Adding SME1 based kernel for FP32 GEMM support

[binneswa]

Added SME1 FP32 fp32_mopa_2VLx2VL GEMM kernel support

[binneswa]

Any update on this Pull Request ?

[morgolock]

Hi @binneswa

Thanks for your contribution.

Could you please tell us more about the use case for this kernel? which model will use it, details of the workload (types, shapes,ect)

I'll discuss your PR with the team.

[binneswa]

Hi @morgolock , Thanks
This kernel is being used in GeekBenchAi workloads, Device: sme1 devices and Os is Android
models like Mobilenet uses it,

[binneswa]

Any update on this Pull Request ?

[morgolock]

Hi @binneswa

We're looking into this. The problem is that these kernels are generated with an internal tool, we are considering uploading a new PR with the SME1 kernels for you to test and confirm that it solves the problem.

Hope this helps

[morgolock]

Hi @binneswa

We have just merged PR 1216 implementing the sme1 kernels.

I'm closing your PR in favor of PR 1216.

Thanks for your contribution. If you experience any problems, please raise an issue https://github.com/ARM-software/ComputeLibrary/issues

Hope this helps