zenoyang commented on code in PR #10084:
URL: https://github.com/apache/doris/pull/10084#discussion_r969180648
##########
be/src/util/simd/bits.h:
##########
@@ -89,16 +92,31 @@ inline size_t count_zero_num(const int8_t* __restrict data,
size_t size) {
return num;
}
-inline size_t count_zero_num(const int8_t* __restrict data, const uint8_t*
__restrict null_map,
Review Comment:
Because `__restrict` is redundant for `const` pointers. I see this function
is called in `VRuntimeFilterWrapper::execute` and cannot be remove.
##########
be/src/util/simd/bits.h:
##########
@@ -89,16 +92,31 @@ inline size_t count_zero_num(const int8_t* __restrict data,
size_t size) {
return num;
}
-inline size_t count_zero_num(const int8_t* __restrict data, const uint8_t*
__restrict null_map,
Review Comment:
Because `__restrict` is redundant for `const` pointers. I see this function
is called in `VRuntimeFilterWrapper::execute` and cannot be remove.
##########
be/src/vec/columns/column_decimal.cpp:
##########
@@ -240,13 +240,13 @@ ColumnPtr ColumnDecimal<T>::filter(const IColumn::Filter&
filt, ssize_t result_s
* completely pass or do not pass the filter.
* Therefore, we will optimistically check the parts of `SIMD_BYTES`
values.
*/
- static constexpr size_t SIMD_BYTES = 32;
+ static constexpr size_t SIMD_BYTES = 64;
const UInt8* filt_end_sse = filt_pos + size / SIMD_BYTES * SIMD_BYTES;
while (filt_pos < filt_end_sse) {
- uint32_t mask = simd::bytes32_mask_to_bits32_mask(filt_pos);
+ auto mask = simd::bytes64_mask_to_bits64_mask(filt_pos);
- if (0xFFFFFFFF == mask) {
+ if (0xFFFFFFFFFFFFFFFF == mask) {
Review Comment:
done
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
