BgemmKernel for ARM32

[honglh]

No description provided.

[honglh]

Hi,

Based from BgemmKernel 64 and Ruy I have implemented the 32 bit BgemmKernel. I have tested on RPI2 32bit, RPI3 32-bit and they have comparable performance against RPI3 64-bit, though I have not run full benchmark.

Please check out the PR 432 for the changes I made, mostly in the new file bgemm_kernels_arm32.h This is a specific implementation for kernel layout 4x4

Most of my development and testing are done with v0.3.1 When I try to use master I still run into issue #408 thus I cannot test on master even though my PR Is based from master

Overall I think this would be helpful for users of ARM devices where 32 bit is the only option e.g. RPI2

[honglh]

I can see a couple CI/CD builds failed on github. Not sure how to run the tests offline. Please let me know if you have more details of the error or some instructions I can test the build offline in my local setup. Thanks!

[lgeiger]

Thank you very much for contributing and sorry for the missing instructions on how to properly setup a dev environment we'll try to add a proper guides soon. For now I hope the following will get you started:

Most of my development and testing are done with v0.3.1 When I try to use master I still run into issue #408 thus I cannot test on master even though my PR Is based from master

Please checkout and let me know if that works for you #408 (comment)

[lgeiger]

Closed via #432