[gcc r16-2217] Fix some auto-profile issues

https://gcc.gnu.org/g:f3186568d09c02a6d8915e43c0f5d7df704dfa0d commit r16-2217-gf3186568d09c02a6d8915e43c0f5d7df704dfa0d Author: Jan Hubicka Date: Sat Jul 12 17:57:25 2025 +0200 Fix some auto-profile issues This patch fixes minor things that has cumulated in my tree. Except for

[gcc r16-2197] Rewrite assign_discriminators

https://gcc.gnu.org/g:385d9937f0e23cbf9c62f0b2553a33ff70e56ecf commit r16-2197-g385d9937f0e23cbf9c62f0b2553a33ff70e56ecf Author: Jan Hubicka Date: Fri Jul 11 13:01:13 2025 +0200 Rewrite assign_discriminators To assign debug locations to corresponding statements auto-fdo uses d

[gcc r16-2196] Fix ICE in speculative devirtualization

https://gcc.gnu.org/g:52d9c2272f6366cd5b30e9540ce8ef16b482cee5 commit r16-2196-g52d9c2272f6366cd5b30e9540ce8ef16b482cee5 Author: Jan Hubicka Date: Fri Jul 11 12:37:24 2025 +0200 Fix ICE in speculative devirtualization This patch fixes ICE bilding lto1 with autoprofiledbootstrap an

[gcc r16-2176] Fixes to auto-profile and Gimple matching.

https://gcc.gnu.org/g:50f3a6a437ad4f2438191b6d9aa9aed8575b9372 commit r16-2176-g50f3a6a437ad4f2438191b6d9aa9aed8575b9372 Author: Jan Hubicka Date: Thu Jul 10 16:56:21 2025 +0200 Fixes to auto-profile and Gimple matching. This patch fixes several issues I noticed in gimple matching

[gcc r16-2150] Fix ICE in afdo_adjust_guessed_profile

https://gcc.gnu.org/g:18324422fdd790b0c11ba300a706a86df1023b74 commit r16-2150-g18324422fdd790b0c11ba300a706a86df1023b74 Author: Jan Hubicka Date: Wed Jul 9 18:30:09 2025 +0200 Fix ICE in afdo_adjust_guessed_profile gcc/ChangeLog: * auto-profile.cc (afdo_adjust_gu

[gcc r16-2126] Improve afdo_adjust_guessed_profile

https://gcc.gnu.org/g:3c0db87b13ed034196d8b77f1acdf40a538d585f commit r16-2126-g3c0db87b13ed034196d8b77f1acdf40a538d585f Author: Jan Hubicka Date: Wed Jul 9 11:51:03 2025 +0200 Improve afdo_adjust_guessed_profile This patch makes afdo_adjust_guessed_profile more robust. Instead o

[gcc r16-2125] Fix profile scaling in tree-inline.cc:initialize_cfun

https://gcc.gnu.org/g:8bd7504cab6fc3289700c1cdb7d03b5e6e9c6c54 commit r16-2125-g8bd7504cab6fc3289700c1cdb7d03b5e6e9c6c54 Author: Jan Hubicka Date: Mon Jul 7 19:20:25 2025 +0200 Fix profile scaling in tree-inline.cc:initialize_cfun initialize_cfun calls profile_count::adjust_f

[gcc r16-2124] Fix auto-profile.cc:get_original_name

https://gcc.gnu.org/g:4de3524f9e88b7b22bdb481163b05a624f090cf9 commit r16-2124-g4de3524f9e88b7b22bdb481163b05a624f090cf9 Author: Jan Hubicka Date: Mon Jul 7 17:18:23 2025 +0200 Fix auto-profile.cc:get_original_name There are two bugs in get_original_name. FIrst the for loop walki

[gcc r16-2048] Print discriminators in dump_scope_block

https://gcc.gnu.org/g:7891c0b450d5ba662fa1817af667b2ba35dee661 commit r16-2048-g7891c0b450d5ba662fa1817af667b2ba35dee661 Author: Jan Hubicka Date: Mon Jul 7 10:07:53 2025 +0200 Print discriminators in dump_scope_block gcc/ChangeLog: * tree-ssa-live.cc (dump_scope_

[gcc r16-2037] Add cutoff information to profile_info and use it when forcing non-zero value

https://gcc.gnu.org/g:5c0758c174c596215857427092e33353f4c1fa72 commit r16-2037-g5c0758c174c596215857427092e33353f4c1fa72 Author: Jan Hubicka Date: Sun Jul 6 14:42:54 2025 +0200 Add cutoff information to profile_info and use it when forcing non-zero value Main difference between n

gcc-cvs@gcc.gnu.org

https://gcc.gnu.org/g:1757c320badc92c0628eafcd07d54585659692ed commit r16-2036-g1757c320badc92c0628eafcd07d54585659692ed Author: Jan Hubicka Date: Sun Jul 6 10:48:46 2025 +0200 Fix overflow check in profile_count::operator* (const sreal &num). gcc/ChangeLog: * pro

[gcc r16-2002] Regenerate common.opt.urls and add period into common.opt

https://gcc.gnu.org/g:70e302d92a6856d1e9bcdca1265203dc3cbd6395 commit r16-2002-g70e302d92a6856d1e9bcdca1265203dc3cbd6395 Author: Jan Hubicka Date: Fri Jul 4 09:14:52 2025 +0200 Regenerate common.opt.urls and add period into common.opt gcc/ChangeLog: * common.opt:

[gcc r16-1970] Add -Wauto-profile warning

https://gcc.gnu.org/g:99f9e90160cb83b09ec7421e9b53e4fffe3ee5ec commit r16-1970-g99f9e90160cb83b09ec7421e9b53e4fffe3ee5ec Author: Jan Hubicka Date: Thu Jul 3 12:05:45 2025 +0200 Add -Wauto-profile warning this patch adds new warning -Wauto-profile which warns about mismatches betw

