[gcc r16-4188] Disallow mask reduction vectorization

https://gcc.gnu.org/g:e5ae5b5ba6f2af408717034d82c40839642667b0 commit r16-4188-ge5ae5b5ba6f2af408717034d82c40839642667b0 Author: Richard Biener Date: Thu Oct 2 14:41:01 2025 +0200 Disallow mask reduction vectorization When you trick bool pattern recognition to use a mask type for

[gcc r16-4180] tree-optimization/122079 - PRE antic compute doesn't converge

https://gcc.gnu.org/g:0f8c6f479e65af985d4d374cc9b3405327124d66 commit r16-4180-g0f8c6f479e65af985d4d374cc9b3405327124d66 Author: Richard Biener Date: Wed Oct 1 14:55:17 2025 +0200 tree-optimization/122079 - PRE antic compute doesn't converge The following re-instantiates the pruni

[gcc r16-4179] tree-optimization/122079 - PRE antic_compute doesn't converge

https://gcc.gnu.org/g:0e91910c0708a60b3ac521a4ea74e44301109f4a commit r16-4179-g0e91910c0708a60b3ac521a4ea74e44301109f4a Author: Richard Biener Date: Wed Oct 1 14:16:50 2025 +0200 tree-optimization/122079 - PRE antic_compute doesn't converge The following fixes another case of us

[gcc r16-4156] tree-optimization/122110 - do not reject all bit-precision reductions

https://gcc.gnu.org/g:c6865e7e15bc9a1337df00d2ca03604e1712a2dd commit r16-4156-gc6865e7e15bc9a1337df00d2ca03604e1712a2dd Author: Richard Biener Date: Wed Oct 1 11:26:45 2025 +0200 tree-optimization/122110 - do not reject all bit-precision reductions We can handle bitwise-operation

[gcc r16-4075] Fix get_load_store_type wrt VMAT_ELEMENTWISE classification

https://gcc.gnu.org/g:191d8b846c999e538f34552d14cb96f424f5490a commit r16-4075-g191d8b846c999e538f34552d14cb96f424f5490a Author: Richard Biener Date: Wed Sep 24 12:24:11 2025 +0200 Fix get_load_store_type wrt VMAT_ELEMENTWISE classification We may not classify a BB vectorization l

[gcc r15-10357] tree-optimization/122016 - PRE insertion breaks abnormal coalescing

https://gcc.gnu.org/g:b9e703f0b42b2a86c9e18cd12497a0c19d2eb96c commit r15-10357-gb9e703f0b42b2a86c9e18cd12497a0c19d2eb96c Author: Richard Biener Date: Mon Sep 22 10:14:31 2025 +0200 tree-optimization/122016 - PRE insertion breaks abnormal coalescing When PRE asks VN to simplify a

[gcc r16-4076] tree-optimization/116816 - improve VMAT_ELEMENTWISE with SLP

https://gcc.gnu.org/g:1f6b1ed047105103c2fba9696fd0ed17ec5e1405 commit r16-4076-g1f6b1ed047105103c2fba9696fd0ed17ec5e1405 Author: Richard Biener Date: Wed Sep 24 12:19:17 2025 +0200 tree-optimization/116816 - improve VMAT_ELEMENTWISE with SLP The following implements VMAT_ELEMENTWI

[gcc r15-10353] tree-optimization/121527 - wrong SRA with aggregate copy

https://gcc.gnu.org/g:96777e13668a829dcbd3c75071b0fcfd7849e649 commit r15-10353-g96777e13668a829dcbd3c75071b0fcfd7849e649 Author: Richard Biener Date: Mon Aug 18 13:38:37 2025 +0200 tree-optimization/121527 - wrong SRA with aggregate copy SRA handles outermost VIEW_CONVERT_EXPRs b

[gcc r15-10352] tree-optimization/121370 - avoid UB in building a CHREC

https://gcc.gnu.org/g:6348c1c263b6b70d11cbe4f178463db7fba185b6 commit r15-10352-g6348c1c263b6b70d11cbe4f178463db7fba185b6 Author: Richard Biener Date: Tue Aug 5 08:59:18 2025 +0200 tree-optimization/121370 - avoid UB in building a CHREC When there is obvious UB involved in the pro

[gcc r16-4020] tree-optimization/122016 - PRE insertion breaks abnormal coalescing

https://gcc.gnu.org/g:52b53cfb5392257536fce8dd22ea60412ceaee7d commit r16-4020-g52b53cfb5392257536fce8dd22ea60412ceaee7d Author: Richard Biener Date: Mon Sep 22 10:14:31 2025 +0200 tree-optimization/122016 - PRE insertion breaks abnormal coalescing When PRE asks VN to simplify a N

[gcc r16-4019] tree-optimization/122023 - rotate pattern with reductions

https://gcc.gnu.org/g:32a8a1c26bd0bf340ab4a5138b694d4102290a2e commit r16-4019-g32a8a1c26bd0bf340ab4a5138b694d4102290a2e Author: Richard Biener Date: Mon Sep 22 09:41:05 2025 +0200 tree-optimization/122023 - rotate pattern with reductions The following disables the use of rotate p

[gcc r16-3983] Remove accidentially left if (0) block

https://gcc.gnu.org/g:95fc85aff1c26184d0fe52ea4e9f47064279126a commit r16-3983-g95fc85aff1c26184d0fe52ea4e9f47064279126a Author: Richard Biener Date: Fri Sep 19 11:11:41 2025 +0200 Remove accidentially left if (0) block The following removes a block I added (and disabled again) wh

[gcc r16-3968] Remove DR_GROUP_STORE_COUNT

https://gcc.gnu.org/g:49431232988d8294bc351ff2edcaad8fce088b70 commit r16-3968-g49431232988d8294bc351ff2edcaad8fce088b70 Author: Richard Biener Date: Fri Sep 19 09:29:04 2025 +0200 Remove DR_GROUP_STORE_COUNT This was only used for non-SLP. * tree-vectorizer.h (_s

