Show patches with: Submitter = Liu, Hongtao       |    State = Action Required       |    Archived = No       |   701 patches
« 1 2 3 4 … 7 8 »
Patch Series A/F/R/T S/W/F Date Submitter Delegate State
Add TARGET_MMX_WITH_SSE to the condition of all 64-bit _Float16 vector related patterns. Add TARGET_MMX_WITH_SSE to the condition of all 64-bit _Float16 vector related patterns. - - - - --- 2026-01-12 Liu, Hongtao New
[v2,x86] Transform std::max(t, float(0)) into vmaxps w/o fast-math. [v2,x86] Transform std::max(t, float(0)) into vmaxps w/o fast-math. - - - - --- 2025-12-03 Liu, Hongtao New
[x86] Transform std::max(t, float(0)) into vmaxps w/o fast-math. [x86] Transform std::max(t, float(0)) into vmaxps w/o fast-math. - - - - --- 2025-12-02 Liu, Hongtao New
Refactor mgather/mscatter implementation. Refactor mgather/mscatter implementation. - - - - --- 2025-11-26 Liu, Hongtao New
Adjust testcase. Adjust testcase. - - - - --- 2025-11-24 Liu, Hongtao New
Also handle vptestnm + and15/and3 to just vptestnm. Also handle vptestnm + and15/and3 to just vptestnm. - - - - --- 2025-11-17 Liu, Hongtao New
Optimize kmov + kmov + or to kortest. Optimize kmov + kmov + or to kortest. - - - - --- 2025-11-13 Liu, Hongtao New
[x86] Fix unrecognizable insn of vpdpwuud. [x86] Fix unrecognizable insn of vpdpwuud. - - - - --- 2025-10-30 Liu, Hongtao New
[2/2,x86] Support reduc_sbool_and_scal_m for V{QI, SI, DI}mode. [1/2,x86] Support reduc_sbool_{and, ior, xor}_scal_m for avx512 kmask. - - - - --- 2025-10-22 Liu, Hongtao New
[1/2,x86] Support reduc_sbool_{and, ior, xor}_scal_m for avx512 kmask. [1/2,x86] Support reduc_sbool_{and, ior, xor}_scal_m for avx512 kmask. - - - - --- 2025-10-22 Liu, Hongtao New
Simplify avx512 vector integer comparison when 2 operands are known equal Simplify avx512 vector integer comparison when 2 operands are known equal - - - - --- 2025-10-21 Liu, Hongtao New
[v3] Deprecate -mstore-max= and related tuning. [v3] Deprecate -mstore-max= and related tuning. - - - - --- 2025-09-28 Liu, Hongtao New
[V2] Deprecate -mstore-max= and related tuning. [V2] Deprecate -mstore-max= and related tuning. - - - - --- 2025-09-28 Liu, Hongtao New
Deprecate -mmove-max= and related tuning. Deprecate -mmove-max= and related tuning. - - - - --- 2025-09-26 Liu, Hongtao New
[v2,x86] Disable vect unroll for znver2/Znver1. [v2,x86] Disable vect unroll for znver2/Znver1. - - - - --- 2025-09-23 Liu, Hongtao New
[x86] Disable vect unroll for znver2. [x86] Disable vect unroll for znver2. - - - - --- 2025-09-22 Liu, Hongtao New
[v3] Remove SPR/GNR/DMR from avx512_{move, store}_by pieces tune. [v3] Remove SPR/GNR/DMR from avx512_{move, store}_by pieces tune. - - - - --- 2025-09-17 Liu, Hongtao New
[v2] Remove SPR/GNR/DMR from avx512_move_by_pieces tune. [v2] Remove SPR/GNR/DMR from avx512_move_by_pieces tune. - - - - --- 2025-09-16 Liu, Hongtao New
Remove SPR/GNR/DMR from avx512_{move,store}_by pieces tune. Remove SPR/GNR/DMR from avx512_{move,store}_by pieces tune. - - - - --- 2025-09-16 Liu, Hongtao New
[x86] Optimize vpermpd to vbroadcastf128 for specific permutations. [x86] Optimize vpermpd to vbroadcastf128 for specific permutations. - - - - --- 2025-09-15 Liu, Hongtao New