[gcc r16-1969] Make inliner loop hints more agressive

https://gcc.gnu.org/g:d52b0327a17f5ed12038bd83002627aac5b4b944 commit r16-1969-gd52b0327a17f5ed12038bd83002627aac5b4b944 Author: Jan Hubicka Date: Thu Jul 3 12:00:05 2025 +0200 Make inliner loop hints more agressive This patch makes loop inline hints more agressive. If we know it

[gcc r16-1968] Fix division by zero in ipa-cp.cc:update_profiling_info

https://gcc.gnu.org/g:a52484f1ac34dbb604dc862407d9abb32df444dd commit r16-1968-ga52484f1ac34dbb604dc862407d9abb32df444dd Author: Jan Hubicka Date: Thu Jul 3 11:56:28 2025 +0200 Fix division by zero in ipa-cp.cc:update_profiling_info This ICE has triggered for me during autoprofile

[gcc r16-1963] Enable ipa-cp cloning for cold wrappers of hot functions

https://gcc.gnu.org/g:328ef9aaede3c59224e52a1337416e5489e7c6c8 commit r16-1963-g328ef9aaede3c59224e52a1337416e5489e7c6c8 Author: Jan Hubicka Date: Thu Jul 3 10:25:39 2025 +0200 Enable ipa-cp cloning for cold wrappers of hot functions ipa-cp cloning disables itself for all function

[gcc r16-1962] Fix overlfow in ipa-cp heuristics

https://gcc.gnu.org/g:bba817adbfde5c44fb77cc284c1917d33407ec2e commit r16-1962-gbba817adbfde5c44fb77cc284c1917d33407ec2e Author: Jan Hubicka Date: Thu Jul 3 10:19:31 2025 +0200 Fix overlfow in ipa-cp heuristics ipa-cp converts sreal times to int, while point of sreal is to accomod

[gcc r16-1961] Auto-FDO/FDO profile comparator

https://gcc.gnu.org/g:09db37f7cea79f1cfcede455763e5e2da28ae2d5 commit r16-1961-g09db37f7cea79f1cfcede455763e5e2da28ae2d5 Author: Jan Hubicka Date: Tue Jul 1 08:32:56 2025 +0200 Auto-FDO/FDO profile comparator the patch I sent from airport only worked if you produced the gcda files

[gcc r16-1772] Impove diagnostics of mismatched discriminators in auto-profile

https://gcc.gnu.org/g:483acdc188012cf5b1fc14a82402764c920470aa commit r16-1772-g483acdc188012cf5b1fc14a82402764c920470aa Author: Jan Hubicka Date: Sun Jun 29 07:05:16 2025 +0200 Impove diagnostics of mismatched discriminators in auto-profile We are missing discriminator info in au

[gcc r16-1773] Re-add logic to mitigate some afdo profile inconsistencies

https://gcc.gnu.org/g:f0b3c57ea6b87f51eb0a9bfd7c4438625cb7ed17 commit r16-1773-gf0b3c57ea6b87f51eb0a9bfd7c4438625cb7ed17 Author: Jan Hubicka Date: Mon Jun 30 09:14:46 2025 +0200 Re-add logic to mitigate some afdo profile inconsistencies This patch re-adds logic to increase counts

[gcc r16-1756] Fix handling of dwarf name and duplicated names

https://gcc.gnu.org/g:61f07bcb509cca051e39b698a1bd9a5907f26cf4 commit r16-1756-g61f07bcb509cca051e39b698a1bd9a5907f26cf4 Author: Jan Hubicka Date: Sat Jun 28 05:02:27 2025 +0200 Fix handling of dwarf name and duplicated names I have tested Kugan's patch on exchange2 and noticed mu

[gcc r16-1734] Fix afdo profiles for functions that was not early-inlined

https://gcc.gnu.org/g:ad18f7f88aee15b3f15aa74483ca2ebdc89e18cb commit r16-1734-gad18f7f88aee15b3f15aa74483ca2ebdc89e18cb Author: Jan Hubicka Date: Fri Jun 27 16:10:31 2025 +0200 Fix afdo profiles for functions that was not early-inlined This patch should finish the oflining infras

[gcc r16-1699] Add testcase for afdo offlining and fix two bugs

https://gcc.gnu.org/g:7b28a7dc9dfb277ef1f053dda84899837f8ed0c1 commit r16-1699-g7b28a7dc9dfb277ef1f053dda84899837f8ed0c1 Author: Jan Hubicka Date: Thu Jun 26 10:48:20 2025 +0200 Add testcase for afdo offlining and fix two bugs This patch adds a testcase that offlining works and pr

[gcc r16-1695] Avoid some lost AFDO profiles with LTO

https://gcc.gnu.org/g:884030d2de767c491223c9cfcf2dd14fdfa93df4 commit r16-1695-g884030d2de767c491223c9cfcf2dd14fdfa93df4 Author: Jan Hubicka Date: Thu Jun 26 09:06:52 2025 +0200 Avoid some lost AFDO profiles with LTO This patch fixes some of cases where we lose profile info becaus

[gcc r16-1663] Remove early inlining from afdo pass

https://gcc.gnu.org/g:3fde750a29820a1ccd6dd399bdfa0cf3d97a2c30 commit r16-1663-g3fde750a29820a1ccd6dd399bdfa0cf3d97a2c30 Author: Jan Hubicka Date: Wed Jun 25 02:59:54 2025 +0200 Remove early inlining from afdo pass This pass removes early-inlining from afdo pass since all inlining