[gcc r16-3966] Remove SLP_TREE_NUMBER_OF_VEC_STMTS

https://gcc.gnu.org/g:140a29ae847aa2e0a4aa1ec4d3066454f6db5e9c commit r16-3966-g140a29ae847aa2e0a4aa1ec4d3066454f6db5e9c Author: Richard Biener Date: Thu Sep 18 11:32:09 2025 +0200 Remove SLP_TREE_NUMBER_OF_VEC_STMTS The following removes the redundant SLP_TREE_NUMBER_OF_VEC_STMTS

[gcc r16-3967] Cleanup vect_get_num_copies API

https://gcc.gnu.org/g:11e09c9ad814c1aa23e87d78246d35eb5ba28d66 commit r16-3967-g11e09c9ad814c1aa23e87d78246d35eb5ba28d66 Author: Richard Biener Date: Thu Sep 18 14:19:40 2025 +0200 Cleanup vect_get_num_copies API The following removes the dual non-SLP/SLP API in favor of only

[gcc r16-3663] tree-optimization/121830 - SLP cycle detection confused by nested cycle

https://gcc.gnu.org/g:f7426ba6c0d4f779ff0f2f84e8beeadc88ebe47c commit r16-3663-gf7426ba6c0d4f779ff0f2f84e8beeadc88ebe47c Author: Richard Biener Date: Mon Sep 8 13:25:37 2025 +0200 tree-optimization/121830 - SLP cycle detection confused by nested cycle The SLP reduc-index computati

[gcc r16-3947] tree-optimization/87615 - VN predication is expensive

https://gcc.gnu.org/g:d5e80bf7571836bb307f38e45f0e6ba392cdcc3d commit r16-3947-gd5e80bf7571836bb307f38e45f0e6ba392cdcc3d Author: Richard Biener Date: Thu Sep 18 10:41:45 2025 +0200 tree-optimization/87615 - VN predication is expensive The following restricts the number of location

[gcc r16-3945] tree-optimization/121720 - missed PRE hoisting

https://gcc.gnu.org/g:c30f58c3f7ec25a0ecb8d9bd36f8c90270e61993 commit r16-3945-gc30f58c3f7ec25a0ecb8d9bd36f8c90270e61993 Author: Richard Biener Date: Wed Sep 17 12:56:56 2025 +0200 tree-optimization/121720 - missed PRE hoisting The following re-implements the fix for PR84830 where

[gcc r16-3846] Avoid VMAT_ELEMENTWISE for negative stride SLP

https://gcc.gnu.org/g:57c97de978332fa5683393b38a2db77b78bf0e2c commit r16-3846-g57c97de978332fa5683393b38a2db77b78bf0e2c Author: Richard Biener Date: Fri Sep 12 14:15:59 2025 +0200 Avoid VMAT_ELEMENTWISE for negative stride SLP The following makes us always use VMAT_STRIDED_SLP fo

[gcc r16-3930] [gimplefe] fix SSA operand creation

https://gcc.gnu.org/g:f8cf09130b75ea8e310debbc0a165f5906ee56d6 commit r16-3930-gf8cf09130b75ea8e310debbc0a165f5906ee56d6 Author: Richard Biener Date: Wed Sep 17 10:22:26 2025 +0200 [gimplefe] fix SSA operand creation When transitioning gcc.dg/torture/pr84830.c to a GIMPLE testcase

[gcc r16-3794] Deal with prior EH/abormal cleanup when fixing up noreturn calls

https://gcc.gnu.org/g:5c4f1313e753aeb6920a48c62c7c99ad36e1adae commit r16-3794-g5c4f1313e753aeb6920a48c62c7c99ad36e1adae Author: Richard Biener Date: Wed Sep 10 17:14:07 2025 +0200 Deal with prior EH/abormal cleanup when fixing up noreturn calls When a dead EH or abnormal edge mak

[gcc r16-3908] Unify last two vect_transform_slp_perm_load calls

https://gcc.gnu.org/g:e717aabb6eff417a613a1247124bbd817d1414b3 commit r16-3908-ge717aabb6eff417a613a1247124bbd817d1414b3 Author: Richard Biener Date: Mon Sep 15 14:03:00 2025 +0200 Unify last two vect_transform_slp_perm_load calls The following unifies the vect_transform_slp_perm_

[gcc r16-3849] lto/121935 - visit all DECL_ARGUMENTS in free-lang-data

https://gcc.gnu.org/g:38666cbccff5114e2f23930fae180f03c385eb45 commit r16-3849-g38666cbccff5114e2f23930fae180f03c385eb45 Author: Richard Biener Date: Sun Sep 14 11:01:11 2025 +0200 lto/121935 - visit all DECL_ARGUMENTS in free-lang-data With no longer visiting TREE_CHAIN for decls

[gcc r16-3847] Integrate SLP permute transform into vect_transform_stmt

https://gcc.gnu.org/g:21d1bb1922fa7cd98f2a27d002e48b488e772176 commit r16-3847-g21d1bb1922fa7cd98f2a27d002e48b488e772176 Author: Richard Biener Date: Fri Sep 12 13:52:51 2025 +0200 Integrate SLP permute transform into vect_transform_stmt This adds permute_info_type and removes the

[gcc r16-3845] Do less redundant vect_transform_slp_perm_load calls

https://gcc.gnu.org/g:80f9440b5ecb1c3ab943b5c862f84e06fbeabd89 commit r16-3845-g80f9440b5ecb1c3ab943b5c862f84e06fbeabd89 Author: Richard Biener Date: Fri Sep 12 13:20:46 2025 +0200 Do less redundant vect_transform_slp_perm_load calls The following tries to do vect_transform_slp_pe

