AMDMIGraphX
AMDMIGraphX copied to clipboard
ROCmSoftwarePlatform
AMD's graph optimization engine.
The issue is related to the GridSample operator ([PR](https://github.com/ROCm/AMDMIGraphX/pull/2909)). When the interpolation mode is linear the following opreators are fused: `equal`, `logical_and`, `where`, `mul`, `add` in a way that the...
* Creates a ref onehot operator that handles dynamic shapes * New simplify matcher that replaces the new ref onehot operator with other ops and scatter * Refactored to use...
Add support to save and load pre compiled MIGraphX graph to From the MIGraphX EP to speed up time to inference. - [ ] Save our precompiled graphs - [...
### Problem Description ### Parsing OCP FP8 Model This would require MIGraphX to expose E4M3FN data type into the IR. Currently only E4M3FNUZ type is exposed. It is probably not...
Streams allows async runs, and events to use more precise timing. Also fixes some dependency issues.
Add support for [GridSample](https://github.com/onnx/onnx/blob/main/docs/Operators.md#gridsample) onnx operator. Currently the following feature set is supported: * inputs: 4-D inputs * align_corners: both `0` and `1` * mode: `linear` and `nearest` * padding_mode:...
This change adds 2 "Model Zoo" to test: - ONNX Zoo (https://onnx.ai/models/ and https://github.com/onnx/models) - The script will use the repository, and does automatically the following steps: - Downloads the...
Remove the shifts added on the parser to maintain separation of concerns between our parser/code. Instead allow for us to add a pass to handle int8/uint8 interactions so that all...
~Related to #1904 ~ Originally from when we were parsing in dynamicquantizelinear but we've found other operators weren't handling inputs of uint8 correctly and thus we need to add a...