Skip to content

fix(kernel): GGUF JIT extension fails to build with gcc under C++17 - compile as C++20 #159

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

Open
avlp12 wants to merge 1 commit into
base: main
Choose a base branch
from
+7 −1
Open

fix(kernel): GGUF JIT extension fails to build with gcc under C++17 - compile as C++20 #159

Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
8 changes: 7 additions & 1 deletion python/freetoken/kernel/gguf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -51,7 +51,13 @@ def _c_compiler_for(cxx: str) -> str:
def _module():
from torch.utils.cpp_extension import load

extra_cuda_cflags = ["-O3", "--expt-relaxed-constexpr"]
# ``-std=c++20`` is load-bearing, not a modernization: nvcc's host pass rewrites
# ``static_cast<typename decltype(impl_->list)::difference_type>`` in libtorch's
# ``ATen/core/List_inl.h`` into a form that drops the ``typename``, which every
# g++ we have (12/13/15) rejects under C++17. C++20 (P0634) makes ``typename``
# implicit in a static_cast type-id, so the rewritten form compiles. torch only
# appends its own ``-std=c++17`` when no ``-std=`` is present, so this wins.
extra_cuda_cflags = ["-O3", "--expt-relaxed-constexpr", "-std=c++20"]
host_cxx = _host_compiler()
if host_cxx is not None:
# Point both nvcc's host pass (-ccbin) and torch's C++ compile (CXX) at a
Expand Down