[gcc r16-1664] Add -fauto-profile-inlining

https://gcc.gnu.org/g:aaf55e09b3d97164615e783d98cfa842f382559d commit r16-1664-gaaf55e09b3d97164615e783d98cfa842f382559d Author: Jan Hubicka Date: Wed Jun 25 03:01:29 2025 +0200 Add -fauto-profile-inlining this patch adds -fauto-profile-inlining which can be used to control th

[gcc r16-1642] Copy discriminators when inlining

https://gcc.gnu.org/g:0235b6d41ace62064d5cd42553028136b49ad947 commit r16-1642-g0235b6d41ace62064d5cd42553028136b49ad947 Author: Jan Hubicka Date: Tue Jun 24 05:07:42 2025 +0200 Copy discriminators when inlining When inlining disciriminator info about the call statement is lost wh

[gcc r16-1641] Fix AFDO zero profile handling

https://gcc.gnu.org/g:c24eb5e01da5ce07f6b616aff1129d4acbff69e6 commit r16-1641-gc24eb5e01da5ce07f6b616aff1129d4acbff69e6 Author: Jan Hubicka Date: Tue Jun 24 05:00:01 2025 +0200 Fix AFDO zero profile handling This patch fixes roms autofdo regression I introduced yesterday. What

[gcc r16-1610] Fix some problems with afdo propagation

https://gcc.gnu.org/g:acd5f587ca6a5deaebab19f6b0935321853c9f6f commit r16-1610-gacd5f587ca6a5deaebab19f6b0935321853c9f6f Author: Jan Hubicka Date: Sun Jun 22 11:06:12 2025 +0200 Fix some problems with afdo propagation This patch fixes problems I noticed by exploring profiles of so

[gcc r16-1609] Handle functions with 0 profile in auto-profile

https://gcc.gnu.org/g:ed02d762d51d9544aeca52a8b306c2da2f36bd7f commit r16-1609-ged02d762d51d9544aeca52a8b306c2da2f36bd7f Author: Jan Hubicka Date: Sun Jun 22 06:55:41 2025 +0200 Handle functions with 0 profile in auto-profile This is the last part of the infrastructure to allow fu

[gcc r16-1607] Prevent possible overflows in ipa-profile

https://gcc.gnu.org/g:6e38bef16bbfaa7743d1ec8937ed9dfba669136d commit r16-1607-g6e38bef16bbfaa7743d1ec8937ed9dfba669136d Author: Jan Hubicka Date: Sun Jun 22 03:32:29 2025 +0200 Prevent possible overflows in ipa-profile The bug in scaling profile of fnsplit produced clones made

[gcc r16-1605] Add GUESSED_GLOBAL0_AFDO

https://gcc.gnu.org/g:eb8ee105706569c9a03f3de9519f6ab8006c3f1e commit r16-1605-geb8ee105706569c9a03f3de9519f6ab8006c3f1e Author: Jan Hubicka Date: Sun Jun 22 03:12:55 2025 +0200 Add GUESSED_GLOBAL0_AFDO This patch adds GUESSED_GLOBAL0_AFDO profile quality. It can be used to pr

[gcc r16-1606] Scale up auto-profile counts

https://gcc.gnu.org/g:dda86c80bca2300a47f91bcfc589951df9c7f1be commit r16-1606-gdda86c80bca2300a47f91bcfc589951df9c7f1be Author: Jan Hubicka Date: Sun Jun 22 03:26:36 2025 +0200 Scale up auto-profile counts This patch makes auto-profile counts to scale up when the train run has

[gcc r16-1603] Fix profile after fnsplit

https://gcc.gnu.org/g:cd589516b12e28ee30aefc4c51500f634f1b888e commit r16-1603-gcd589516b12e28ee30aefc4c51500f634f1b888e Author: Jan Hubicka Date: Sat Jun 21 22:29:50 2025 +0200 Fix profile after fnsplit when splitting functions, tree-inline determined correctly entry count of th

[gcc r16-1599] Extend afdo inliner to introduce speculative calls

https://gcc.gnu.org/g:d29cf57f9e4e9e16285a627a1717269ef7cf131b commit r16-1599-gd29cf57f9e4e9e16285a627a1717269ef7cf131b Author: Jan Hubicka Date: Sat Jun 21 05:37:24 2025 +0200 Extend afdo inliner to introduce speculative calls This patch makes the AFDO's VPT to happen during ear

[gcc r16-1598] Implement afdo inliner

https://gcc.gnu.org/g:8f40a8e8f8d1ebe931d52f914533036c2f950814 commit r16-1598-g8f40a8e8f8d1ebe931d52f914533036c2f950814 Author: Jan Hubicka Date: Wed Jun 18 12:10:25 2025 +0200 Implement afdo inliner This patch moves afdo inlining from early inliner into specialized one. The

[gcc r16-1545] Improve static and AFDO profile combination

https://gcc.gnu.org/g:260252e7dc07bd6e201c76c24d858efaea4a1a78 commit r16-1545-g260252e7dc07bd6e201c76c24d858efaea4a1a78 Author: Jan Hubicka Date: Tue Jun 17 17:26:18 2025 +0200 Improve static and AFDO profile combination This patch makes afdo_adjust_guessed_profile more agressive

[gcc r16-1544] Fix cgraph_node::apply_scale

https://gcc.gnu.org/g:30e8f97a25f44b1e11275a9f46aef78d7e4e4316 commit r16-1544-g30e8f97a25f44b1e11275a9f46aef78d7e4e4316 Author: Jan Hubicka Date: Tue Jun 17 17:20:04 2025 +0200 Fix cgraph_node::apply_scale while working on auto-FDO I noticed that we may run into ICE because we i