[v3,x86] Exclude fake cross-lane permutation from avx256_avoid_vec_perm. [v3,x86] Exclude fake cross-lane permutation from avx256_avoid_vec_perm. - - - - --- 2025-09-08 Liu, Hongtao New
[v2,2/2,x86] Use vpermil{ps, pd} instead of vperm{d, q} when permutation is in-lane. [v2,1/2,x86] Exclude fake cross-lane permutation from avx256_avoid_vec_perm. - - - - --- 2025-09-05 Liu, Hongtao New
[v2,1/2,x86] Exclude fake cross-lane permutation from avx256_avoid_vec_perm. [v2,1/2,x86] Exclude fake cross-lane permutation from avx256_avoid_vec_perm. - - - - --- 2025-09-05 Liu, Hongtao New
[x86] Fix ICE due to wrong operand is passed to ix86_vgf2p8affine_shift_matrix. [x86] Fix ICE due to wrong operand is passed to ix86_vgf2p8affine_shift_matrix. - - - - --- 2025-08-30 Liu, Hongtao New
Document -param=ix86-vect-unroll-limit. Document -param=ix86-vect-unroll-limit. - - - - --- 2025-08-29 Liu, Hongtao New
Fix _Decimal128 arithmetic error under FE_UPWARD. Fix _Decimal128 arithmetic error under FE_UPWARD. - - - - --- 2025-08-28 Liu, Hongtao New
Restrict avx256_avoid_vec_perm only for loop vectorization. Restrict avx256_avoid_vec_perm only for loop vectorization. - - - - --- 2025-08-27 Liu, Hongtao New
[v2,x86] Enable unroll in the vectorizer when there's reduction for FMA/DOT_PROD_EXPR/SAD_EXPR [v2,x86] Enable unroll in the vectorizer when there's reduction for FMA/DOT_PROD_EXPR/SAD_EXPR - - - - --- 2025-08-11 Liu, Hongtao New
[x86] Enable unroll in the vectorizer when there's reduction for FMA/DOT_PROD_EXPR [x86] Enable unroll in the vectorizer when there's reduction for FMA/DOT_PROD_EXPR - - - - --- 2025-07-30 Liu, Hongtao New
Remove V64SFmode and V64SImode. Remove V64SFmode and V64SImode. - - - - --- 2025-07-30 Liu, Hongtao New
Eliminate redundant vpextrq/vpinsrq when move TI to V4SI. Eliminate redundant vpextrq/vpinsrq when move TI to V4SI. - - - - --- 2025-07-30 Liu, Hongtao New
Don't duplicate setup code cost when do group-candidate cost calucalution. Don't duplicate setup code cost when do group-candidate cost calucalution. - - - - --- 2025-06-24 Liu, Hongtao New
[x86,PR103750] Also handle avx512 kmask & immediate 15 or 3 when VF is 4/2. [x86,PR103750] Also handle avx512 kmask & immediate 15 or 3 when VF is 4/2. - - - - --- 2025-06-05 Liu, Hongtao New
[V2] For datarefs with big gap, split them into different groups. [V2] For datarefs with big gap, split them into different groups. - - - - --- 2025-05-27 Liu, Hongtao New
[AUTOFDO] Don't scale bb_count with ipa_count when ipa_count is zero but count_max is not [AUTOFDO] Don't scale bb_count with ipa_count when ipa_count is zero but count_max is not - - - - --- 2025-05-19 Liu, Hongtao New
[v3] Extend vect_recog_cond_expr_convert_pattern to handle REAL_CST [v3] Extend vect_recog_cond_expr_convert_pattern to handle REAL_CST - - - - --- 2025-05-19 Liu, Hongtao New
For datarefs with big gap, split them into different groups. For datarefs with big gap, split them into different groups. - - - - --- 2025-05-16 Liu, Hongtao New
Add pattern match in match.pd for .AVG_CEIL Add pattern match in match.pd for .AVG_CEIL - - - - --- 2025-05-15 Liu, Hongtao New
[v3] Extend vect_recog_cond_expr_convert_pattern to handle REAL_CST [v3] Extend vect_recog_cond_expr_convert_pattern to handle REAL_CST - - - - --- 2025-05-14 Liu, Hongtao New
