[RFC] Improve quantized convolution performance for armv8 architectures#5754

Merged

FrozenGene merged 15 commits intoapache:masterfrom giuseros:conv2d_quantized_improvements

Jun 23, 2020

+1,065-14

Commits on Jun 19, 2020

Improve quantized conv2d performance for armv8
Giuseppe Rossini
committed
Add ASF header to conv2d_gemm.py
Giuseppe Rossini
committed
Run clang-format-10 on c++ files
Giuseppe Rossini
committed
Fix pylint errors/warnings
Giuseppe Rossini
committed
Fix pylint errors/warnings in topi
Giuseppe Rossini
committed
Fix legalizations tests for aarch64
Giuseppe Rossini
committed
Reintroduce conv2d_nhwc_spatial_pack.arm_cpu and int16 cast
Giuseppe Rossini
committed
Switch type of legalization depending on the strategy used
Giuseppe Rossini
committed
Revert last commit
Giuseppe Rossini
committed
Fix the auto-tuner by registering the correct schedules
Giuseppe Rossini
committed
Address review comments
Giuseppe Rossini
committed
Improve usability and readability of conv2d_gemm_weight_transform
Giuseppe Rossini
committed
Change variable name to weight in Conv2DGemmWeightTransformRel
Giuseppe Rossini
committed
Fix clang-10 linting errors
Giuseppe Rossini
committed

Commits on Jun 22, 2020

Trigger tests
Giuseppe Rossini
committed