Skip to content
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

Jump to bottom

fix node issue 49960 #519

Merged
merged 4 commits into from
Sep 30, 2023
Merged

fix node issue 49960 #519

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
ee47ec1
fix node issue 49960
lemire Sep 30, 2023
bda632a
add: slow path for short strings.
lemire Sep 30, 2023
50cf4d6
fixing alignment/sse
lemire Sep 30, 2023
5d87a05
format
Sep 30, 2023
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
2 changes: 1 addition & 1 deletion benchmarks/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -226,7 +226,7 @@ endif(ADA_BOOST_URL)
endif(Boost_FOUND)

# Zuri
find_package(ZURI)
find_package(ZURI QUIET)
if(ZURI_FOUND)
message(STATUS "Zuri found")
target_link_libraries(bench PRIVATE zuri)
Expand Down
32 changes: 26 additions & 6 deletions src/unicode.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,6 +48,17 @@ constexpr bool to_lower_ascii(char* input, size_t length) noexcept {
#if ADA_NEON
ada_really_inline bool has_tabs_or_newline(
std::string_view user_input) noexcept {
// first check for short strings in which case we do it naively.
if (user_input.size() < 16) { // slow path
for (size_t i = 0; i < user_input.size(); i++) {
if (user_input[i] == '\r' || user_input[i] == '\n' ||
user_input[i] == '\t') {
return true;
}
}
return false;
}
// fast path for long strings (expected to be common)
size_t i = 0;
const uint8x16_t mask1 = vmovq_n_u8('\r');
const uint8x16_t mask2 = vmovq_n_u8('\n');
Expand All @@ -60,9 +71,8 @@ ada_really_inline bool has_tabs_or_newline(
vceqq_u8(word, mask3));
}
if (i < user_input.size()) {
uint8_t buffer[16]{};
memcpy(buffer, user_input.data() + i, user_input.size() - i);
uint8x16_t word = vld1q_u8((const uint8_t*)user_input.data() + i);
uint8x16_t word =
vld1q_u8((const uint8_t*)user_input.data() + user_input.length() - 16);
running = vorrq_u8(vorrq_u8(running, vorrq_u8(vceqq_u8(word, mask1),
vceqq_u8(word, mask2))),
vceqq_u8(word, mask3));
Expand All @@ -72,6 +82,17 @@ ada_really_inline bool has_tabs_or_newline(
#elif ADA_SSE2
ada_really_inline bool has_tabs_or_newline(
std::string_view user_input) noexcept {
// first check for short strings in which case we do it naively.
if (user_input.size() < 16) { // slow path
for (size_t i = 0; i < user_input.size(); i++) {
if (user_input[i] == '\r' || user_input[i] == '\n' ||
user_input[i] == '\t') {
return true;
}
}
return false;
}
// fast path for long strings (expected to be common)
size_t i = 0;
const __m128i mask1 = _mm_set1_epi8('\r');
const __m128i mask2 = _mm_set1_epi8('\n');
Expand All @@ -85,9 +106,8 @@ ada_really_inline bool has_tabs_or_newline(
_mm_cmpeq_epi8(word, mask3));
}
if (i < user_input.size()) {
alignas(16) uint8_t buffer[16]{};
memcpy(buffer, user_input.data() + i, user_input.size() - i);
__m128i word = _mm_load_si128((const __m128i*)buffer);
__m128i word = _mm_loadu_si128(
(const __m128i*)(user_input.data() + user_input.length() - 16));
running = _mm_or_si128(
_mm_or_si128(running, _mm_or_si128(_mm_cmpeq_epi8(word, mask1),
_mm_cmpeq_epi8(word, mask2))),
Expand Down
Loading