[Bug c++/96709] New: cmov and vectorization

2020-08-19 Thread g.peterh...@t-online.de
Priority: P3 Component: c++ Assignee: unassigned at gcc dot gnu.org Reporter: g.peterh...@t-online.de Target Milestone: --- Hello gcc team, I noticed 2 problems: 1) the compiler does not generate cmov commands 2) the auto-vectorization is very unreliable I would like to

[Bug target/96709] cmov and vectorization

2020-08-24 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=96709 --- Comment #2 from g.peterh...@t-online.de --- You can choose the boost version on godbolt.org. The example uses 1.73, but only the macros #define BOOST_FORCEINLINE inline __attribute__ ((__always_inline__)) and #define BOOST_NOINLINE

[Bug target/90492] simple array-copy not use available simd-registers

2019-12-20 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90492 g.peterh...@t-online.de changed: What|Removed |Added Known to fail||10.0 --- Comment #6 from

[Bug c++/90491] New: simple operation with unsigned integer and conversion to float/double not vectorized

2019-05-15 Thread g.peterh...@t-online.de
Severity: normal Priority: P3 Component: c++ Assignee: unassigned at gcc dot gnu.org Reporter: g.peterh...@t-online.de Target Milestone: --- snip #include #include int main(const int argc, const char** argv) { using value_type = float; // or double

[Bug c++/90492] New: simple array-copy not use available simd-registers

2019-05-15 Thread g.peterh...@t-online.de
Component: c++ Assignee: unassigned at gcc dot gnu.org Reporter: g.peterh...@t-online.de Target Milestone: --- snip #include #include int main(const int argc, const char** argv) { using value_type = int; // type does not matter using array_type = std::array

[Bug target/90492] simple array-copy not use available simd-registers

2019-05-15 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90492 --- Comment #3 from g.peterh...@t-online.de --- Am 15.05.19 um 21:20 schrieb glisse at gcc dot gnu.org: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90492 > > --- Comment #1 from Marc Glisse --- >> copy's use only sse

[Bug target/90492] simple array-copy not use available simd-registers

2019-05-15 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90492 --- Comment #4 from g.peterh...@t-online.de --- #include #include int main(const int argc, const char** argv) { using value_type = int64_t; using array_type = std::array; array_type a, b; for (size_t i=0

[Bug tree-optimization/90491] simple operation with unsigned integer and conversion to float/double not vectorized

2019-05-15 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90491 --- Comment #2 from g.peterh...@t-online.de --- example: #include #include int main(const int argc, const char** argv) { using value_type = float; using array_type = std::array; array_type a; for (size_t

[Bug target/90600] New: incompatible 64-bit-types in x86-intrinsics

2019-05-23 Thread g.peterh...@t-online.de
: P3 Component: target Assignee: unassigned at gcc dot gnu.org Reporter: g.peterh...@t-online.de Target Milestone: --- Host: x86-64 Target: x86-64 Build: 9.1.1 COLLECT_GCC=gcc-9 COLLECT_LTO_WRAPPER=/usr/lib64/gcc/x86_64-suse-linux

[Bug target/90600] incompatible 64-bit-types in x86-intrinsics

2019-05-23 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90600 --- Comment #2 from g.peterh...@t-online.de --- Am 23.05.19 um 19:04 schrieb jakub at gcc dot gnu.org: > Note, clang agrees with gcc here, and I don't think it is a good idea to > change > this incompatibly. I think it would be b

[Bug target/90600] incompatible 64-bit-types in x86-intrinsics

2019-05-23 Thread g.peterh...@t-online.de
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90600 --- Comment #4 from g.peterh...@t-online.de --- Am 23.05.19 um 20:11 schrieb glisse at gcc dot gnu.org: > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90600 > > --- Comment #3 from Marc Glisse --- > Intel documents that it us