[gcc r16-1543] Add testcase for AFDO early inlining and indirect call promotion

https://gcc.gnu.org/g:3d1d893a00d6a46e8f090ad29715693b6c04a78f commit r16-1543-g3d1d893a00d6a46e8f090ad29715693b6c04a78f Author: Jan Hubicka Date: Tue Jun 17 15:43:13 2025 +0200 Add testcase for AFDO early inlining and indirect call promotion gcc/testsuite/ChangeLog:

[gcc r16-1521] Combine static and afdo branch predictions

https://gcc.gnu.org/g:2ef043c5a05d9914e3c3dbff6f2c521eb665d971 commit r16-1521-g2ef043c5a05d9914e3c3dbff6f2c521eb665d971 Author: Jan Hubicka Date: Mon Jun 16 10:19:05 2025 +0200 Combine static and afdo branch predictions Currently afdo reads the profile and anotates basic blocks c

[gcc r16-1401] More API for IPA profile manipulation

https://gcc.gnu.org/g:e416c8097fc87513e05c2d104c63488f733758c0 commit r16-1401-ge416c8097fc87513e05c2d104c63488f733758c0 Author: Jan Hubicka Date: Tue Jun 10 21:32:40 2025 +0200 More API for IPA profile manipulation This patch attempts to make IPA profile manipulation easier. It

[gcc r16-1346] Fix profile updating in ipa-cp

https://gcc.gnu.org/g:b0d50cbb42ab2ce5fab8a832cb82fc54b371c914 commit r16-1346-gb0d50cbb42ab2ce5fab8a832cb82fc54b371c914 Author: Jan Hubicka Date: Fri Jun 6 17:57:00 2025 +0200 Fix profile updating in ipa-cp Bootstrapping with autoprofiledbootstrap, LTO and checking enables ICEs i

[gcc r16-1246] More of autofdo 0 issues

https://gcc.gnu.org/g:22f1f765db056527076a3b7e785b068f3ea28cf0 commit r16-1246-g22f1f765db056527076a3b7e785b068f3ea28cf0 Author: Jan Hubicka Date: Fri Jun 6 10:43:38 2025 +0200 More of autofdo 0 issues This patch fixes ICE seen when building spec2k17 with autofdo and enable ch

[gcc r16-1249] Avoid useless reading of profile data in LTO

https://gcc.gnu.org/g:572dd0dd8db4bec69883c0b3e53013a7842e886e commit r16-1249-g572dd0dd8db4bec69883c0b3e53013a7842e886e Author: Jan Hubicka Date: Fri Jun 6 10:56:51 2025 +0200 Avoid useless reading of profile data in LTO New auto-profile merging dumps made me notice that we read

[gcc r16-1190] Fix typo in afdo_calculate_branch_prob

https://gcc.gnu.org/g:6fb88ff622ca2d12229870e9b0b3781bb930d2ad commit r16-1190-g6fb88ff622ca2d12229870e9b0b3781bb930d2ad Author: Jan Hubicka Date: Thu Jun 5 18:10:16 2025 +0200 Fix typo in afdo_calculate_branch_prob gcc/ChangeLog: * auto-profile.cc (afdo_calculate

[gcc r16-1183] Fix handling of GUESSED_LOCAL in auto-fdo and preserve more static profile

https://gcc.gnu.org/g:9399a3b5f512af6dbbe05be076e726efc91413f1 commit r16-1183-g9399a3b5f512af6dbbe05be076e726efc91413f1 Author: Jan Hubicka Date: Thu Jun 5 15:24:36 2025 +0200 Fix handling of GUESSED_LOCAL in auto-fdo and preserve more static profile This patch fixes ICE where GU

[gcc r16-953] i386: Use Shuffles instead of shifts for Reduction in AMD znver4/5

https://gcc.gnu.org/g:5080d98a383de244a7b78ae50456fd41881268c2 commit r16-953-g5080d98a383de244a7b78ae50456fd41881268c2 Author: Pranav Gorantla Date: Thu May 29 15:02:24 2025 +0200 i386: Use Shuffles instead of shifts for Reduction in AMD znver4/5 In AMD znver4, znver5 targets vps

[gcc r16-941] Set znver5 addss cost to 2 again

https://gcc.gnu.org/g:8859a06fc8143561dd94ef2342234dbc5ccfd937 commit r16-941-g8859a06fc8143561dd94ef2342234dbc5ccfd937 Author: Jan Hubicka Date: Wed May 28 23:43:51 2025 +0200 Set znver5 addss cost to 2 again since uses of addss for other purposes then modelling FP addition/subt

[gcc r16-930] Handle auto-fdo 0 more carefully

https://gcc.gnu.org/g:17f7b6250628c31182fd4f71c9ecdeca9568ffd1 commit r16-930-g17f7b6250628c31182fd4f71c9ecdeca9568ffd1 Author: Jan Hubicka Date: Wed May 28 14:26:11 2025 +0200 Handle auto-fdo 0 more carefully This patch fixes few other places where auto-fdo 0 should be be treated

[gcc r16-929] Do not recompute profile when entry block has afdo count of 0

https://gcc.gnu.org/g:a4dc4001999a8b504cde468618af12c9b870589b commit r16-929-ga4dc4001999a8b504cde468618af12c9b870589b Author: Jan Hubicka Date: Wed May 28 14:18:39 2025 +0200 Do not recompute profile when entry block has afdo count of 0 With normal profile feedback checking entr

[gcc r16-927] Fix profile_probability quality of switch

https://gcc.gnu.org/g:8058e1f8857d580698c29590273e72c61de2c121 commit r16-927-g8058e1f8857d580698c29590273e72c61de2c121 Author: Jan Hubicka Date: Wed May 28 12:23:48 2025 +0200 Fix profile_probability quality of switch This fixes ages old bug I noticed only now where switch cases,

