Rebuild for CUDA 12

[h-vetinari]

Bot couldn't open the PR for some reason (so the restart from #2 never worked), let's try doing it manually; based on #12

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

[jakirkham]

Thanks Axel! 🙏

Had a couple minor suggestions below that may help this clear

[jakirkham]

If this is loading libcuda.so (which requires a working GPU) and it may be doing this, we would need to skip this test

Suggested change

	imports:
	- awq
	imports: # [cuda_compiler_version != "None"]
	- awq # [cuda_compiler_version != "None"]

Alternatively we could switch to a Quansight GPU runner

[h-vetinari]

It fails to find libtorch_python.so:

import: 'awq_ext'
Traceback (most recent call last):
  File "/home/conda/feedstock_root/build_artifacts/autoawq_1723203836924/test_tmp/run_test.py", line 2, in <module>
    import awq_ext
ImportError: libtorch_python.so: cannot open shared object file: No such file or directory

I don't understand how the build manages to link at buildtime, but not at runtime. Well, -Wl,--allow-shlib-undefined might explain some of it (didn't succeed in stripping this it seems), but -L$SP_DIR/torch/lib is in LDFLAGS, so I have no idea where this fails. I doubt it's the cuda drivers though.

[jakirkham]

IIUC this was an issue in PyTorch packaging that was being worked on in PR: conda-forge/pytorch-cpu-feedstock#246

AIUI this was added to PyTorch 2.4.0: conda-forge/pytorch-cpu-feedstock#250

So likely we need to update PyTorch here too

[jakirkham]

Also worth noting the migrator and version updates got a bit clogged up due to a duplicate key in conda-forge.yml: #14

That is now resolved. So a bunch of bot PRs just got added to this feedstock

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

I do have some suggestions for making it better though...

For recipe/meta.yaml:

• No valid build backend found for Python recipe for package autoawq_kernels using pip. Python recipes using pip need to explicitly specify a build backend in the host section. If your recipe has built with only pip in the host section in the past, you likely should add setuptools to the host section of your recipe.

[conda-forge-webservices]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe/meta.yaml) and found it was in an excellent condition.

[h-vetinari]

Given conda/conda-build#5467, I'm going to merge this for now and then see if I can layer the rest on top.