[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 Richard Biener changed: What|Removed |Added Known to work||13.1.1, 14.0 Target Milestone|--- |11.5 Known to fail||13.1.0 Status|UNCONFIRMED |RESOLVED Resolution|--- |FIXED --- Comment #8 from Richard Biener --- Fixed.
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 --- Comment #7 from CVS Commits --- The releases/gcc-13 branch has been updated by hongtao Liu : https://gcc.gnu.org/g:ecc1af1f5b2c0fbcfa8840c79aa6102d413850b2 commit r13-7499-gecc1af1f5b2c0fbcfa8840c79aa6102d413850b2 Author: liuhongt Date: Tue Jun 20 15:41:00 2023 +0800 Refine maskloadmn pattern with UNSPEC_MASKLOAD. If mem_addr points to a memory region with less than whole vector size bytes of accessible memory and k is a mask that would prevent reading the inaccessible bytes from mem_addr, add UNSPEC_MASKLOAD to prevent it to be transformed to vpblendd. gcc/ChangeLog: PR target/110309 * config/i386/sse.md (maskload): Refine pattern with UNSPEC_MASKLOAD. (maskload): Ditto. (*_load_mask): Extend mode iterator to VI12HFBF_AVX512VL. (*_load): Ditto. gcc/testsuite/ChangeLog: * gcc.target/i386/pr110309.c: New test.
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 --- Comment #6 from CVS Commits --- The releases/gcc-12 branch has been updated by hongtao Liu : https://gcc.gnu.org/g:1f5591a9578b8cacda9d4c73a25d93598d68e028 commit r12-9742-g1f5591a9578b8cacda9d4c73a25d93598d68e028 Author: liuhongt Date: Tue Jun 20 15:41:00 2023 +0800 Refine maskloadmn pattern with UNSPEC_MASKLOAD. If mem_addr points to a memory region with less than whole vector size bytes of accessible memory and k is a mask that would prevent reading the inaccessible bytes from mem_addr, add UNSPEC_MASKLOAD to prevent it to be transformed to vpblendd. gcc/ChangeLog: PR target/110309 * config/i386/sse.md (maskload): Refine pattern with UNSPEC_MASKLOAD. (maskload): Ditto. (*_load_mask): Extend mode iterator to VI12HF_AVX512VL. (*_load): Ditto. gcc/testsuite/ChangeLog: * gcc.target/i386/pr110309.c: New test.
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 --- Comment #5 from CVS Commits --- The releases/gcc-11 branch has been updated by hongtao Liu : https://gcc.gnu.org/g:f59565f5dc2cdb5ac5a0b2b75404a36771232f86 commit r11-10883-gf59565f5dc2cdb5ac5a0b2b75404a36771232f86 Author: liuhongt Date: Tue Jun 20 15:41:00 2023 +0800 Refine maskloadmn pattern with UNSPEC_MASKLOAD. If mem_addr points to a memory region with less than whole vector size bytes of accessible memory and k is a mask that would prevent reading the inaccessible bytes from mem_addr, add UNSPEC_MASKLOAD to prevent it to be transformed to vpblendd. gcc/ChangeLog: PR target/110309 * config/i386/sse.md (maskload): Refine pattern with UNSPEC_MASKLOAD. (maskload): Ditto.
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 --- Comment #4 from Hongtao.liu --- Fixed for GCC14. Note: unspec is not added to maskstore since vpblendd doesn't support memeory dest, so there's no chance for a maskstore be optimized to vpblendd?
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 --- Comment #3 from CVS Commits --- The master branch has been updated by hongtao Liu : https://gcc.gnu.org/g:c79476da46728e2ab17e0e546262d2f6377081aa commit r14-2070-gc79476da46728e2ab17e0e546262d2f6377081aa Author: liuhongt Date: Tue Jun 20 15:41:00 2023 +0800 Refine maskloadmn pattern with UNSPEC_MASKLOAD. If mem_addr points to a memory region with less than whole vector size bytes of accessible memory and k is a mask that would prevent reading the inaccessible bytes from mem_addr, add UNSPEC_MASKLOAD to prevent it to be transformed to vpblendd. gcc/ChangeLog: PR target/110309 * config/i386/sse.md (maskload): Refine pattern with UNSPEC_MASKLOAD. (maskload): Ditto. (*_load_mask): Extend mode iterator to VI12HFBF_AVX512VL. (*_load): Ditto. gcc/testsuite/ChangeLog: * gcc.target/i386/pr110309.c: New test.
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309
--- Comment #2 from Hongtao.liu ---
(In reply to Richard Biener from comment #1)
> I think the two
>
> (define_expand "maskload"
> [(set (match_operand:V48H_AVX512VL 0 "register_operand")
> (vec_merge:V48H_AVX512VL
> (match_operand:V48H_AVX512VL 1 "memory_operand")
> (match_dup 0)
> (match_operand: 2 "register_operand")))]
> "TARGET_AVX512F")
>
> (define_expand "maskload"
> [(set (match_operand:VI12_AVX512VL 0 "register_operand")
> (vec_merge:VI12_AVX512VL
> (match_operand:VI12_AVX512VL 1 "memory_operand")
> (match_dup 0)
> (match_operand: 2 "register_operand")))]
> "TARGET_AVX512BW")
>
> patterns are wrong (all others use UNSPEC_MASKMOV)
UNSPEC_MASKMOV is used for MASKMOV, UNSPEC_MASK_LOAD is used for avx512 mask,
5;; If mem_addr points to a memory region with less than whole vector size
bytes
1386;; of accessible memory and k is a mask that would prevent reading the
inaccessible
1387;; bytes from mem_addr, add UNSPEC_MASKLOAD to prevent it to be
transformed to vpblendd
1388;; See pr97642.
1389(define_expand "_load_mask"
1390 [(set (match_operand:V48_AVX512VL 0 "register_operand")
1391(vec_merge:V48_AVX512VL
1392 (match_operand:V48_AVX512VL 1 "nonimmediate_operand")
1393 (match_operand:V48_AVX512VL 2 "nonimm_or_0_operand")
1394 (match_operand: 3
"register_or_constm1_operand")))]
1395 "TARGET_AVX512F"
1396{
1397 if (CONST_INT_P (operands[3]))
1398{
1399 emit_insn (gen_rtx_SET (operands[0], operands[1]));
1400 DONE;
1401}
1402 else if (MEM_P (operands[1]))
1403operands[1] = gen_rtx_UNSPEC (mode,
1404 gen_rtvec(1, operands[1]),
1405 UNSPEC_MASKLOAD);
1406})
1407
Yes, those 2 patterns needs to be fixed.
[Bug target/110309] Wrong code for masked load expansion
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110309 --- Comment #1 from Richard Biener --- I think the two (define_expand "maskload" [(set (match_operand:V48H_AVX512VL 0 "register_operand") (vec_merge:V48H_AVX512VL (match_operand:V48H_AVX512VL 1 "memory_operand") (match_dup 0) (match_operand: 2 "register_operand")))] "TARGET_AVX512F") (define_expand "maskload" [(set (match_operand:VI12_AVX512VL 0 "register_operand") (vec_merge:VI12_AVX512VL (match_operand:VI12_AVX512VL 1 "memory_operand") (match_dup 0) (match_operand: 2 "register_operand")))] "TARGET_AVX512BW") patterns are wrong (all others use UNSPEC_MASKMOV)