[gcc r16-926] Do not erase static profile by 0 autofdo profile

https://gcc.gnu.org/g:aa1b47aa41af2a6a094035bd446504cf352c4d71 commit r16-926-gaa1b47aa41af2a6a094035bd446504cf352c4d71 Author: Jan Hubicka Date: Wed May 28 12:15:32 2025 +0200 Do not erase static profile by 0 autofdo profile This patch makes auto-fdo more careful about keeping in

[gcc r16-912] Enable afdo testing on AMD Zen3+

https://gcc.gnu.org/g:7b76965df10c47616c8f65e1c1134356d635f9ed commit r16-912-g7b76965df10c47616c8f65e1c1134356d635f9ed Author: Jan Hubicka Date: Tue May 27 19:53:50 2025 +0200 Enable afdo testing on AMD Zen3+ contrib/ChangeLog: * gen_autofdo_event.py: Add support

[gcc r16-911] Remove dead code in auto-profile.cc

https://gcc.gnu.org/g:fbe6dea1a1b8cad114aeec07c6f8966300417cd9 commit r16-911-gfbe6dea1a1b8cad114aeec07c6f8966300417cd9 Author: Jan Hubicka Date: Tue May 27 19:14:21 2025 +0200 Remove dead code in auto-profile.cc This code to track what locations were used when reading auto-fdo pr

[gcc r16-869] Make i386 construcotr vectorizer costs more realistics

https://gcc.gnu.org/g:e3d3d6d7d2c8ab73ff597f4c82514c3217256567 commit r16-869-ge3d3d6d7d2c8ab73ff597f4c82514c3217256567 Author: Jan Hubicka Date: Sun May 25 14:33:17 2025 +0200 Make i386 construcotr vectorizer costs more realistics this patch attempts to make vectorizer costs of v

[gcc r16-531] i386: Fix move costs in vectorizer cost model.

https://gcc.gnu.org/g:37e61c793c1b22bdcfbf142cd6086da2745be596 commit r16-531-g37e61c793c1b22bdcfbf142cd6086da2745be596 Author: Jan Hubicka Date: Sun May 11 23:49:11 2025 +0200 i386: Fix move costs in vectorizer cost model. This patch complements the change to stv and uses COSTS_N

[gcc r16-517] i386: Fix some problems in stv cost model

https://gcc.gnu.org/g:993aa0bd28722c7f01fb8310f1c79814aef217ed commit r16-517-g993aa0bd28722c7f01fb8310f1c79814aef217ed Author: Jan Hubicka Date: Sat May 10 22:23:48 2025 +0200 i386: Fix some problems in stv cost model this patch fixes some of problems with cosint in scalar to vec

[gcc r16-447] i386: implement costs for float<->int conversions in ix86_vector_costs::add_stmt_cost

https://gcc.gnu.org/g:2c8d632d9ed4e3aeee2156ba17fe631ecbc90dbf commit r16-447-g2c8d632d9ed4e3aeee2156ba17fe631ecbc90dbf Author: Jan Hubicka Date: Wed May 7 15:33:44 2025 +0200 i386: implement costs for float<->int conversions in ix86_vector_costs::add_stmt_cost This patch adds pa

[gcc r16-429] i386: Add costs for integer<->float conversions

https://gcc.gnu.org/g:62ffaef514aad0eae5d3bc548d8e86d21a3095e7 commit r16-429-g62ffaef514aad0eae5d3bc548d8e86d21a3095e7 Author: Jan Hubicka Date: Wed May 7 00:24:04 2025 +0200 i386: Add costs for integer<->float conversions Extend ix86_rtx_costs to cost FLOAT, UNSIGNED_FLOAT, FIX,

[gcc r16-401] Fix i386 bootstrap on non-Windows platforms

https://gcc.gnu.org/g:673d446894c063c92cafce9ba41340c82e960a11 commit r16-401-g673d446894c063c92cafce9ba41340c82e960a11 Author: Jan Hubicka Date: Tue May 6 12:07:15 2025 +0200 Fix i386 bootstrap on non-Windows platforms * config/i386/i386.cc (ix86_tls_index): Add ifdef. D

[gcc r16-372] Improve maybe_hot handling in inliner heuristics

https://gcc.gnu.org/g:064cac730f88dc71c6da578f9ae5b8e092ab6cd4 commit r16-372-g064cac730f88dc71c6da578f9ae5b8e092ab6cd4 Author: Jan Hubicka Date: Sun May 4 10:52:35 2025 +0200 Improve maybe_hot handling in inliner heuristics Inliner currently applies different heuristics to hot an

[gcc r16-367] Improve ix86 VEC_MERGE costs

https://gcc.gnu.org/g:20d184e3f84d859e7e9f44a8d91772a02b658872 commit r16-367-g20d184e3f84d859e7e9f44a8d91772a02b658872 Author: Jan Hubicka Date: Sat May 3 00:26:29 2025 +0200 Improve ix86 VEC_MERGE costs ix86_rtx_costs VEC_MERGE by special casing AVX512 mask operations and other

[gcc r16-358] Make ix86 cost of VEC_SELECT equivalent to SUBREG cost 1

https://gcc.gnu.org/g:c85148d036d17295bb2560e10020c924c83a5d13 commit r16-358-gc85148d036d17295bb2560e10020c924c83a5d13 Author: Jan Hubicka Date: Fri May 2 15:53:35 2025 +0200 Make ix86 cost of VEC_SELECT equivalent to SUBREG cost 1 This patch fixes regression of imagick with PGO

[gcc r16-291] Fix cs_interesting_for_ipcp_p wrt flag_profile_partial_training.