[v3] Consider frequency in cost estimation when converting scalar to vector. [v3] Consider frequency in cost estimation when converting scalar to vector. - - - - --- 2025-05-14 Liu, Hongtao New
Update libbid according to the latest Intel Decimal Floating-Point Math Library. Update libbid according to the latest Intel Decimal Floating-Point Math Library. - - - - --- 2025-05-14 Liu, Hongtao New
[v2,2/2] Extend vect_recog_cond_expr_convert_pattern to handle REAL_CST [v2,1/2] Extend vect_recog_cond_expr_convert_pattern to handle floating point type. - - - - --- 2025-05-13 Liu, Hongtao New
[v2,1/2] Extend vect_recog_cond_expr_convert_pattern to handle floating point type. [v2,1/2] Extend vect_recog_cond_expr_convert_pattern to handle floating point type. - - - - --- 2025-05-13 Liu, Hongtao New
[v3] Consider frequency in cost estimation when converting scalar to vector. [v3] Consider frequency in cost estimation when converting scalar to vector. - - - - --- 2025-05-08 Liu, Hongtao New
[V2] Fix name mismatch for fortran. [V2] Fix name mismatch for fortran. - - - - --- 2025-05-07 Liu, Hongtao New
[V3,autofdo] Annotate empty bb with all debug_stmt with location of phi in the single_succ. [V3,autofdo] Annotate empty bb with all debug_stmt with location of phi in the single_succ. - - - - --- 2025-04-29 Liu, Hongtao New
[v2] Consider frequency in cost estimation when converting scalar to vector. [v2] Consider frequency in cost estimation when converting scalar to vector. - - - - --- 2025-04-29 Liu, Hongtao New
Remove other processors from X86_TUNE_DEST_FALSE_DEP_FOR_GLC except GLC Remove other processors from X86_TUNE_DEST_FALSE_DEP_FOR_GLC except GLC - - - - --- 2025-04-29 Liu, Hongtao New
Extend vect_recog_cond_expr_convert_pattern to handle floating point type. Extend vect_recog_cond_expr_convert_pattern to handle floating point type. - - - - --- 2025-04-28 Liu, Hongtao New
[v2,autofdo] Annotate empty bb with all debug_stmt with location of phi in the single_succ. [v2,autofdo] Annotate empty bb with all debug_stmt with location of phi in the single_succ. - - - - --- 2025-04-28 Liu, Hongtao New
[autofdo] Annotate bb with all debug_stmt with location of phi in the single_succ. [autofdo] Annotate bb with all debug_stmt with location of phi in the single_succ. - - - - --- 2025-04-28 Liu, Hongtao New
Fix name mismatch for fortran. Fix name mismatch for fortran. - - - - --- 2025-04-28 Liu, Hongtao New
Refactor msse4 and mno-sse4. Refactor msse4 and mno-sse4. - - - - --- 2025-04-25 Liu, Hongtao New
target: [PR103750] Also handle avx512 kmask & immediate 15 or 3 when VF is 4/2. target: [PR103750] Also handle avx512 kmask & immediate 15 or 3 when VF is 4/2. - - - - --- 2025-04-23 Liu, Hongtao New
Accept allones or 0 operand for vcond_mask op1. Accept allones or 0 operand for vcond_mask op1. - - - - --- 2025-04-21 Liu, Hongtao New
[x86] Generate 2 FMA instructions in ix86_expand_swdivsf. [x86] Generate 2 FMA instructions in ix86_expand_swdivsf. - - - - --- 2025-04-21 Liu, Hongtao New
Consider frequency in cost estimation when converting scalar to vector. Consider frequency in cost estimation when converting scalar to vector. - - - - --- 2025-04-18 Liu, Hongtao New
Revert documents from r11-344-g0fec3f62b9bfc0 Revert documents from r11-344-g0fec3f62b9bfc0 - - - - --- 2025-04-14 Liu, Hongtao New
