Fix `USE_STATIC_MKL` lost functionality#138996

Draft

xuhancn wants to merge 16 commits intopytorch:mainfrom

xuhancn:xu_fix_USE_STATIC_MKL_lost_functionality

Commits on Aug 9, 2025

Fix USE_STATIC_MKL

xuhancn
authored and
pytorchmergebot
committed
setup USE_STATIC_MKL=1

xuhancn
authored and
pytorchmergebot
committed
update code.

xuhancn
authored and
pytorchmergebot
committed
clean env

xuhancn
authored and
pytorchmergebot
committed
setup USE_STATIC_MKL=1 for wheel build.

xuhancn
authored and
pytorchmergebot
committed
export USE_STATIC_MKL=1 for linux wheel build.

xuhancn
authored and
pytorchmergebot
committed
update code.

xuhancn
authored and
pytorchmergebot
committed
try to fix torch_cuda_linalg mkl dependency.

xuhancn
authored and
pytorchmergebot
committed
update code.

xuhancn
authored and
pytorchmergebot
committed
test code.

xuhancn
authored and
pytorchmergebot
committed
try to fix issue 146551

xuhancn
authored and
pytorchmergebot
committed
use static mkl for XPU build.

xuhancn
authored and
pytorchmergebot
committed
try to fix xla.

xuhancn
authored and
pytorchmergebot
committed
add comments for torch_cuda_linalg fixing.

xuhancn
authored and
pytorchmergebot
committed
clean up useless files.

xuhancn
authored and
pytorchmergebot
committed
setup mkl build for cuda_v129

xuhancn
authored and
pytorchmergebot
committed