[gcc r16-3802] tree-optimization/121595 - new fabs(a+0.0) -> fabs(a) pattern

https://gcc.gnu.org/g:aa4aafbad5235fd302c39e1d8b7cb9cdea11c67c commit r16-3802-gaa4aafbad5235fd302c39e1d8b7cb9cdea11c67c Author: Matteo Nicoli Date: Fri Aug 22 20:42:12 2025 +0200 tree-optimization/121595 - new fabs(a+0.0) -> fabs(a) pattern With -fno-trapping-math it is safe to o

[gcc r16-3797] tree-optimization/121703 - UBSAN error with moving from uninit data

https://gcc.gnu.org/g:0887961f9e0e7ea6b417bfb02787cfdaea41dbe4 commit r16-3797-g0887961f9e0e7ea6b417bfb02787cfdaea41dbe4 Author: Richard Biener Date: Fri Sep 5 14:47:33 2025 +0200 tree-optimization/121703 - UBSAN error with moving from uninit data The PR reports vectorize

[gcc r16-3795] tree-optimization/121829 - bogus CFG with asm goto

https://gcc.gnu.org/g:e9c2a69fe6fe1a20f8ad7b37ecbb3877cd14c514 commit r16-3795-ge9c2a69fe6fe1a20f8ad7b37ecbb3877cd14c514 Author: Richard Biener Date: Mon Sep 8 12:40:30 2025 +0200 tree-optimization/121829 - bogus CFG with asm goto When the vectorizer removes a forwarder created ea

[gcc r16-3666] tree-optimization/121844 - IVOPTs and asm goto in latch

https://gcc.gnu.org/g:00cd34b1046076a3272f8e8e85c97dc8f4d2ea44 commit r16-3666-g00cd34b1046076a3272f8e8e85c97dc8f4d2ea44 Author: Richard Biener Date: Mon Sep 8 14:32:38 2025 +0200 tree-optimization/121844 - IVOPTs and asm goto in latch When there's an asm goto in the latch of a lo

[gcc r16-3662] tree-optimization/121829 - bogus CFG with asm goto

https://gcc.gnu.org/g:a632becefad29206a980cc080eee74ed808f9cd3 commit r16-3662-ga632becefad29206a980cc080eee74ed808f9cd3 Author: Richard Biener Date: Mon Sep 8 12:40:30 2025 +0200 tree-optimization/121829 - bogus CFG with asm goto When the vectorizer removes a forwarder created ea

[gcc r16-3562] tree-optimization/61247 - handle peeled converted IV in SCEV

https://gcc.gnu.org/g:cba74d2cde528af210e37a06652493b2ca38079c commit r16-3562-gcba74d2cde528af210e37a06652493b2ca38079c Author: Richard Biener Date: Fri Aug 29 12:15:09 2025 +0200 tree-optimization/61247 - handle peeled converted IV in SCEV The following handles SCEV analysis of

[gcc r16-3578] TLC to vect_create_epilog_for_reduction

https://gcc.gnu.org/g:bad57f78679e485cc18fd03d4df8e3e15d009544 commit r16-3578-gbad57f78679e485cc18fd03d4df8e3e15d009544 Author: Richard Biener Date: Thu Sep 4 14:54:52 2025 +0200 TLC to vect_create_epilog_for_reduction The following removes back-and-forth of state in vect_cre

[gcc r16-3598] Avoid costing vector stmts with count == 0

https://gcc.gnu.org/g:2965a67c7ef34de74904b600a3e4a4cc858ea37b commit r16-3598-g2965a67c7ef34de74904b600a3e4a4cc858ea37b Author: Richard Biener Date: Fri Sep 5 11:55:50 2025 +0200 Avoid costing vector stmts with count == 0 This avoids confusing the backends. * tre

[gcc r16-3597] tree-optimization/121802 - fix vect_setup_realignment

https://gcc.gnu.org/g:dad6f6eb9b3c96a7fd6c13836a85a4f66373ff0c commit r16-3597-gdad6f6eb9b3c96a7fd6c13836a85a4f66373ff0c Author: Richard Biener Date: Fri Sep 5 10:35:07 2025 +0200 tree-optimization/121802 - fix vect_setup_realignment The following avoids looking at STMT_VINFO_VECT

[gcc r16-3572] TLC for vectorizable_reduction

https://gcc.gnu.org/g:6cc5328cd67472fb7565ee6c3274f227c66dfcf5 commit r16-3572-g6cc5328cd67472fb7565ee6c3274f227c66dfcf5 Author: Richard Biener Date: Thu Sep 4 13:58:56 2025 +0200 TLC for vectorizable_reduction The following removes never taken paths and consolidates the neste

[gcc r16-3569] tree-optimization/121768 - bogus double reduction detected

https://gcc.gnu.org/g:56e7010eef88d941c59e3d677b7bc67a890c210d commit r16-3569-g56e7010eef88d941c59e3d677b7bc67a890c210d Author: Richard Biener Date: Thu Sep 4 09:33:58 2025 +0200 tree-optimization/121768 - bogus double reduction detected The following changes how we detect double

[gcc r16-3568] tree-optimization/121685 - accesses to *this are not trapping

https://gcc.gnu.org/g:e8969a86c1052e2949028f38daf8e273e7ecb106 commit r16-3568-ge8969a86c1052e2949028f38daf8e273e7ecb106 Author: Richard Biener Date: Thu Aug 28 14:27:01 2025 +0200 tree-optimization/121685 - accesses to *this are not trapping When inside a method then we know the

[gcc r16-3561] tree-optimization/121740 - handle aggregate zeroing as skipped may-def

https://gcc.gnu.org/g:95afbe4c7633459b59d434550888decc41b4d44f commit r16-3561-g95afbe4c7633459b59d434550888decc41b4d44f Author: Richard Biener Date: Wed Sep 3 12:54:58 2025 +0200 tree-optimization/121740 - handle aggregate zeroing as skipped may-def The following makes value-numb

