[gcc r15-992] Remove value_range typedef.

https://gcc.gnu.org/g:6fb43d1b90e2683f6f41f04a65341fc2dab4b495 commit r15-992-g6fb43d1b90e2683f6f41f04a65341fc2dab4b495 Author: Aldy Hernandez Date: Fri May 31 15:49:26 2024 +0200 Remove value_range typedef. Now that pointers and integers have been disambiguated from irange,

[gcc r15-785] [prange] Use type agnostic range in phiopt [PR115191]

https://gcc.gnu.org/g:35a293a6454ac0cd88735036f536d8f4ec65951a commit r15-785-g35a293a6454ac0cd88735036f536d8f4ec65951a Author: Aldy Hernandez Date: Wed May 22 22:32:57 2024 +0200 [prange] Use type agnostic range in phiopt [PR115191] Fix a use of int_range_max in phiopt that

[gcc r15-632] [prange] Drop range to VARYING if the bitmask intersection made it so [PR115131]

https://gcc.gnu.org/g:1accf4454a2ab57c4d681d1f6db332c46c61c058 commit r15-632-g1accf4454a2ab57c4d681d1f6db332c46c61c058 Author: Aldy Hernandez Date: Fri May 17 13:44:08 2024 +0200 [prange] Drop range to VARYING if the bitmask intersection made it so [PR115131] If the

[gcc r15-627] [prange] Avoid looking at type() for undefined ranges

https://gcc.gnu.org/g:bc6e336cb7c85094ddc77757be97c3d8588f35ca commit r15-627-gbc6e336cb7c85094ddc77757be97c3d8588f35ca Author: Aldy Hernandez Date: Fri May 17 10:30:03 2024 +0200 [prange] Avoid looking at type() for undefined ranges Undefined ranges have no type. This patch

[gcc r15-575] Revert "Revert: "Enable prange support.""

https://gcc.gnu.org/g:da73261ce7731be7f2b164f1db796878cdc23365 commit r15-575-gda73261ce7731be7f2b164f1db796878cdc23365 Author: Aldy Hernandez Date: Fri May 10 00:38:51 2024 +0200 Revert "Revert: "Enable prange support."" This reverts commit

[gcc r15-574] Cleanup prange sanity checks.

https://gcc.gnu.org/g:b8e3574e68310f68116f157a35d5650600d13718 commit r15-574-gb8e3574e68310f68116f157a35d5650600d13718 Author: Aldy Hernandez Date: Thu May 16 09:47:56 2024 +0200 Cleanup prange sanity checks. The pointers_handled_p() code was a temporary sanity check, and not

[gcc r15-573] Use a boolean type when folding conditionals in simplify_using_ranges.

https://gcc.gnu.org/g:f6bed6d3fcc13880ffa786b6c616e2306efe2bf3 commit r15-573-gf6bed6d3fcc13880ffa786b6c616e2306efe2bf3 Author: Aldy Hernandez Date: Thu May 16 09:22:55 2024 +0200 Use a boolean type when folding conditionals in simplify_using_ranges. In adding some traps for

[gcc r15-504] [prange] Default pointers_handled_p() to false.

https://gcc.gnu.org/g:c400b2100719d0a9e5989c63e0827b9e98919df3 commit r15-504-gc400b2100719d0a9e5989c63e0827b9e98919df3 Author: Aldy Hernandez Date: Tue May 14 16:21:50 2024 +0200 [prange] Default pointers_handled_p() to false. The pointers_handled_p() method is an internal

[gcc r15-379] [prange] Default unimplemented prange operators to false.

https://gcc.gnu.org/g:18c93c65a9fbaaf3762198e78fb3c24b9b6fd9fc commit r15-379-g18c93c65a9fbaaf3762198e78fb3c24b9b6fd9fc Author: Aldy Hernandez Date: Fri May 10 18:55:34 2024 +0200 [prange] Default unimplemented prange operators to false. The canonical way to indicate that a range

