https://gcc.gnu.org/bugzilla/show_bug.cgi?id=101471
Bug ID: 101471 Summary: AVX512 incorrect writemask generated for _mm512_fpclass_ps_mask in O0 Product: gcc Version: 11.1.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: target Assignee: unassigned at gcc dot gnu.org Reporter: kastianoza at gmail dot com Target Milestone: --- It seems that in debug mode gcc generates an 8-bit write opmask, which causes half of the result to be skipped when calling '_mm512_fpclass_ps_mask' (which takes 16 floats and produces 16 bit result). It doesn't happen in O1, when gcc generates instruction that doesn't take writemask. GCC info: Using built-in specs. COLLECT_GCC=g++-11 COLLECT_LTO_WRAPPER=/usr/lib/gcc/x86_64-linux-gnu/11/lto-wrapper OFFLOAD_TARGET_NAMES=nvptx-none:amdgcn-amdhsa OFFLOAD_TARGET_DEFAULT=1 Target: x86_64-linux-gnu Configured with: ../src/configure -v --with-pkgversion='Ubuntu 11.1.0-1ubuntu1~18.04.1' --with-bugurl=file:///usr/share/doc/gcc-11/README.Bugs --enable-languages=c,ada,c++,go,brig,d,fortran,objc,obj-c++,m2 --prefix=/usr --with-gcc-major-version-only --program-suffix=-11 --program-prefix=x86_64-linux-gnu- --enable-shared --enable-linker-build-id --libexecdir=/usr/lib --without-included-gettext --enable-threads=posix --libdir=/usr/lib --enable-nls --enable-clocale=gnu --enable-libstdcxx-debug --enable-libstdcxx-time=yes --with-default-libstdcxx-abi=new --enable-gnu-unique-object --disable-vtable-verify --enable-plugin --enable-default-pie --with-system-zlib --enable-libphobos-checking=release --with-target-system-zlib=auto --enable-objc-gc=auto --enable-multiarch --disable-werror --disable-cet --with-arch-32=i686 --with-abi=m64 --with-multilib-list=m32,m64,mx32 --enable-multilib --with-tune=generic --enable-offload-targets=nvptx-none=/build/gcc-11-YRKbe7/gcc-11-11.1.0/debian/tmp-nvptx/usr,amdgcn-amdhsa=/build/gcc-11-YRKbe7/gcc-11-11.1.0/debian/tmp-gcn/usr --without-cuda-driver --enable-checking=release --build=x86_64-linux-gnu --host=x86_64-linux-gnu --target=x86_64-linux-gnu Thread model: posix Supported LTO compression algorithms: zlib zstd gcc version 11.1.0 (Ubuntu 11.1.0-1ubuntu1~18.04.1) Compilation command: g++-11 fpclass_instruction.cpp -mavx512f -mavx512vl -mavx512bw -mavx512dq -ggdb -O0 minimal example: #include <immintrin.h> #include <iostream> int main() { __m512 x { 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, }; auto ret = _mm512_fpclass_ps_mask(x, 0x26); std::cout << ret << std::endl; } relevant part of preprocessed code: (notice incorrect mask type : __mmask8) # 7 "fpclass_instruction.cpp" int main() { __m512 x { 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, }; auto ret = # 13 "fpclass_instruction.cpp" 3 4 ((__mmask16) __builtin_ia32_fpclassps512_mask ((__v16sf) (__m512) ( # 13 "fpclass_instruction.cpp" x # 13 "fpclass_instruction.cpp" 3 4 ), (int) ( # 13 "fpclass_instruction.cpp" 0x26 # 13 "fpclass_instruction.cpp" 3 4 ),(__mmask8)-1)) # 13 "fpclass_instruction.cpp" ; std::cout << ret << std::endl; }