-
Windows 10, with AMD Radeon RX 6600.
Version 524 of the exe is functional. Version 700, and (because I tried it in hopes of a slightly older version still working) 693, both go through the exact sa…
-
Hi all,
We (IREE@Google) are hitting a gap in the RISC-V backend when lowering a masked vector reduction for a vector of size 1.
`repro.ll`:
```ll
declare float @llvm.vp.reduce.fadd.v1f32(floa…
-
### What happened?
When trying to compile mlir file exported from mixed percision LLama2, getting the below error:
```
root@aiinfra-C9X299-PGF:/home/admin/iree-dist# ./bin/iree-compile --iree-hal-t…
-
### What happened?
For the given IR
```
module {
func.func @main_graph(%arg0: !torch.vtensor, %arg1: !torch.vtensor, %arg2: !torch.vtensor, %arg3: !torch.vtensor, %arg4: !torch.vtensor, %arg5:…
-
### What happened?
Running stable diffusion FP16 on CPU crashes the IREE compiler.
anush@anush-5900xt ~/github/SHARK
% IREE_SAVE_TEMPS=save_tmps python3.10 shark/examples/shark_inference/sta…
-
There's a few missing bits to get our API really C compatible; or at least there are some places that are not 32-bit (x86) compatible.
Example: mismatched calling conventions:
```
[build] D:\Dev\…
-
I tried to implement i4 -> f16 conversion with a lookup table implemented using a dynamic vector extract:
```mlir
func.func @main(%30 : tensor
kuhar updated
6 months ago
-
### What happened?
Got error `INTERNAL; CUDA driver error 'CUDA_ERROR_INVALID_VALUE' (1): invalid argument` when running iree runtime
### Steps to reproduce your issue
```
import numpy as np
impo…
-
### What happened?
We have IR for SDXL that preserves the torch.aten._scaled_dot_product_attention op as tm_tensor.attention -> iree_linalg_ext.attention, but I'm seeing issues trying to lower the …
-
### What happened?
I am trying to run the colab example but it fails.
### Steps to reproduce your issue
```
python3 -mvenv venv
source venv/bin/activate
pip install --pre torch-mlir -f htt…