[gcc r15-378] [prange] Do not trap by default on range dispatch mismatches.

https://gcc.gnu.org/g:45ef24f2702dac4f8707ca682ed364019ee90c44 commit r15-378-g45ef24f2702dac4f8707ca682ed364019ee90c44 Author: Aldy Hernandez Date: Fri May 10 23:21:29 2024 +0200 [prange] Do not trap by default on range dispatch mismatches. The trap in the range-op dispatch code

[gcc r15-363] [prange] Fix thinko in prange::update_bitmask() [PR115026]

https://gcc.gnu.org/g:cbd420a1c3e2bb549dc83b53cc9a31aa6b23952c commit r15-363-gcbd420a1c3e2bb549dc83b53cc9a31aa6b23952c Author: Aldy Hernandez Date: Fri May 10 12:26:49 2024 +0200 [prange] Fix thinko in prange::update_bitmask() [PR115026] gcc/ChangeLog: PR

[gcc r15-357] [prange] Do not assume all pointers are the same size [PR115009]

https://gcc.gnu.org/g:ac255c7afeb8a558bd6224ff77277eebcd849d6e commit r15-357-gac255c7afeb8a558bd6224ff77277eebcd849d6e Author: Aldy Hernandez Date: Thu May 9 23:37:30 2024 +0200 [prange] Do not assume all pointers are the same size [PR115009] In a world with same sized pointers

[gcc r15-353] Revert: "Enable prange support." [PR114985]

https://gcc.gnu.org/g:d7bb8eaade3cd3aa70715c8567b4d7b08098e699 commit r15-353-gd7bb8eaade3cd3aa70715c8567b4d7b08098e699 Author: Aldy Hernandez Date: Fri May 10 00:29:13 2024 +0200 Revert: "Enable prange support." [PR114985] This reverts commit

[gcc r15-336] [ranger] Force buffer alignment in Value_Range [PR114912]

https://gcc.gnu.org/g:d7ff8ae5313bea755f5960786b33a7b151e7b663 commit r15-336-gd7ff8ae5313bea755f5960786b33a7b151e7b663 Author: Aldy Hernandez Date: Fri May 3 11:17:32 2024 +0200 [ranger] Force buffer alignment in Value_Range [PR114912] gcc/ChangeLog: PR

[gcc r15-335] [prange] Reword dispatch error message

https://gcc.gnu.org/g:be3df704ce7de417682d57bc3e819dfcf0fdd501 commit r15-335-gbe3df704ce7de417682d57bc3e819dfcf0fdd501 Author: Aldy Hernandez Date: Wed May 8 22:50:22 2024 +0200 [prange] Reword dispatch error message After reading the ICE for the PR, it's obvious the error

[gcc r15-312] Enable prange support.

https://gcc.gnu.org/g:36e877996936abd8bd08f8b1d983c8d1023a5842 commit r15-312-g36e877996936abd8bd08f8b1d983c8d1023a5842 Author: Aldy Hernandez Date: Tue May 7 14:05:50 2024 +0200 Enable prange support. This throws the switch on prange. After this patch, it is no longer valid

[gcc r15-270] Minor range type fixes for IPA in preparation for prange.

https://gcc.gnu.org/g:24853cd854eb9b8a5c7b9706ad0908221bf964ce commit r15-270-g24853cd854eb9b8a5c7b9706ad0908221bf964ce Author: Aldy Hernandez Date: Tue Mar 19 17:55:58 2024 +0100 Minor range type fixes for IPA in preparation for prange. The polymorphic Value_Range object takes a

[gcc r15-164] Add prange entries in gimple-range-op.cc.

https://gcc.gnu.org/g:6cec31d44a426fa031ca07266fd2723b0038de83 commit r15-164-g6cec31d44a426fa031ca07266fd2723b0038de83 Author: Aldy Hernandez Date: Wed Mar 20 11:27:21 2024 +0100 Add prange entries in gimple-range-op.cc. gcc/ChangeLog: * gimple-range-op.cc