https://gcc.gnu.org/g:1efd525e828d5c30d0962d05ba81f9d7c8176ca7 commit r16-291-g1efd525e828d5c30d0962d05ba81f9d7c8176ca7 Author: Jan Hubicka Date: Tue Apr 29 22:43:45 2025 +0200 Fix cs_interesting_for_ipcp_p wrt flag_profile_partial_training. As noticed by Martin Jambor, I introduc

[gcc r16-162] Fix i386 vectorizer cost of FP scalar MAX_EXPR and MIN_EXPR

https://gcc.gnu.org/g:1d635e79b3c2d26f864964b79717132bffbcad20 commit r16-162-g1d635e79b3c2d26f864964b79717132bffbcad20 Author: Jan Hubicka Date: Sat Apr 26 22:10:19 2025 +0200 Fix i386 vectorizer cost of FP scalar MAX_EXPR and MIN_EXPR I introduced a bug by last minute cleanups u

[gcc r16-117] Fix i386 vectorizer cost of COND_EXPR and MIN_MAX with one of parameters 0 or -1

https://gcc.gnu.org/g:2056d52d74070f50c5f8a22e4a600fcc3974fd88 commit r16-117-g2056d52d74070f50c5f8a22e4a600fcc3974fd88 Author: Jan Hubicka Date: Thu Apr 24 18:37:55 2025 +0200 Fix i386 vectorizer cost of COND_EXPR and MIN_MAX with one of parameters 0 or -1 gcc/ChangeLog:

[gcc r16-116] Fix ICE building deepsjeng with -fprofile-use

https://gcc.gnu.org/g:cfb04e0de6aa438df9d8b83a3d8c7f93789b5c9f commit r16-116-gcfb04e0de6aa438df9d8b83a3d8c7f93789b5c9f Author: Jan Hubicka Date: Thu Apr 24 18:35:54 2025 +0200 Fix ICE building deepsjeng with -fprofile-use The problem here is division by zero, since adjusted 0 > p

[gcc r16-101] Enable ip-cp cloning over non-hot edges

https://gcc.gnu.org/g:132d01d96ea9d617aaffdd5dfba3284a8958e529 commit r16-101-g132d01d96ea9d617aaffdd5dfba3284a8958e529 Author: Jan Hubicka Date: Wed Apr 23 18:39:14 2025 +0200 Enable ip-cp cloning over non-hot edges Currently enabling profile feedback regresses x264 and exchange.

[gcc r16-100] Cost truth_value exprs in i386 vectorizer costs.

https://gcc.gnu.org/g:9b9d605d68cf27a24e8ed9d4f1ead1f00131cec1 commit r16-100-g9b9d605d68cf27a24e8ed9d4f1ead1f00131cec1 Author: Jan Hubicka Date: Wed Apr 23 17:04:32 2025 +0200 Cost truth_value exprs in i386 vectorizer costs. this patch implements costing of truth_value exprs. I.

[gcc r16-89] Fix vectorizer costs of COND_EXPR, MIN_EXPR, MAX_EXPR, ABS_EXPR, ABSU_EXPR

https://gcc.gnu.org/g:0650ea627399a0ef23db434d4fce6b52b9faf557 commit r16-89-g0650ea627399a0ef23db434d4fce6b52b9faf557 Author: Jan Hubicka Date: Tue Apr 22 23:47:14 2025 +0200 Fix vectorizer costs of COND_EXPR, MIN_EXPR, MAX_EXPR, ABS_EXPR, ABSU_EXPR this patch adds special cases

[gcc r16-54] Fix cost of vectorized double->float conversion

https://gcc.gnu.org/g:0907a810f586b07636cc5b83dba6025eb5240655 commit r16-54-g0907a810f586b07636cc5b83dba6025eb5240655 Author: Jan Hubicka Date: Mon Apr 21 20:16:50 2025 +0200 Fix cost of vectorized double->float conversion In previous patch I miscomputed costs of cvtpd2pf instruc

[gcc r16-39] Add tables for SSE fp conversion costs

https://gcc.gnu.org/g:f6859fb621179ec9bf5631eb8902619ab8d4467b commit r16-39-gf6859fb621179ec9bf5631eb8902619ab8d4467b Author: Jan Hubicka Date: Sat Apr 19 18:51:27 2025 +0200 Add tables for SSE fp conversion costs as disucssed, I will proceed adding costs for common SSE operation

[gcc r15-9522] Stream ipa_return_value_summary

https://gcc.gnu.org/g:eabba7be040e81690332070873d59d23e8c93e11 commit r15-9522-geabba7be040e81690332070873d59d23e8c93e11 Author: Jan Hubicka Date: Wed Apr 16 15:28:32 2025 +0200 Stream ipa_return_value_summary Add streaming of return summaries from compile time to ltrans which

[gcc r15-9496] Set znver5 issue rate to 4.

https://gcc.gnu.org/g:4a01869b963520d689fe9242cf2ff24984ea41d8 commit r15-9496-g4a01869b963520d689fe9242cf2ff24984ea41d8 Author: Jan Hubicka Date: Tue Apr 15 19:09:20 2025 +0200 Set znver5 issue rate to 4. this patch sets issue rate of znver5 to 4. With current model, unless a r

[gcc r15-9495] Set ADDSS cost to 3 for znver5

https://gcc.gnu.org/g:e2011ab13de3e70774f869b356f5f9c750780b34 commit r15-9495-ge2011ab13de3e70774f869b356f5f9c750780b34 Author: Jan Hubicka Date: Tue Apr 15 19:04:15 2025 +0200 Set ADDSS cost to 3 for znver5 Znver5 has latency of addss 2 in typical case while all earlier versions