Use ix86_fp_comparison_operator in cbranchbf4 to avoid ICE. Use ix86_fp_comparison_operator in cbranchbf4 to avoid ICE. - - - - --- 2025-03-19 Liu, Hongtao New
[testsuite] Mark gcc.target/i386/apx-ndd-tls-1b.c as xfail. [testsuite] Mark gcc.target/i386/apx-ndd-tls-1b.c as xfail. - - - - --- 2025-03-17 Liu, Hongtao New
[3/3] Adjust testcases after better RA decision. GCC13/GCC12 backport [PR108707][PR109610] - - - - --- 2025-02-10 Liu, Hongtao New
[2/3] Only use NO_REGS in cost calculation when !hard_regno_mode_ok for GENERAL_REGS and mode. GCC13/GCC12 backport [PR108707][PR109610] - - - - --- 2025-02-10 Liu, Hongtao New
[1/3] Use NO_REGS in cost calculation when the preferred register class are not known yet. GCC13/GCC12 backport [PR108707][PR109610] - - - - --- 2025-02-10 Liu, Hongtao New
[x86,avx512] Fix typo to avoid ICE. [x86,avx512] Fix typo to avoid ICE. - - - - --- 2025-01-16 Liu, Hongtao New
Refactor ix86_expand_vecop_qihi2. Refactor ix86_expand_vecop_qihi2. - - - - --- 2025-01-10 Liu, Hongtao New
[V2] Fix inaccuracy in cunroll/cunrolli when considering what's innermost loop. [V2] Fix inaccuracy in cunroll/cunrolli when considering what's innermost loop. - - - - --- 2024-12-10 Liu, Hongtao New
Fix inaccuracy in cunroll/cunrolli when considering what's innermost loop. Fix inaccuracy in cunroll/cunrolli when considering what's innermost loop. - - - - --- 2024-12-06 Liu, Hongtao New
[x86,RFC] Prevent loop vectorization if it's in a deeply nested big loop. [x86,RFC] Prevent loop vectorization if it's in a deeply nested big loop. - - - - --- 2024-11-27 Liu, Hongtao New
[x86] Fix uninitialized operands[2] in vec_unpacks_hi_v4sf. [x86] Fix uninitialized operands[2] in vec_unpacks_hi_v4sf. - - - - --- 2024-11-22 Liu, Hongtao New
Guard truncate from vector float to vector __bf16 with !flag_rounding_math && HONOR_NANS (BFmode). Guard truncate from vector float to vector __bf16 with !flag_rounding_math && HONOR_NANS (BFmode). - - - - --- 2024-11-08 Liu, Hongtao New
Make ix86_align_loops uarch-specific tune. Make ix86_align_loops uarch-specific tune. - - - - --- 2024-11-07 Liu, Hongtao New
Fix ICE due to subreg:us_truncate. Fix ICE due to subreg:us_truncate. - - - - --- 2024-10-30 Liu, Hongtao New
[2/2] Support vector float_extend from __bf16 to float. [1/2,x86] Support vector float_truncate for SF to BF. - - - - --- 2024-10-30 Liu, Hongtao New
[1/2,x86] Support vector float_truncate for SF to BF. [1/2,x86] Support vector float_truncate for SF to BF. - - - - --- 2024-10-30 Liu, Hongtao New
[x86] Fix ICE due to isa mismatch for the builtins. [x86] Fix ICE due to isa mismatch for the builtins. - - - - --- 2024-10-23 Liu, Hongtao New
[GCC13/GCC12] Fix testcase. [GCC13/GCC12] Fix testcase. - - - - --- 2024-10-22 Liu, Hongtao New
i386: Optimize EQ/NE comparison between avx512 kmask and -1. i386: Optimize EQ/NE comparison between avx512 kmask and -1. - - - - --- 2024-10-22 Liu, Hongtao New
[AVX512] Refine splitters related to "combine vpcmpuw + zero_extend to vpcmpuw" [AVX512] Refine splitters related to "combine vpcmpuw + zero_extend to vpcmpuw" - - - - --- 2024-10-17 Liu, Hongtao New
Adjust testcase to avoid scan FIX in REG_EQUIV. Adjust testcase to avoid scan FIX in REG_EQUIV. - - - - --- 2024-10-15 Liu, Hongtao New