[gcc r16-3542] tree-optimization/121756 - handle irreducible regions when sinking

https://gcc.gnu.org/g:df64893e7082d7fae5d6863fd02371b37c78557f commit r16-3542-gdf64893e7082d7fae5d6863fd02371b37c78557f Author: Richard Biener Date: Wed Sep 3 10:41:17 2025 +0200 tree-optimization/121756 - handle irreducible regions when sinking The sinking code currently does no

[gcc r16-3540] Remove vector type setting from vect_recog_cond_expr_convert_pattern

https://gcc.gnu.org/g:dd6fe9f5575ca01842e1809dd765bda7b1e8140e commit r16-3540-gdd6fe9f5575ca01842e1809dd765bda7b1e8140e Author: Richard Biener Date: Thu Jul 24 11:13:12 2025 +0200 Remove vector type setting from vect_recog_cond_expr_convert_pattern This pattern doesn't do any tar

[gcc r16-3539] tree-optimization/121767 - modvar pattern breaking reductions

https://gcc.gnu.org/g:6acd5534ad3ab6208cda68fddb48eb73260d6508 commit r16-3539-g6acd5534ad3ab6208cda68fddb48eb73260d6508 Author: Richard Biener Date: Wed Sep 3 11:04:49 2025 +0200 tree-optimization/121767 - modvar pattern breaking reductions The a % b -> a - a / b pattern breaks r

[gcc r16-3538] tree-optimization/121758 - fix pattern stmt REDUC_IDX updating

https://gcc.gnu.org/g:41e2fe9be1ff5ef2eafd49b30882898d26b9bf36 commit r16-3538-g41e2fe9be1ff5ef2eafd49b30882898d26b9bf36 Author: Richard Biener Date: Wed Sep 3 10:04:58 2025 +0200 tree-optimization/121758 - fix pattern stmt REDUC_IDX updating The following fixes a corner case of p

[gcc r16-3526] tree-optimization/121753 - ICE with pattern breaking reduction constraints

https://gcc.gnu.org/g:edeb5b8258382a58e1ce8e1e79e61386250d42b2 commit r16-3526-gedeb5b8258382a58e1ce8e1e79e61386250d42b2 Author: Richard Biener Date: Tue Sep 2 10:16:28 2025 +0200 tree-optimization/121753 - ICE with pattern breaking reduction constraints The recent change to vect_

[gcc r16-3512] Pass vectype to vect_check_gather_scatter

https://gcc.gnu.org/g:d5b1baa7f2251895447d99cd2f6a1eabb2c94601 commit r16-3512-gd5b1baa7f2251895447d99cd2f6a1eabb2c94601 Author: Richard Biener Date: Mon Sep 1 15:01:24 2025 +0200 Pass vectype to vect_check_gather_scatter The strided-store path needs to have the SLP trees vector t

[gcc r16-3515] Restore STMT_VINFO_VECTYPE during analysis, set to NULL for all stmts

https://gcc.gnu.org/g:3f868fb150957e0510d37d5f39effbddc2f8a4f1 commit r16-3515-g3f868fb150957e0510d37d5f39effbddc2f8a4f1 Author: Richard Biener Date: Mon Sep 1 15:49:59 2025 +0200 Restore STMT_VINFO_VECTYPE during analysis, set to NULL for all stmts The following makes vect_analyz

[gcc r16-3513] Avoid touching STMT_VINFO_VECTYPE in bump_vector_ptr

https://gcc.gnu.org/g:0b6f1de05e2b1e25ac4cf8659bc3c4b5b3bac734 commit r16-3513-g0b6f1de05e2b1e25ac4cf8659bc3c4b5b3bac734 Author: Richard Biener Date: Mon Sep 1 15:45:57 2025 +0200 Avoid touching STMT_VINFO_VECTYPE in bump_vector_ptr bump is always specified, so remove the STMT_VIN

[gcc r16-3514] tree-optimization/121754 - ICE with vect_reduc_type and nested cycle

https://gcc.gnu.org/g:861b7c054e5f356722e6e1a1b043de5d7a412499 commit r16-3514-g861b7c054e5f356722e6e1a1b043de5d7a412499 Author: Richard Biener Date: Tue Sep 2 09:50:14 2025 +0200 tree-optimization/121754 - ICE with vect_reduc_type and nested cycle The reduction guard isn't correc

[gcc r16-3491] tree-optimization/121744 - handle CST << var in shift pattern recog

https://gcc.gnu.org/g:20cd438f2f49ed4816cba9ecc558ee3377b342ba commit r16-3491-g20cd438f2f49ed4816cba9ecc558ee3377b342ba Author: Richard Biener Date: Mon Sep 1 13:29:23 2025 +0200 tree-optimization/121744 - handle CST << var in shift pattern recog We currently do not handle promot

[gcc r16-3489] Remove no longer needed STMT_VINFO_REDUC_DEF sets

https://gcc.gnu.org/g:3b83232c672cd969ec1f7420da99f7a054554958 commit r16-3489-g3b83232c672cd969ec1f7420da99f7a054554958 Author: Richard Biener Date: Wed Aug 27 15:20:03 2025 +0200 Remove no longer needed STMT_VINFO_REDUC_DEF sets The following removes no longer needed extra sets

[gcc r16-3490] Eliminate some STMT_VINFO_REDUC_IDX for SLP_TREE_REDUC_IDX

https://gcc.gnu.org/g:7dff87e0f535406451467991477b79ffa8e45009 commit r16-3490-g7dff87e0f535406451467991477b79ffa8e45009 Author: Richard Biener Date: Wed Aug 27 14:59:34 2025 +0200 Eliminate some STMT_VINFO_REDUC_IDX for SLP_TREE_REDUC_IDX The following uses SLP_TREE_REDUC_IDX whe

