Charlie Lin

Seems to be failing `test_spacetodepth_example_cpu` from ONNX backend. The test says CPU but the compiled program looks to be using the GPU...

No updates right now. In our long backlog right now.

Still seeing this dangling quantizelinear after FP8 OCP->FNUZ changes on MI300 but now it's merged with the elementwise kernels from the OCP->FNUZ conversion: ``` @26 = gpu::code_object[code_object=6592,symbol_name=quantizelinear_bit_cast_equal_where_equal_equal_logical_or_where│@207 = gpu::gemm[alpha=1,beta=0,compute_fp32=1,trans_batch=0,solution_idx=0](@204,@206,@205) ->...

With how the current performance report for fp8 and int8 on MI300 look this is a marginal effect current compared to the time taken on fp8/int8 GEMMs. Would be better...

Here's a picture of the situation in ONNX before and the issue:  See this internal discussion for more elaboration: https://github.com/ROCm/AMDMIGraphX-internal/discussions/81

Resolved by https://github.com/ROCm/AMDMIGraphX/pull/3920.

> CI Hit a failure, not sure why it is not showing up.... > > [2024-07-30T21:21:39.312Z] [ RUN ] test_bitwise_andmigraphx::shape::bool_type [2024-07-30T21:21:39.312Z] [2024-07-30T21:21:39.312Z] module: "main" [2024-07-30T21:21:39.312Z] y = @param:y -> bool_type,...

Where is this spec for this operator?

What's the status on this PR? Is there additional work needed or is the removal of the allowed types the only thing that needs review?

I fixed the timing logic. Need doc update for the new environment variable.