[gcc r15-163] Implement operator_ge for prange....

https://gcc.gnu.org/g:ff306c77b7cf5b7b09914f38b7351328835ac4ce commit r15-163-gff306c77b7cf5b7b09914f38b7351328835ac4ce Author: Aldy Hernandez Date: Wed Mar 20 11:15:03 2024 +0100 Implement operator_ge for prange gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-162] Implement operator_gt for prange.

https://gcc.gnu.org/g:76fae4051a72b2d417d50f1980dff8ab0c50d0c5 commit r15-162-g76fae4051a72b2d417d50f1980dff8ab0c50d0c5 Author: Aldy Hernandez Date: Wed Mar 20 11:10:03 2024 +0100 Implement operator_gt for prange. gcc/ChangeLog: * range-op-mixed.h: Add overloaded

[gcc r15-161] Implement operator_le for prange.

https://gcc.gnu.org/g:3a4ee6ea8627efe0d34a71d0ea4ce9b70d34df18 commit r15-161-g3a4ee6ea8627efe0d34a71d0ea4ce9b70d34df18 Author: Aldy Hernandez Date: Wed Mar 20 11:07:30 2024 +0100 Implement operator_le for prange. gcc/ChangeLog: * range-op-mixed.h: Add overloaded

[gcc r15-160] Implement operator_lt for prange.

https://gcc.gnu.org/g:bfa2323d1ddf96a4f40a26aa39b8e1e3bd9b6d98 commit r15-160-gbfa2323d1ddf96a4f40a26aa39b8e1e3bd9b6d98 Author: Aldy Hernandez Date: Wed Mar 20 11:03:24 2024 +0100 Implement operator_lt for prange. gcc/ChangeLog: * range-op-mixed.h: Add overloaded

[gcc r15-158] Implement operator_not_equal for prange.

https://gcc.gnu.org/g:d1be4c907fa47d868d6ef31e8fd7ede0535420ca commit r15-158-gd1be4c907fa47d868d6ef31e8fd7ede0535420ca Author: Aldy Hernandez Date: Wed Mar 20 10:49:11 2024 +0100 Implement operator_not_equal for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-159] Implement operator_equal for prange.

https://gcc.gnu.org/g:bcb226924f50d9b84ca32b6f39abb63b43e23229 commit r15-159-gbcb226924f50d9b84ca32b6f39abb63b43e23229 Author: Aldy Hernandez Date: Wed Mar 20 10:54:39 2024 +0100 Implement operator_equal for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-157] Implement operator_bitwise_or for prange.

https://gcc.gnu.org/g:6b9e640d49138183aaeee67f2dcf3de329318d27 commit r15-157-g6b9e640d49138183aaeee67f2dcf3de329318d27 Author: Aldy Hernandez Date: Wed Mar 20 10:29:50 2024 +0100 Implement operator_bitwise_or for prange. We seem to have a range-op entry for pointer bitwise OR

[gcc r15-156] Implement operator_bitwise_and for prange.

https://gcc.gnu.org/g:e58f14916954411628eb122da996383b8c996b57 commit r15-156-ge58f14916954411628eb122da996383b8c996b57 Author: Aldy Hernandez Date: Wed Mar 20 10:23:31 2024 +0100 Implement operator_bitwise_and for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-155] Implement operator_pointer_diff for prange.

https://gcc.gnu.org/g:f803b93feef60c8c2d4f7f7270bfc94650dbc8f5 commit r15-155-gf803b93feef60c8c2d4f7f7270bfc94650dbc8f5 Author: Aldy Hernandez Date: Wed Mar 20 10:12:47 2024 +0100 Implement operator_pointer_diff for prange. gcc/ChangeLog: * range-op-ptr.cc

[gcc r15-154] Implement pointer_plus_operator for prange.