[wwwdoc] Mention O2 vectorization enhancement. [wwwdoc] Mention O2 vectorization enhancement. - - - - --- 2024-10-15 Liu, Hongtao New
[2/2,x86] Canonicalize (vec_merge (fma: op2 op1 op3) (match_dup 1)) mask) to (vec_merge (fma: op1 o… Canonicalize (vec_merge (fma op1 op2 op3) op1 mask) to (vec_merge (fma op1 op2 op3) op1 mask) - - - - --- 2024-10-15 Liu, Hongtao New
[1/2,Middle-end] Canonicalize (vec_merge (fma op2 op1 op3) op1 mask) to (vec_merge (fma op1 op2 op3… Canonicalize (vec_merge (fma op1 op2 op3) op1 mask) to (vec_merge (fma op1 op2 op3) op1 mask) - - - - --- 2024-10-15 Liu, Hongtao New
[v3,2/2] Adjust testcase after relax O2 vectorization. [v3,1/2] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vec… - - - - --- 2024-10-09 Liu, Hongtao New
[v3,1/2] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vec… [v3,1/2] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vec… - - - - --- 2024-10-09 Liu, Hongtao New
Don't lower vpcmpu to pcmpgt since the latter is for signed comparison. Don't lower vpcmpu to pcmpgt since the latter is for signed comparison. - - - - --- 2024-10-09 Liu, Hongtao New
[2/2,x86] Add a new tune avx256_avoid_vec_perm for SRF. Enable more SRF tuning - - - - --- 2024-10-08 Liu, Hongtao New
[1/2,x86] Add new microarchitecture tune for SRF/GRR/CWF. Enable more SRF tuning - - - - --- 2024-10-08 Liu, Hongtao New
[v2,2/2] Adjust testcase after relax O2 vectorization. [v2,1/2] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vec… - - - - --- 2024-10-08 Liu, Hongtao New
[v2,1/2] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vec… [v2,1/2] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vec… - - - - --- 2024-10-08 Liu, Hongtao New
[x86] Define VECTOR_STORE_FLAG_VALUE [x86] Define VECTOR_STORE_FLAG_VALUE - - - - --- 2024-09-24 Liu, Hongtao New
[RFC] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vector… [RFC] Enable vectorization for unknown tripcount in very cheap cost model but disable epilog vector… - - - - --- 2024-09-11 Liu, Hongtao New
Enable tune fuse_move_and_alu for GNR/GNR-D. Enable tune fuse_move_and_alu for GNR/GNR-D. - - - - --- 2024-09-10 Liu, Hongtao New
Don't force_reg operands[3] when it's not const0_rtx. Don't force_reg operands[3] when it's not const0_rtx. - - - - --- 2024-09-09 Liu, Hongtao New
Handle const0_operand for *avx2_pcmp<mode>3_1. Handle const0_operand for *avx2_pcmp<mode>3_1. - - - - --- 2024-09-05 Liu, Hongtao New
[x86] Check avx upper register for parallel. [x86] Check avx upper register for parallel. - - - - --- 2024-08-30 Liu, Hongtao New
[x86] Check avx upper register for parallel. [x86] Check avx upper register for parallel. - - - - --- 2024-08-29 Liu, Hongtao New
[v2,2/2,x86] Update ix86_mode_tieable_p and ix86_rtx_costs. - - - - --- 2024-08-27 Liu, Hongtao New
[v2,1/2] Enhance cse_insn to handle all-zeros and all-ones for vector mode. [v2,1/2] Enhance cse_insn to handle all-zeros and all-ones for vector mode. - - - - --- 2024-08-27 Liu, Hongtao New
[2/2,x86] Update ix86_mode_tieable_p and ix86_rtx_costs. [1/2] Enhance cse_insn to handle all-zeros and all-ones for vector mode. - - - - --- 2024-08-26 Liu, Hongtao New
[1/2] Enhance cse_insn to handle all-zeros and all-ones for vector mode. [1/2] Enhance cse_insn to handle all-zeros and all-ones for vector mode. - - - - --- 2024-08-26 Liu, Hongtao New
« 1 2 3 4 … 7 8 »