[gcc r16-3488] Introduce abstraction for vect reduction info, tracked from SLP nodes

https://gcc.gnu.org/g:b3355804c7054bce288b643ce127ec5e9d0c2061 commit r16-3488-gb3355804c7054bce288b643ce127ec5e9d0c2061 Author: Richard Biener Date: Fri Aug 22 12:29:35 2025 +0200 Introduce abstraction for vect reduction info, tracked from SLP nodes While we have already the acce

[gcc r16-3486] Simplify vectorizer IV analysis

https://gcc.gnu.org/g:5f127979f34d89844efe3489cc4f44b3cb04725d commit r16-3486-g5f127979f34d89844efe3489cc4f44b3cb04725d Author: Richard Biener Date: Fri Aug 29 13:50:32 2025 +0200 Simplify vectorizer IV analysis The following simplifies the flow of IV analysis a bit.

[gcc r16-3440] Dump niter assumption versioning when vectorizing

https://gcc.gnu.org/g:48ef4af2274a0a1ad945ac14217806c1c1d8ccb2 commit r16-3440-g48ef4af2274a0a1ad945ac14217806c1c1d8ccb2 Author: Richard Biener Date: Thu Aug 28 15:15:53 2025 +0200 Dump niter assumption versioning when vectorizing The following emits the assumption that is used fo

[gcc r16-3437] Compute reduction var in vectorize_fold_left_reduction

https://gcc.gnu.org/g:2649c79496a47109b9821d9db7f69aa94e8d3f04 commit r16-3437-g2649c79496a47109b9821d9db7f69aa94e8d3f04 Author: Richard Biener Date: Wed Aug 27 14:01:36 2025 +0200 Compute reduction var in vectorize_fold_left_reduction Instead of going via the PHI node accessible

[gcc r16-3427] The divmod pattern will break reduction constraints

https://gcc.gnu.org/g:674b1d73e437cbdbe86da45dbb67364297bb2216 commit r16-3427-g674b1d73e437cbdbe86da45dbb67364297bb2216 Author: Richard Biener Date: Wed Aug 27 14:39:00 2025 +0200 The divmod pattern will break reduction constraints When we apply a divmod pattern this will break r

[gcc r16-3428] Avoid mult pattern if that will break reduction constraints

https://gcc.gnu.org/g:fa03e49399a8b202b7d6700604c8fbe5ee33474f commit r16-3428-gfa03e49399a8b202b7d6700604c8fbe5ee33474f Author: Richard Biener Date: Wed Aug 27 14:40:37 2025 +0200 Avoid mult pattern if that will break reduction constraints synth-mult introduces multiple uses of a

[gcc r16-3415] Remove dead code

https://gcc.gnu.org/g:5de587c1f7afa2f7a619441a7b8887853662a655 commit r16-3415-g5de587c1f7afa2f7a619441a7b8887853662a655 Author: Richard Biener Date: Wed Aug 27 15:19:44 2025 +0200 Remove dead code The following removes trivially dead code. * tree-vect-loop.cc (ve

[gcc r16-3412] tree-optimization/121686 - failed SLP discovery for live recurrence

https://gcc.gnu.org/g:2027574bacf4fb665a4cb42a33608e169363b39f commit r16-3412-g2027574bacf4fb665a4cb42a33608e169363b39f Author: Richard Biener Date: Wed Aug 27 09:53:49 2025 +0200 tree-optimization/121686 - failed SLP discovery for live recurrence The following adjusts the SLP bu

[gcc r16-3390] Compute vect_reduc_type off SLP node instead of stmt-info

https://gcc.gnu.org/g:670e2b4d3e05f58ce573f719b74dfa49962ed175 commit r16-3390-g670e2b4d3e05f58ce573f719b74dfa49962ed175 Author: Richard Biener Date: Tue Aug 26 09:04:36 2025 +0200 Compute vect_reduc_type off SLP node instead of stmt-info The following changes the vect_reduc_type

[gcc r16-3388] Fix UBSAN issue with load-store data refactoring

https://gcc.gnu.org/g:b01ae7a2b589e015617bed7110b6291c5ecd420c commit r16-3388-gb01ae7a2b589e015617bed7110b6291c5ecd420c Author: Richard Biener Date: Tue Aug 26 10:43:42 2025 +0200 Fix UBSAN issue with load-store data refactoring The following makes sure to read from the lanes_ifn

[gcc r16-3389] tree-optimization/121659 - bogus swap of reduction operands

https://gcc.gnu.org/g:68e692eed9e8e8c47d83586ee08f40c27fa3a78d commit r16-3389-g68e692eed9e8e8c47d83586ee08f40c27fa3a78d Author: Richard Biener Date: Tue Aug 26 10:34:01 2025 +0200 tree-optimization/121659 - bogus swap of reduction operands The following addresses a bogus swapping

[gcc r16-3387] Remove STMT_VINFO_REDUC_VECTYPE_IN

https://gcc.gnu.org/g:0240c52eb6772c3139dfd27fc8fd4ac73c1014b2 commit r16-3387-g0240c52eb6772c3139dfd27fc8fd4ac73c1014b2 Author: Richard Biener Date: Mon Aug 25 14:40:27 2025 +0200 Remove STMT_VINFO_REDUC_VECTYPE_IN This was added when invariants/externals outside of SLP didn't ha

[gcc r16-3367] tree-optimization/121638 - missed SLP discovery of live induction

https://gcc.gnu.org/g:2d4b7b983c7232325a9a7659e37143bdd5112dcc commit r16-3367-g2d4b7b983c7232325a9a7659e37143bdd5112dcc Author: Richard Biener Date: Mon Aug 25 11:02:52 2025 +0200 tree-optimization/121638 - missed SLP discovery of live induction The following fixes a missed SLP d

