[Optimizer] DQ + MatMul to MatMulNBits support: kernel changes (#21342) #1

	name: Windows_CI
	on:
	push:
	branches:
	- main
	- rel-*
	pull_request:

	concurrency:
	group: ${{ github.workflow }}-${{ github.head_ref \|\| github.run_id }}
	cancel-in-progress: true

	env:
	AZCOPY_AUTO_LOGIN_TYPE: MSI
	AZCOPY_MSI_CLIENT_ID: 63b63039-6328-442f-954b-5a64d124e5b4

	jobs:
	Windows-CUDA-12:
	runs-on: ["self-hosted", "1ES.Pool=onnxruntime-github-vs2022-mms"]
	steps:
	- uses: actions/checkout@v4
	with:
	submodules: false
	- uses: actions/setup-python@v5
	with:
	python-version: '3.11.x'
	architecture: 'x64'

	- uses: actions/setup-node@v4
	with:
	node-version: 18

	- name: Download cuda
	run: azcopy.exe cp --recursive "https://lotusscus.blob.core.windows.net/models/cuda_sdk/v12.2" cuda_sdk


	- name: Delete build folder
	run: \|
	if (Test-Path D:\b) { Remove-Item -Recurse -Force D:\b }
	&tools\ci_build\github\windows\install_third_party_deps.ps1 -cpu_arch x64 -install_prefix D:\b\Debug\installed -build_config Debug

	# The build machine doesn't have a GPU. So the value of CMAKE_CUDA_ARCHITECTURES doesn't matter.
	- name: Build code
	run: python tools\ci_build\build.py --windows_sdk_version 10.0.22621.0 --enable_training --build_java --config Debug --build_dir D:\b --skip_submodule_sync --build_csharp --update --build --parallel --cmake_generator "Visual Studio 17 2022" --build_shared_lib --enable_pybind --use_cuda --cuda_home=${{ github.workspace }}\cuda_sdk\v12.2 --enable_cuda_profiling --cmake_extra_defines CMAKE_CUDA_ARCHITECTURES=75

Provide feedback