Use in-place ops in _quantize_affine_float8 to reduce peak memory (#4068) #4155

	name: Run TorchAO Experimental MPS Tests
	on:
	push:
	branches:
	- main
	- 'gh/**'
	pull_request:
	branches:
	- main
	- 'gh/**'

	jobs:
	test-mps-ops:
	name: test-mps-ops
	uses: pytorch/test-infra/.github/workflows/macos_job.yml@main
	with:
	runner: macos-m1-stable
	python-version: '3.11'
	submodules: 'recursive'
	ref: ${{ github.event_name == 'pull_request' && github.event.pull_request.head.sha \|\| github.sha }}
	timeout: 90
	script: \|
	set -eux

	echo "::group::Install Torch"
	${CONDA_RUN} pip install torch --index-url "https://download.pytorch.org/whl/nightly/cpu"
	echo "::endgroup::"

	echo "::group::Install requirements"
	${CONDA_RUN} pip install -r dev-requirements.txt
	echo "::endgroup::"

	echo "::group::Install experimental MPS ops"
	${CONDA_RUN} USE_CPP=1 TORCHAO_BUILD_EXPERIMENTAL_MPS=1 pip install . --no-build-isolation
	echo "::endgroup::"

	echo "::group::Run lowbit tests"
	${CONDA_RUN} python -m pytest torchao/experimental/ops/mps/test/test_lowbit.py
	echo "::endgroup::"

	echo "::group::Run quantizer tests"
	${CONDA_RUN} python -m pytest torchao/experimental/ops/mps/test/test_quantizer.py
	echo "::endgroup::"

Provide feedback