[gcc r16-3349] Pass representative of live SLP node to vect_create_epilog_for_reduction

https://gcc.gnu.org/g:bd7c238e3c36c872111eb6ffb0e8f75a4a4d2f41 commit r16-3349-gbd7c238e3c36c872111eb6ffb0e8f75a4a4d2f41 Author: Richard Biener Date: Fri Aug 22 14:40:12 2025 +0200 Pass representative of live SLP node to vect_create_epilog_for_reduction We passed the reduc_info wh

[gcc r16-3348] Fixups around reduction info and STMT_VINFO_REDUC_VECTYPE_IN

https://gcc.gnu.org/g:af0af4cf81307240cafdf642e913128ad7ac6b81 commit r16-3348-gaf0af4cf81307240cafdf642e913128ad7ac6b81 Author: Richard Biener Date: Fri Aug 22 14:29:49 2025 +0200 Fixups around reduction info and STMT_VINFO_REDUC_VECTYPE_IN STMT_VINFO_REDUC_VECTYPE_IN exists on r

[gcc r16-3344] Use REDUC_GROUP_FIRST_ELEMENT less

https://gcc.gnu.org/g:60fca68165bb66943ac0f70e28b3d9f2df93c219 commit r16-3344-g60fca68165bb66943ac0f70e28b3d9f2df93c219 Author: Richard Biener Date: Thu Aug 21 13:41:29 2025 +0200 Use REDUC_GROUP_FIRST_ELEMENT less REDUC_GROUP_FIRST_ELEMENT is often checked to see whether we are

[gcc r16-3342] Decouple parloops from vect reduction infra some more

https://gcc.gnu.org/g:3858dcd7a05e7a1143c9ac74a42c30c2fb1dbe4c commit r16-3342-g3858dcd7a05e7a1143c9ac74a42c30c2fb1dbe4c Author: Richard Biener Date: Fri Aug 22 10:39:33 2025 +0200 Decouple parloops from vect reduction infra some more The following removes the use of STMT_VINFO_RE

[gcc r16-3328] tree-optimization/111494 - reduction vectorization with signed UB

https://gcc.gnu.org/g:3182e95eda4a1d612b910b3248c997c8acc7add3 commit r16-3328-g3182e95eda4a1d612b910b3248c997c8acc7add3 Author: Richard Biener Date: Thu Jan 23 14:29:26 2025 +0100 tree-optimization/111494 - reduction vectorization with signed UB The following makes sure to pun ar

[gcc r16-3310] Merge BB and loop path in vect_analyze_stmt

https://gcc.gnu.org/g:bf864b450e8eff4e0fab30da2e6b69ee42d04e6f commit r16-3310-gbf864b450e8eff4e0fab30da2e6b69ee42d04e6f Author: Richard Biener Date: Wed Aug 20 14:07:34 2025 +0200 Merge BB and loop path in vect_analyze_stmt We have now common patterns for most of the vectorizable

[gcc r16-3298] tree-optimization/114480 - speedup IDF compute

https://gcc.gnu.org/g:fc23b539caa16a108bd16bcfcb86fe261a9aa174 commit r16-3298-gfc23b539caa16a108bd16bcfcb86fe261a9aa174 Author: Richard Biener Date: Wed Aug 20 11:06:53 2025 +0200 tree-optimization/114480 - speedup IDF compute The testcase in the PR shows that it's worth splittin

[gcc r16-3296] Thread the remains of vect_analyze_slp_instance

https://gcc.gnu.org/g:c548abddf5e0aa5cd3a9cd936f926dc1e2c7af7b commit r16-3296-gc548abddf5e0aa5cd3a9cd936f926dc1e2c7af7b Author: Richard Biener Date: Tue Aug 19 14:41:30 2025 +0200 Thread the remains of vect_analyze_slp_instance vect_analyze_slp_instance still handles stores and r

[gcc r16-3295] Enable gather/scatter for epilogues of vector epilogues

https://gcc.gnu.org/g:1bf102afed70d88d757d6f7b3014523e66ab42ff commit r16-3295-g1bf102afed70d88d757d6f7b3014523e66ab42ff Author: Richard Biener Date: Wed Jul 30 15:38:03 2025 +0200 Enable gather/scatter for epilogues of vector epilogues The restriction no longer applies, so remove

[gcc r16-3294] Remove most of the epilogue vinfo fixup

https://gcc.gnu.org/g:893d29cf16d5eb62e46e80aa991f9fc6c44be949 commit r16-3294-g893d29cf16d5eb62e46e80aa991f9fc6c44be949 Author: Richard Biener Date: Wed Jul 30 15:14:44 2025 +0200 Remove most of the epilogue vinfo fixup The following removes the fixup we apply to pattern stmt ope

[gcc r16-3293] Record get_load_store_info results from analysis

https://gcc.gnu.org/g:f30aa394e4b14e6d55e349344fe8e0b23e0a7d06 commit r16-3293-gf30aa394e4b14e6d55e349344fe8e0b23e0a7d06 Author: Richard Biener Date: Mon Jul 28 13:48:39 2025 +0200 Record get_load_store_info results from analysis The following is a patch to make us record the get_

[gcc r16-3284] Remove reduction chain detection from parloops

https://gcc.gnu.org/g:f647d4f58a968079c342a0a3f7e71444d3836614 commit r16-3284-gf647d4f58a968079c342a0a3f7e71444d3836614 Author: Richard Biener Date: Tue Aug 19 13:38:58 2025 +0200 Remove reduction chain detection from parloops Historically SLP reduction chains were the only multi

