Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

exclude marlin & exllama on windows #898

Merged
merged 5 commits into from
Dec 18, 2024
Merged

exclude marlin & exllama on windows #898

Changes from 1 commit
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
f103b69
exclude marlin & exllama on windows
CSY-ModelCloud Dec 18, 2024
bf7b8ee
validate as false when qlinear was not installed
CSY-ModelCloud Dec 18, 2024
0f130e5
fix list combination
CSY-ModelCloud Dec 18, 2024
5404a6b
simplify validate args
CSY-ModelCloud Dec 18, 2024
6022887
Merge remote-tracking branch 'origin/main' into CSY/exclude-ext
CSY-ModelCloud Dec 18, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
71 changes: 39 additions & 32 deletions setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -174,40 +174,47 @@ def get_version_tag(is_cuda_release: bool = True) -> str:
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
),
cpp_ext.CUDAExtension(
"gptqmodel_marlin_kernels",
[
"gptqmodel_ext/marlin/marlin_cuda.cpp",
"gptqmodel_ext/marlin/marlin_cuda_kernel.cu",
"gptqmodel_ext/marlin/marlin_repack.cu",
],
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
),
cpp_ext.CUDAExtension(
"gptqmodel_exllama_kernels",
[
"gptqmodel_ext/exllama/exllama_ext.cpp",
"gptqmodel_ext/exllama/cuda_buffers.cu",
"gptqmodel_ext/exllama/cuda_func/column_remap.cu",
"gptqmodel_ext/exllama/cuda_func/q4_matmul.cu",
"gptqmodel_ext/exllama/cuda_func/q4_matrix.cu",
],
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
),
cpp_ext.CUDAExtension(
"gptqmodel_exllamav2_kernels",
[
"gptqmodel_ext/exllamav2/ext.cpp",
"gptqmodel_ext/exllamav2/cuda/q_matrix.cu",
"gptqmodel_ext/exllamav2/cuda/q_gemm.cu",
],
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
)
]

if sys.platform != "win32":
extensions.append([
# TODO: VC++: fatal error C1061: compiler limit : blocks nested too deeply
cpp_ext.CUDAExtension(
"gptqmodel_marlin_kernels",
[
"gptqmodel_ext/marlin/marlin_cuda.cpp",
"gptqmodel_ext/marlin/marlin_cuda_kernel.cu",
"gptqmodel_ext/marlin/marlin_repack.cu",
],
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
),
# TODO: VC++: error lnk2001 unresolved external symbol cublasHgemm
cpp_ext.CUDAExtension(
"gptqmodel_exllama_kernels",
[
"gptqmodel_ext/exllama/exllama_ext.cpp",
"gptqmodel_ext/exllama/cuda_buffers.cu",
"gptqmodel_ext/exllama/cuda_func/column_remap.cu",
"gptqmodel_ext/exllama/cuda_func/q4_matmul.cu",
"gptqmodel_ext/exllama/cuda_func/q4_matrix.cu",
],
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
),
# TODO: VC++: error lnk2001 unresolved external symbol cublasHgemm
cpp_ext.CUDAExtension(
"gptqmodel_exllamav2_kernels",
[
"gptqmodel_ext/exllamav2/ext.cpp",
"gptqmodel_ext/exllamav2/cuda/q_matrix.cu",
"gptqmodel_ext/exllamav2/cuda/q_gemm.cu",
],
extra_link_args=extra_link_args,
extra_compile_args=extra_compile_args,
)
])

additional_setup_kwargs = {"ext_modules": extensions, "cmdclass": {"build_ext": cpp_ext.BuildExtension}}


Expand Down
Loading