[gcc r15-9176] Fix costs of x86 move instructions at -Os

https://gcc.gnu.org/g:564e4e0819022925dd160e455ee44baf0fda5805 commit r15-9176-g564e4e0819022925dd160e455ee44baf0fda5805 Author: Jan Hubicka Date: Thu Apr 3 13:06:07 2025 +0200 Fix costs of x86 move instructions at -Os This patch fixes problem with size costs declaring all moves t

[gcc r15-9047] Optimize string constructor

https://gcc.gnu.org/g:9c5505a35d9d71705464f9254f55407192d31ec3 commit r15-9047-g9c5505a35d9d71705464f9254f55407192d31ec3 Author: Jan Hubicka Date: Sun Mar 30 23:49:49 2025 +0200 Optimize string constructor this patch improves code generation on string constructors. We currently

[gcc r15-8041] Fix speculation_useful_p

https://gcc.gnu.org/g:57dbbdd8e34b80926e06b352b6c442c555b303ed commit r15-8041-g57dbbdd8e34b80926e06b352b6c442c555b303ed Author: Jan Hubicka Date: Thu Mar 13 20:11:02 2025 +0100 Fix speculation_useful_p This patch fixes issue with speculation and x264. With profile feedback w

[gcc r15-7813] Break false dependency chain on Zen5

https://gcc.gnu.org/g:8c4a00f9a48f1b2af10448c9f2058b44b8cb7234 commit r15-7813-g8c4a00f9a48f1b2af10448c9f2058b44b8cb7234 Author: Jan Hubicka Date: Tue Mar 4 16:22:01 2025 +0100 Break false dependency chain on Zen5 Zen5 on some variants has false dependency on tzcnt, blsi, blsr and

[gcc r15-7811] Make ix86_macro_fusion_pair_p and ix86_fuse_mov_alu_p match current CPUs

https://gcc.gnu.org/g:c84be624e079cd748df93a3dc0b5168865fefee9 commit r15-7811-gc84be624e079cd748df93a3dc0b5168865fefee9 Author: Jan Hubicka Date: Mon Mar 3 19:12:20 2025 +0100 Make ix86_macro_fusion_pair_p and ix86_fuse_mov_alu_p match current CPUs The current implementation of f

[gcc r12-10923] Zen5 tuning part 2: disable gather and scatter

https://gcc.gnu.org/g:e909afe8a8a2924dd6ced6bdf7d8e397f14310b5 commit r12-10923-ge909afe8a8a2924dd6ced6bdf7d8e397f14310b5 Author: Jan Hubicka Date: Tue Sep 3 15:07:41 2024 +0200 Zen5 tuning part 2: disable gather and scatter We disable gathers for zen4. It seems that gather has i

[gcc r12-10901] Zen5 tuning part 5: update instruction latencies in x86-tune-costs

https://gcc.gnu.org/g:3911b6366ee49dffe2f16578093b49664b3a2d72 commit r12-10901-g3911b6366ee49dffe2f16578093b49664b3a2d72 Author: Jan Hubicka Date: Wed Sep 4 09:19:08 2024 +0200 Zen5 tuning part 5: update instruction latencies in x86-tune-costs there is nothing exciting in this pa

[gcc r13-9309] Zen5 tuning part 5: update instruction latencies in x86-tune-costs

https://gcc.gnu.org/g:f10d381dfc983ea32e5f72faadc7eb8126f114f6 commit r13-9309-gf10d381dfc983ea32e5f72faadc7eb8126f114f6 Author: Jan Hubicka Date: Wed Sep 4 09:19:08 2024 +0200 Zen5 tuning part 5: update instruction latencies in x86-tune-costs there is nothing exciting in this pat

[gcc r12-10888] Zen5 tuning part 4: update reassocation width

https://gcc.gnu.org/g:4d7efc031fbd925565b049670bf755aca21bd2e3 commit r12-10888-g4d7efc031fbd925565b049670bf755aca21bd2e3 Author: Jan Hubicka Date: Tue Sep 3 18:20:34 2024 +0200 Zen5 tuning part 4: update reassocation width Zen5 has 6 instead of 4 ALUs and the integer multiplicati

[gcc r13-9285] Zen5 tuning part 4: update reassocation width

https://gcc.gnu.org/g:52507e15aa31bc66e99f2273306f1b45be919bba commit r13-9285-g52507e15aa31bc66e99f2273306f1b45be919bba Author: Jan Hubicka Date: Tue Sep 3 18:20:34 2024 +0200 Zen5 tuning part 4: update reassocation width Zen5 has 6 instead of 4 ALUs and the integer multiplicatio

[gcc r12-10887] Zen5 tuning part 3: scheduler tweaks

https://gcc.gnu.org/g:f0718f1d7815c7845243a182c66f4a454efbfb72 commit r12-10887-gf0718f1d7815c7845243a182c66f4a454efbfb72 Author: Jan Hubicka Date: Tue Sep 3 16:26:16 2024 +0200 Zen5 tuning part 3: scheduler tweaks this patch adds support for new fussion in znver5 documented in th

[gcc r13-9284] Zen5 tuning part 3: scheduler tweaks

https://gcc.gnu.org/g:7392e9e480afe3143e72a99f7b5ac99b2f49c284 commit r13-9284-g7392e9e480afe3143e72a99f7b5ac99b2f49c284 Author: Jan Hubicka Date: Tue Sep 3 16:26:16 2024 +0200 Zen5 tuning part 3: scheduler tweaks this patch adds support for new fussion in znver5 documented in the

[gcc r15-5694] improve std::deque::_M_reallocate_map

https://gcc.gnu.org/g:1729147f2b1a05ad2a4e8e2d91d1d2ba914039d6 commit r15-5694-g1729147f2b1a05ad2a4e8e2d91d1d2ba914039d6 Author: Jan Hubicka Date: Tue Nov 26 13:52:09 2024 +0100 improve std::deque::_M_reallocate_map Looking into reason why we still do throw_bad_alloc in clang bina