https://gcc.gnu.org/g:86ff3c45ea82452888244476f26a4f628b148ace commit r15-154-g86ff3c45ea82452888244476f26a4f628b148ace Author: Aldy Hernandez Date: Wed Mar 20 10:04:41 2024 +0100 Implement pointer_plus_operator for prange. gcc/ChangeLog: * range-op-ptr.cc (class

[gcc r15-153] Implement operator_addr_expr for prange.

https://gcc.gnu.org/g:54d3fd6d9f5d029c23ab376df2f5decb4902907d commit r15-153-g54d3fd6d9f5d029c23ab376df2f5decb4902907d Author: Aldy Hernandez Date: Wed Mar 20 09:51:33 2024 +0100 Implement operator_addr_expr for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-152] Implement operator_min and operator_max for prange.

https://gcc.gnu.org/g:1a4f5d499109d3f2a06bfd1403b6d47d6f55e481 commit r15-152-g1a4f5d499109d3f2a06bfd1403b6d47d6f55e481 Author: Aldy Hernandez Date: Wed Mar 20 08:44:49 2024 +0100 Implement operator_min and operator_max for prange. gcc/ChangeLog: *

[gcc r15-151] Implement operator_cast for prange.

https://gcc.gnu.org/g:95fce0dc73002d5c9557fa7260c7ba1e761136ff commit r15-151-g95fce0dc73002d5c9557fa7260c7ba1e761136ff Author: Aldy Hernandez Date: Wed Mar 20 08:04:32 2024 +0100 Implement operator_cast for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-150] Implement operator_cst for prange.

https://gcc.gnu.org/g:a91fd7b4342dbeaf1d2514beaee3af0bb5680b81 commit r15-150-ga91fd7b4342dbeaf1d2514beaee3af0bb5680b81 Author: Aldy Hernandez Date: Wed Mar 20 07:55:57 2024 +0100 Implement operator_cst for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-149] Implement operator_identity for prange.

https://gcc.gnu.org/g:e7b6e9663e9b31e681fb0302338bcb4bb306a334 commit r15-149-ge7b6e9663e9b31e681fb0302338bcb4bb306a334 Author: Aldy Hernandez Date: Wed Mar 20 07:50:11 2024 +0100 Implement operator_identity for prange. gcc/ChangeLog: * range-op-mixed.h: Add

[gcc r15-147] Add prange implementation for get_legacy_range.

https://gcc.gnu.org/g:ddf039adef5e2695f1cb27e0b6e5056fef4be2ee commit r15-147-gddf039adef5e2695f1cb27e0b6e5056fef4be2ee Author: Aldy Hernandez Date: Wed Mar 20 06:39:48 2024 +0100 Add prange implementation for get_legacy_range. gcc/ChangeLog: * value-range.cc

[gcc r15-148] Implement range-op dispatch for prange.

https://gcc.gnu.org/g:31377eed515506c9e8ba2ac8fa3ab4e743f8c1f3 commit r15-148-g31377eed515506c9e8ba2ac8fa3ab4e743f8c1f3 Author: Aldy Hernandez Date: Wed Mar 20 07:19:45 2024 +0100 Implement range-op dispatch for prange. This patch adds the range-op dispatch code for prange, and

[gcc r15-146] Add hashing support for prange.

https://gcc.gnu.org/g:7c1b136630790eb34d57c45d9a816b32fd904e3f commit r15-146-g7c1b136630790eb34d57c45d9a816b32fd904e3f Author: Aldy Hernandez Date: Wed Mar 20 06:39:07 2024 +0100 Add hashing support for prange. gcc/ChangeLog: * value-range.cc (add_vrange): Add

[gcc r15-145] Add storage support for prange.

https://gcc.gnu.org/g:f859996a976ca185f371a8ed395e9c8e459e05b0 commit r15-145-gf859996a976ca185f371a8ed395e9c8e459e05b0 Author: Aldy Hernandez Date: Wed Mar 20 06:38:06 2024 +0100 Add storage support for prange. gcc/ChangeLog: * value-range-storage.cc

[gcc r15-144] Add streaming support for prange.

