whytofly
/
qt6-bb10
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

qHash: update the pre-AVX512 mask to use the cacheline size 

No change in behavior. But instead of trying to guess if we're going to
cross a page boundary, we check for the cacheline: if we don't cross a
cacheline boundary, then we can't cross a page boundary either. Testing
bit 4 can be implemented with a shorter instruction than a test for bit
11.

   f43de:       40 f6 c7 20             test   $0x20,%dil

Change-Id: I664b9f014ffc48cbb49bfffd17b04817a0fb8c6b
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>
bb10
Thiago Macieira 2024-02-02 22:58:45 -08:00
parent 55959aefab
commit fff9f5047a
1 changed files with 4 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
src/corelib/tools/qhash.cpp
View File

@ -756,16 +756,16 @@ aeshash128_lt16(__m128i state0, const __m128i *src, const __m128i *srcend, size_
// including NULLs at the end because the length is in the key)
// WARNING: this may produce valgrind warnings, but it's safe
constexpr quintptr PageSize = 4096;
constexpr quintptr CachelineSize = 64;
__m128i data;
if ((quintptr(src) & (PageSize / 2)) == 0) {
// lower half of the page:
if ((quintptr(src) & (CachelineSize / 2)) == 0) {
// lower half of the cacheline:
__m128i mask = _mm_loadu_si128(reinterpret_cast<const __m128i *>(maskarray + 15 - len));
data = loadu128<ZX>(src);
data = _mm_and_si128(data, mask);
} else {
// upper half of the page:
// upper half of the cacheline:
__m128i control = _mm_loadu_si128(reinterpret_cast<const __m128i *>(shufflecontrol + 15 - len));
data = loadu128<ZX>(advance<ZX>(srcend, -1));
data = _mm_shuffle_epi8(data, control);