[gcc r16-3283] A few missing SLP node passings to vector costing

https://gcc.gnu.org/g:3bc63918cec91b0e1cd40ed7d4990a93f254678e commit r16-3283-g3bc63918cec91b0e1cd40ed7d4990a93f254678e Author: Richard Biener Date: Tue Aug 19 12:52:39 2025 +0200 A few missing SLP node passings to vector costing The following fixes another few missed cases to pa

[gcc r16-3282] tree-optimization/121592 - failed reduction SLP discovery

https://gcc.gnu.org/g:05284f73cff1685e2513b415e8da644193acf8cd commit r16-3282-g05284f73cff1685e2513b415e8da644193acf8cd Author: Richard Biener Date: Tue Aug 19 12:31:53 2025 +0200 tree-optimization/121592 - failed reduction SLP discovery The testcase in the PR shows that when we

[gcc r16-3281] Fix riscv build, no longer works with python2

https://gcc.gnu.org/g:fc8e2846c24bd62091b31cc35c143b990e79d557 commit r16-3281-gfc8e2846c24bd62091b31cc35c143b990e79d557 Author: Richard Biener Date: Tue Aug 19 10:12:31 2025 +0200 Fix riscv build, no longer works with python2 Building riscv no longer works with python2:

[gcc r16-3280] tree-optimization/121527 - wrong SRA with aggregate copy

https://gcc.gnu.org/g:1d0a0173cd3e48f1c7c7e98893d440527fc198d0 commit r16-3280-g1d0a0173cd3e48f1c7c7e98893d440527fc198d0 Author: Richard Biener Date: Mon Aug 18 13:38:37 2025 +0200 tree-optimization/121527 - wrong SRA with aggregate copy SRA handles outermost VIEW_CONVERT_EXPRs bu

[gcc r16-3182] Introduce SLP_TREE_PERMUTE_P

https://gcc.gnu.org/g:be65c9f13d4bac7289524f97005003f39c7942ea commit r16-3182-gbe65c9f13d4bac7289524f97005003f39c7942ea Author: Richard Biener Date: Wed Aug 13 13:53:41 2025 +0200 Introduce SLP_TREE_PERMUTE_P The following wraps SLP_TREE_CODE checks against VEC_PERM_EXPR (the

[gcc r16-3181] Remove use of STMT_VINFO_DEF_TYPE in vect_analyze_stmt

https://gcc.gnu.org/g:7f6da2623b8739da61d8e57dc175005c5885c6ff commit r16-3181-g7f6da2623b8739da61d8e57dc175005c5885c6ff Author: Richard Biener Date: Wed Aug 13 14:43:28 2025 +0200 Remove use of STMT_VINFO_DEF_TYPE in vect_analyze_stmt This removes a use of STMT_VINFO_DEF_TYPE.

[gcc r16-3180] Fold GATHER_SCATTER_*_P into vect_memory_access_type

https://gcc.gnu.org/g:b8ccad471e51056d442794b9301480de9cd7a19c commit r16-3180-gb8ccad471e51056d442794b9301480de9cd7a19c Author: Richard Biener Date: Tue Aug 12 13:34:30 2025 +0200 Fold GATHER_SCATTER_*_P into vect_memory_access_type The following splits up VMAT_GATHER_SCATTER int

[gcc r16-3179] Simplify vect_supportable_dr_alignment API

https://gcc.gnu.org/g:2c0ce83c9faa1cbea9136a35de14a9f07b4fe6d4 commit r16-3179-g2c0ce83c9faa1cbea9136a35de14a9f07b4fe6d4 Author: Richard Biener Date: Tue Aug 12 15:47:06 2025 +0200 Simplify vect_supportable_dr_alignment API The gather_scatter_info pointer is only used as flag, so

[gcc r16-3164] Cleanup SLP decision during loop analysis

https://gcc.gnu.org/g:6405c5c4419bef9589baf22daf162a6928af5441 commit r16-3164-g6405c5c4419bef9589baf22daf162a6928af5441 Author: Richard Biener Date: Tue Aug 12 11:00:54 2025 +0200 Cleanup SLP decision during loop analysis The following refactors the now misleading slp_done_for_su

[gcc r16-3162] tree-optimization/121509 - failure to detect unvectorizable loop

https://gcc.gnu.org/g:c5dab6fb402c93a92f6aa808c43956dfb9328190 commit r16-3162-gc5dab6fb402c93a92f6aa808c43956dfb9328190 Author: Richard Biener Date: Tue Aug 12 09:51:54 2025 +0200 tree-optimization/121509 - failure to detect unvectorizable loop With the hybrid stmt detection no l

[gcc r16-3161] tree-optimization/121514 - ICE with recent VN improvement

https://gcc.gnu.org/g:a440b382e43203857de9195eb526c4a16f21ceb1 commit r16-3161-ga440b382e43203857de9195eb526c4a16f21ceb1 Author: Richard Biener Date: Tue Aug 12 09:00:48 2025 +0200 tree-optimization/121514 - ICE with recent VN improvement When inserting a compensation stmt during

[gcc r16-3159] Do not set STMT_VINFO_VECTYPE for non-dataref stmts

https://gcc.gnu.org/g:1786be14e94bf1a7806b9dc09186f021737f0227 commit r16-3159-g1786be14e94bf1a7806b9dc09186f021737f0227 Author: Richard Biener Date: Mon Aug 11 11:22:47 2025 +0200 Do not set STMT_VINFO_VECTYPE for non-dataref stmts Now that all STMT_VINFO_VECTYPE uses from vector

[gcc r16-3158] Pass down vector type to avoid STMT_VINFO_VECTYPE on reduc-info

https://gcc.gnu.org/g:5e76090d320a83bf6eeb75c6adb950939aa674e9 commit r16-3158-g5e76090d320a83bf6eeb75c6adb950939aa674e9 Author: Richard Biener Date: Mon Aug 11 15:38:41 2025 +0200 Pass down vector type to avoid STMT_VINFO_VECTYPE on reduc-info The following passes down the vector

