m1k3

tiramisu aligned with llvm 6? Bring Halide up to date.

4

Tiramisu has dependencies with LLVM/Clang 5 because Halide branch is outdated. The same could be said with ISL to be updated to 0.20. Any idea when this updated alignment would...

Regression issue? Many halide tests slower with tiramisu on Mac, or crashing

3

The following tests are significantly slower with Tiramisu vs Halide: * blurxy * convolution, convolution_layer * gaussian * vgg * warp-affine Others fails: * recfilter — segfault * heat2d, heat3d...

Updated code for conda, python 3.7, PyTorch 1.8, cuda 10/11 on ubuntu 20.04

4

This PR updates AIMET to almost latest versions of various dependencies in a conda environment. - aimet environment in packaging/environment.yml - Python 3.7 - google test 1.10 - cuda 10...

DNNL partly implemented - won't compile

2

The code seems to miss BroadcastAdd, FuseConvBatchNormBias, and FuseConvBatchNormAlphaVar operators implementation in C or with DNNL so the code won't compile. Would you plan to support DNNL fully?

multi-gpu results WRONG

1

I couldn't reproduce the results with the examples provided in the readme on my 4 GPUs. So I used batch 256 on only 1 GPU and it works. Adding one...

Using the example with --device cpu leads to crash: python inference/inference_sim.py -a resnet50 -b 512 -pcq_w -pcq_a -sh --qtype int4 -qw int4 --device cpu The bug comes from calling torch.cuda.clear()....

on Mac M3, under conda python 3.11, pytorch 2.2 torch-mlir main branch, built in-tree running projects/pt1/examples/torchscript_stablehlo_backend_tinybert.py - with output_type='tosa' ``` output_type='tosa' Traceback (most recent call last): File "torch-mlir/projects/pt1/examples/torchscript_stablehlo_backend_tinybert.py", line 24,...

Graph or model_exporter doesn't preserve list/tuple and dict outputs

5

### Issue Type Feature Request ### Source source ### MCT Version main ### OS Platform and Distribution ubuntu 20.04, pytorch 2 ### Python version 3.11 ### Describe the issue ```shell...

bad list release?

1

the last release has only 27k sites understood by transmission, while the list clearly has many more. bug?