https://gcc.gnu.org/g:a7f1285380029d2928d61e61032c1948ccabc495 commit r15-144-ga7f1285380029d2928d61e61032c1948ccabc495 Author: Aldy Hernandez Date: Wed Mar 20 06:34:26 2024 +0100 Add streaming support for prange. gcc/ChangeLog: * data-streamer-in.cc

[gcc r15-143] Implement basic prange class.

https://gcc.gnu.org/g:64993a89ad75814ab69addade1b2c0020a180f41 commit r15-143-g64993a89ad75814ab69addade1b2c0020a180f41 Author: Aldy Hernandez Date: Wed Mar 20 06:25:52 2024 +0100 Implement basic prange class. This provides a bare prange class with bounds and bitmasks. It will

[gcc r15-142] Minimal prange class showing inlining degradation to VRP.

https://gcc.gnu.org/g:f5891967947562060076956bd953e5df4c7289bf commit r15-142-gf5891967947562060076956bd953e5df4c7289bf Author: Aldy Hernandez Date: Sat May 4 06:45:18 2024 +0200 Minimal prange class showing inlining degradation to VRP. There is a 2% slowdown to VRP unrelated to

Re: 1.76% performance loss in VRP due to inlining

t; > On 4/30/24 12:22, Jakub Jelinek wrote: > > On Tue, Apr 30, 2024 at 03:09:51PM -0400, Jason Merrill via Gcc wrote: > >> On Fri, Apr 26, 2024 at 5:44 AM Aldy Hernandez via Gcc > >> wrote: > >>> > >>> In implementing prange (pointer ranges)

[gcc r15-88] Reduce startup costs for Value_Range.

https://gcc.gnu.org/g:c60b3e211c555706cdc2dc8bfcdd540152cff350 commit r15-88-gc60b3e211c555706cdc2dc8bfcdd540152cff350 Author: Aldy Hernandez Date: Tue Apr 30 19:39:00 2024 +0200 Reduce startup costs for Value_Range. Value_Range is our polymorphic temporary that can hold any

[gcc r15-87] Cleanups to unsupported_range.

https://gcc.gnu.org/g:1b5732de7e3980aa5197b1ac818f48f1ce9f87ab commit r15-87-g1b5732de7e3980aa5197b1ac818f48f1ce9f87ab Author: Aldy Hernandez Date: Tue Apr 30 18:54:11 2024 +0200 Cleanups to unsupported_range. Here are some cleanups to unsupported_range so the assignment operator

[gcc r15-70] Change int_range<2> to infinite precision.

https://gcc.gnu.org/g:0b2735e0797fee9b4ec5cd74f22afe0483f888dd commit r15-70-g0b2735e0797fee9b4ec5cd74f22afe0483f888dd Author: Aldy Hernandez Date: Tue Apr 30 10:36:58 2024 +0200 Change int_range<2> to infinite precision. In my previous change I mistakenly changed Value_Range to

Re: 1.76% performance loss in VRP due to inlining

On Tue, Apr 30, 2024 at 9:58 AM Richard Biener wrote: > > On Fri, Apr 26, 2024 at 11:45 AM Aldy Hernandez via Gcc > wrote: > > > > Hi folks! > > > > In implementing prange (pointer ranges), I have found a 1.74% slowdown > > in VRP, even without any cod

[gcc r15-41] Callers of irange_bitmask must normalize value/mask pairs.

https://gcc.gnu.org/g:d71308d5a681de00ea291136c162e5b46c7c commit r15-41-gd71308d5a681de00ea291136c162e5b46c7c Author: Aldy Hernandez Date: Tue Apr 23 10:12:56 2024 +0200 Callers of irange_bitmask must normalize value/mask pairs. As per the documentation, irange_bitmask

[gcc r15-40] Remove range_zero and range_nonzero.