[gcc r16-3157] Do not use STMT_VINFO_VECTYPE in vectorizable_reduction

https://gcc.gnu.org/g:bceae48c3c0978fc840169dd4b487c24a0ac5c53 commit r16-3157-gbceae48c3c0978fc840169dd4b487c24a0ac5c53 Author: Richard Biener Date: Mon Aug 11 11:20:41 2025 +0200 Do not use STMT_VINFO_VECTYPE in vectorizable_reduction There's one use of STMT_VINFO_VECTYPE in vec

[gcc r16-3156] tree-optimization/121493 - another missed VN with aggregate copy

https://gcc.gnu.org/g:5294840e3c7bf9bd98016dec07f54ee8dc71714a commit r16-3156-g5294840e3c7bf9bd98016dec07f54ee8dc71714a Author: Richard Biener Date: Mon Aug 11 10:42:47 2025 +0200 tree-optimization/121493 - another missed VN with aggregate copy This is another case where opportun

[gcc r16-3155] Restrict aggregate copy VN generalization

https://gcc.gnu.org/g:6d9c1aad2c46092ac0e5adcaad2ea5af57fad180 commit r16-3155-g6d9c1aad2c46092ac0e5adcaad2ea5af57fad180 Author: Richard Biener Date: Mon Aug 11 14:43:58 2025 +0200 Restrict aggregate copy VN generalization The following avoids ending up with a MEM_REF as component

[gcc r16-3130] tree-optimization/121362 - fixup thinko in earlier change

https://gcc.gnu.org/g:45a56f42b4909e2ae0d1ad35dab05c5c092a14bc commit r16-3130-g45a56f42b4909e2ae0d1ad35dab05c5c092a14bc Author: Richard Biener Date: Mon Aug 11 10:46:12 2025 +0200 tree-optimization/121362 - fixup thinko in earlier change The following fixes the index guards on th

[gcc r16-3131] tree-optimization/121488 - improve BIT_FIELD_REF lookup in VN

https://gcc.gnu.org/g:f2a85db9eee7d761692ca8f69ea1a20d32715bad commit r16-3131-gf2a85db9eee7d761692ca8f69ea1a20d32715bad Author: Richard Biener Date: Mon Aug 11 09:42:09 2025 +0200 tree-optimization/121488 - improve BIT_FIELD_REF lookup in VN When a BIT_FIELD_REF lookup combined w

[gcc r16-3082] Add subsections for vectorization and OpenMP/OpenACC target hooks

https://gcc.gnu.org/g:c9d0953ec88a4a4d2d949042ecb44deaeb65961f commit r16-3082-gc9d0953ec88a4a4d2d949042ecb44deaeb65961f Author: Richard Biener Date: Fri Aug 8 14:09:11 2025 +0200 Add subsections for vectorization and OpenMP/OpenACC target hooks Currently they are grouped under Ad

[gcc r16-3081] Remove setting of STMT_VINFO_VECTYPE on non-dataref stmts

https://gcc.gnu.org/g:8ee642035bf8c3bb62e1bf446761a8aa81b78814 commit r16-3081-g8ee642035bf8c3bb62e1bf446761a8aa81b78814 Author: Richard Biener Date: Thu Jul 10 11:37:14 2025 +0200 Remove setting of STMT_VINFO_VECTYPE on non-dataref stmts The following removes early setting of STM

[gcc r16-3078] tree-optimization/121454 - ICE building libgo

https://gcc.gnu.org/g:43949a5271b6c0b14076b736d0c609235d36f7df commit r16-3078-g43949a5271b6c0b14076b736d0c609235d36f7df Author: Richard Biener Date: Fri Aug 8 08:57:58 2025 +0200 tree-optimization/121454 - ICE building libgo The following avoids building BIT_FIELD_REFs of referen

[gcc r15-10202] Bump BASE-VER

https://gcc.gnu.org/g:47dd30079fd532f7c672870d61d3f2ea8cdb1771 commit r15-10202-g47dd30079fd532f7c672870d61d3f2ea8cdb1771 Author: Richard Biener Date: Fri Aug 8 09:32:56 2025 +0200 Bump BASE-VER * BASE-VER: Set to 15.2.1. Diff: --- gcc/BASE-VER | 2 +- 1 file changed, 1

[gcc r16-3076] Modernize vectorizable_lane_reducing

https://gcc.gnu.org/g:19102eab72ea5bdaa220564033f15a0f354d6254 commit r16-3076-g19102eab72ea5bdaa220564033f15a0f354d6254 Author: Richard Biener Date: Thu Aug 7 14:57:09 2025 +0200 Modernize vectorizable_lane_reducing The following avoids STMT_VINFO_VECTYPE usage in vect_is_emu

[gcc r16-3075] Pass SLP node to promotion/demotion costing

https://gcc.gnu.org/g:f38ebe459af792e097d7923dd49dc19d79d52427 commit r16-3075-gf38ebe459af792e097d7923dd49dc19d79d52427 Author: Richard Biener Date: Thu Aug 7 14:45:52 2025 +0200 Pass SLP node to promotion/demotion costing This one was forgotten. Also constants/externals are cos

[gcc r16-3074] Fix typo in getting SLP_TREE_VECTYPE

https://gcc.gnu.org/g:e4678c4cf843370ce00c46810dcf7f79f8ba7be9 commit r16-3074-ge4678c4cf843370ce00c46810dcf7f79f8ba7be9 Author: Richard Biener Date: Thu Aug 7 14:44:47 2025 +0200 Fix typo in getting SLP_TREE_VECTYPE It was applied on the stmt_info rather than the slp_node.

  1   2   3   4   5   6   7   8   9   10   >