name Average Latency (ms) Median Latency (ms) Latency Standard Deviation (ms)
DeepLabV3 [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 16 (vs. 15, 6.67%↑) 16 0
DeepLabV3 [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 36 (vs. 35, 2.86%↑) 36 1
PoseNet [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 39 (vs. 38, 2.63%↑) 39 1
PoseNet [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 38 (vs. 39, 2.56%↓) 38 1
MobileBertSquad [fp16] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 234 (vs. 238, 1.68%↓) 236 5
MobileBertSquad [fp16] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 139 (vs. 137, 1.46%↑) 139 2
MobileSSD [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 93 (vs. 94, 1.06%↓) 93 2
MobileSSD [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 94 (vs. 95, 1.05%↓) 94 3
PoseNet [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 20 (vs. 20, 0.00%) 21 0
MobileSSD [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 52 (vs. 52, 0.00%) 52 1
DeepLabV3 [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 35 (vs. 35, 0.00%) 35 1
MobileBertSquad [fp32] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 237 (vs. 237, 0.00%) 237 3
MobileNetV3Small [fp32,imagenet] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 25 (vs. 25, 0.00%) 25 0
MobileNetV3Small [fp32,imagenet] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 9 (vs. 9, 0.00%) 9 0
MobileNetV2 [fp32,imagenet] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 32 (vs. 32, 0.00%) 32 1
MobileNetV2 [fp32,imagenet] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 13 (vs. 13, 0.00%) 13 0
Raw
1v4.md

Full Benchmark Summary

Similar Benchmarks

Benchmark Name Average Latency (ms) Median Latency (ms) Latency Standard Deviation (ms)
MobileNetV2 [fp32,imagenet] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 12 (vs. 13, 7.69%↓) 12 0
PoseNet [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 21 (vs. 20, 5.00%↑) 21 1
MobileBertSquad [fp32] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 228 (vs. 237, 3.80%↓) 228 2
MobileBertSquad [fp16] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 231 (vs. 238, 2.94%↓) 232 1
DeepLabV3 [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 36 (vs. 35, 2.86%↑) 36 1
PoseNet [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 39 (vs. 38, 2.63%↑) 39 1
PoseNet [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 38 (vs. 39, 2.56%↓) 38 1
MobileSSD [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 53 (vs. 52, 1.92%↑) 52 1
MobileSSD [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 95 (vs. 94, 1.06%↑) 96 2
MobileSSD [fp32] (TFLite) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 95 (vs. 95, 0.00%) 96 4
DeepLabV3 [fp32] (TFLite) full-inference,default-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 35 (vs. 35, 0.00%) 35 1
DeepLabV3 [fp32] (TFLite) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 15 (vs. 15, 0.00%) 15 0
MobileNetV3Small [fp32,imagenet] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 25 (vs. 25, 0.00%) 25 0
MobileNetV3Small [fp32,imagenet] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 9 (vs. 9, 0.00%) 9 0
MobileBertSquad [fp16] (TensorFlow) kernel-execution,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 137 (vs. 137, 0.00%) 137 1
MobileNetV2 [fp32,imagenet] (TensorFlow) full-inference,experimental-flags with IREE-Vulkan @ Pixel-6 (GPU-Mali-G78) 32 (vs. 32, 0.00%) 32 0
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment