AMDMIGraphX
AMDMIGraphX copied to clipboard
ROCmSoftwarePlatform
AMD's graph optimization engine.
This is a follow up on https://github.com/ROCm/AMDMIGraphX/pull/2909 addresses: https://github.com/ROCm/AMDMIGraphX/issues/2960
``` @25 = gpu::code_object[code_object=5232,symbol_name=mlir_reshape_transpose_reshape_reshape_transpose_slice_reshape_transpose_dot,global=46080,local=256,](@23,@19,@24) -> half_type, {20, 77, 77}, {5929, 77, 1}, target_id=0 @26 = load[offset=591360,end=828520](@1) -> half_type, {1, 20, 77, 77}, {118580, 5929, 77, 1}, target_id=0 @27 = reshape_lazy[dims={1,...
Check repo for instances were we reference older versions of ROCm. Move them up where appropriate Places python versions would also need to be changed... torch==2.1.1 -> torch==2.1.2 ``` ~/code/AMDMIGraphX...
- Implement ONNX parsing support for the [Scan](https://onnx.ai/onnx/operators/onnx__Scan.html) operator - Resolves https://github.com/migraphx-benchmark/AMDMIGraphX/issues/116
https://github.com/ROCmSoftwarePlatform/AMDMIGraphX/pull/2506/files This PR had to disable FP8 tests for the CPU backend. Ref implementation is doing Float -- > Fp8 -- > Float conversion but CPU backend is doing entire...
If the E4M3FN model is quantized using QDQ pairs then, it can be converted into E4M3FNUZ types by multiplying scales with some constants. This would allow using E4M3FN models inside...
Currently, we only eliminate concat if the axis = 0 or the axis > 0 but the data is packed (i.e., all dims up to axis = 1). This PR...
[GridSample](https://github.com/onnx/onnx/blob/main/docs/Operators.md#GridSample) cubic interpolation is not yet supported. Extend the related parsing with a new parser to handle it. A possible solution from TVM: https://github.com/apache/tvm/blob/main/python/tvm/topi/image/grid_sample.py#L208-L277