Alexandre Eichenberger
Alexandre Eichenberger
Our default build listed in the README.txt (https://github.com/onnx/onnx-mlir#installation-on-unix) is as follows: ``` cmake --build . ``` and this attempts to build stuff for running the `check-onnx-backend`, which requires onnx. To...
##Possible areas of contributions Below is a lit of possible new contributions that new developers can contribute. We are always ready to help you perform new tasks, a good way...
Right now, we get very cryptic error messages, esp when we don't support an op. If it passes the frontend, then we get errors in the shape inference; if it...
With the new Keras models translated to TF and then to ONNX, we have the following pattern: ``` onnx.loop (...) { ... onnx.gemm(...) ''' } ``` Simplified to the extreme,...
As our compiler matures, we are tuning for performance and we would like to make sure we have to regressions. The proposal it so have a "performance" run that works...
New PR derived from #1709, which add data layout in ONNX that are good for SIMD exploitation. It is still in the original ONNX dialect as I believe it will...
Handling for leaky relu was recently added for NNPA, but it is not being tested and/or reported in the `NNPA/backend/CMakeList.txt` @imaihal or @negiyas: would you be able to add something...
Flag is ` -enable-compiler-stick-unstick` to enable (at this time) sickify only. When using the `-parallel` flag, it will use several OMP threads. At this time, the f32 to f16 is...
We are interested in generating multiple threading code in the extension library, which we compile when targeting the NNPA accelerator. This runtime is loaded into the .so model, and is...
Operations on the CPU are accelerated ultimately using the `omp` dialect in MLIR. We exploit it using the `krnl.parallel` which in turn activate the `affine.parallel` and `scf.parallel` or `scf.forall`. As...