[gcc r15-5365] Mark asm statements as necessary in ipa-fnsummary

https://gcc.gnu.org/g:bd59f2eeacd41b91e4e79b32dda83cc60d499e25 commit r15-5365-gbd59f2eeacd41b91e4e79b32dda83cc60d499e25 Author: Jan Hubicka Date: Sun Nov 17 15:48:29 2024 +0100 Mark asm statements as necessary in ipa-fnsummary I forgot to mark asm statements as necessary in ipa-f

[gcc r15-5362] ipa-modref bits for unsequenced and reproducible

https://gcc.gnu.org/g:addf02282026cc23d24fc4445b47b408d484d1e7 commit r15-5362-gaddf02282026cc23d24fc4445b47b408d484d1e7 Author: Jan Hubicka Date: Sun Nov 17 11:54:10 2024 +0100 ipa-modref bits for unsequenced and reproducible C attributes reproducible and unsequenced implies that

[gcc r15-5359] Avoid expicit builtion list in tree-ssa-dce

https://gcc.gnu.org/g:cc33f880e553d1aa94d19a349ad755f34c33de9e commit r15-5359-gcc33f880e553d1aa94d19a349ad755f34c33de9e Author: Jan Hubicka Date: Sat Nov 16 23:45:57 2024 +0100 Avoid expicit builtion list in tree-ssa-dce while working on -fmalloc-dce I noticed that tree-ssa-dce.c

[gcc r15-5358] Minor cleanup to cxx_init_decl_processing

https://gcc.gnu.org/g:101f8c73d575b4746e49b0ea35eb6cc24de1dfdc commit r15-5358-g101f8c73d575b4746e49b0ea35eb6cc24de1dfdc Author: Jan Hubicka Date: Sat Nov 16 23:44:13 2024 +0100 Minor cleanup to cxx_init_decl_processing gcc/cp/ChangeLog: * decl.cc (cxx_build_opera

[gcc r15-5336] Ignore conditions guarding __builtin_unreachable in inliner metrics

https://gcc.gnu.org/g:cee7d080d5c2a5fb8125878998b742c040ec88b4 commit r15-5336-gcee7d080d5c2a5fb8125878998b742c040ec88b4 Author: Jan Hubicka Date: Sat Nov 16 14:04:32 2024 +0100 Ignore conditions guarding __builtin_unreachable in inliner metrics This extends my last year attempt t

[gcc r15-5322] Fix type of malloc parameter in trans-expr.cc

https://gcc.gnu.org/g:d5af5657fa4173dd8d3155e44fe6a1e3914cde8f commit r15-5322-gd5af5657fa4173dd8d3155e44fe6a1e3914cde8f Author: Jan Hubicka Date: Fri Nov 15 15:51:14 2024 +0100 Fix type of malloc parameter in trans-expr.cc gcc/fortran/ChangeLog: * trans-expr.cc (

[gcc r15-5281] Fix common.opt.urls

https://gcc.gnu.org/g:236104908257d779b06894379f99e007a1c4c897 commit r15-5281-g236104908257d779b06894379f99e007a1c4c897 Author: Jan Hubicka Date: Thu Nov 14 17:29:14 2024 +0100 Fix common.opt.urls gcc/ChangeLog: * common.opt.urls: Fix. Diff: --- gcc/common.opt.

[gcc r15-5256] New testcase for operator new/delete removal.

https://gcc.gnu.org/g:269b6477744f25ab0a89b364295b86a4bce392f9 commit r15-5256-g269b6477744f25ab0a89b364295b86a4bce392f9 Author: Jan Hubicka Date: Thu Nov 14 17:08:03 2024 +0100 New testcase for operator new/delete removal. * g++.dg/tree-ssa/dce-1.C: New test. Diff: ---

[gcc r15-5255] Remove allocations which are used only for NULL pointer check and free

https://gcc.gnu.org/g:7828dc070510f8f5c06765858815fa6e5d2d1ec6 commit r15-5255-g7828dc070510f8f5c06765858815fa6e5d2d1ec6 Author: Jan Hubicka Date: Thu Nov 14 17:01:12 2024 +0100 Remove allocations which are used only for NULL pointer check and free Extend tree-ssa-dse to remove me

[gcc r15-5238] Add testcases for std::vector optimization

https://gcc.gnu.org/g:20faea418d580d4db99aa5293c976ab16867aca5 commit r15-5238-g20faea418d580d4db99aa5293c976ab16867aca5 Author: Jan Hubicka Date: Thu Nov 14 14:19:04 2024 +0100 Add testcases for std::vector optimization gcc/testsuite/ChangeLog: PR tree-optimizati

[gcc r15-5176] Verify that empty std::vector is optimized away

https://gcc.gnu.org/g:2264b68796aa6f1a609987cf7edb9183bf070e7e commit r15-5176-g2264b68796aa6f1a609987cf7edb9183bf070e7e Author: Jan Hubicka Date: Tue Nov 12 15:58:02 2024 +0100 Verify that empty std::vector is optimized away With __builtin_operator_new we now can optimize away un

[gcc r12-10736] Zen5 tuning part 1: avoid FMA chains

https://gcc.gnu.org/g:be6334fffdf2a7df3b7f92ea933b804664dfc383 commit r12-10736-gbe6334fffdf2a7df3b7f92ea933b804664dfc383 Author: Jan Hubicka Date: Tue Sep 3 13:38:33 2024 +0200 Zen5 tuning part 1: avoid FMA chains testing matrix multiplication benchmarks shows that FMA on a criti

  1   2   >