https://gcc.gnu.org/g:3b9abfd2df5fe720798aab1e21b4a11876607561 commit r15-40-g3b9abfd2df5fe720798aab1e21b4a11876607561 Author: Aldy Hernandez Date: Wed Mar 20 05:51:55 2024 +0100 Remove range_zero and range_nonzero. Remove legacy range_zero and range_nonzero as they return by

[gcc r15-39] Move print_irange_* out of vrange_printer class.

https://gcc.gnu.org/g:df6a1bc59a355c9fee10d29f54c9dca81612afb6 commit r15-39-gdf6a1bc59a355c9fee10d29f54c9dca81612afb6 Author: Aldy Hernandez Date: Tue Mar 19 20:26:27 2024 +0100 Move print_irange_* out of vrange_printer class. Move some code out of the irange pretty printers so

[gcc r15-37] Make some integer specific ranges generic Value_Range's.

https://gcc.gnu.org/g:c284f8d2d16ce9c29defce3329419ccc54605ad4 commit r15-37-gc284f8d2d16ce9c29defce3329419ccc54605ad4 Author: Aldy Hernandez Date: Tue Mar 19 18:22:08 2024 +0100 Make some integer specific ranges generic Value_Range's. There are some irange uses that should be

[gcc r15-38] Accept any vrange in range_includes_zero_p.

https://gcc.gnu.org/g:b102633be7d0b763d106b0a883679bb1497ca17c commit r15-38-gb102633be7d0b763d106b0a883679bb1497ca17c Author: Aldy Hernandez Date: Tue Mar 19 18:29:21 2024 +0100 Accept any vrange in range_includes_zero_p. Accept a vrange, as this will be used for either integers

[gcc r15-36] Move get_bitmask_from_range out of irange class.

https://gcc.gnu.org/g:2caf7a50a6a9de80d2767d82b8cdb69d63469aaf commit r15-36-g2caf7a50a6a9de80d2767d82b8cdb69d63469aaf Author: Aldy Hernandez Date: Tue Mar 19 18:04:55 2024 +0100 Move get_bitmask_from_range out of irange class. prange will also have bitmasks, so it will need to

[gcc r15-34] Verify that reading back from vrange_storage doesn't drop bits.

https://gcc.gnu.org/g:92f74ee21218cab08d7bb7769004a65e8a291fa3 commit r15-34-g92f74ee21218cab08d7bb7769004a65e8a291fa3 Author: Aldy Hernandez Date: Tue Mar 19 16:35:41 2024 +0100 Verify that reading back from vrange_storage doesn't drop bits. We have a sanity check in the irange

[gcc r15-35] Accept a vrange in get_legacy_range.

https://gcc.gnu.org/g:9a2f0d152d98dd55efc9accd07ea507b929c3516 commit r15-35-g9a2f0d152d98dd55efc9accd07ea507b929c3516 Author: Aldy Hernandez Date: Tue Mar 19 17:17:53 2024 +0100 Accept a vrange in get_legacy_range. In preparation for prange, make get_legacy_range take a generic

[gcc r15-33] Change range_includes_zero_p argument to a reference.

https://gcc.gnu.org/g:d883fc7d00ed6bf5ee151de4fd3e05431582bd5f commit r15-33-gd883fc7d00ed6bf5ee151de4fd3e05431582bd5f Author: Aldy Hernandez Date: Tue Mar 19 16:33:47 2024 +0100 Change range_includes_zero_p argument to a reference. Make range_includes_zero_p take an argument

[gcc r15-32] Make fold_cond_with_ops use a boolean type for range_true/range_false.

https://gcc.gnu.org/g:039e88b1aea5723221e8b0b926c35afb2f96a8a9 commit r15-32-g039e88b1aea5723221e8b0b926c35afb2f96a8a9 Author: Aldy Hernandez Date: Wed Feb 7 11:27:29 2024 +0100 Make fold_cond_with_ops use a boolean type for range_true/range_false. Conditional operators are

[gcc r15-31] Remove GTY support for vrange and derived classes.

https://gcc.gnu.org/g:eeef1f69c5e77ecf13fdcf44df5bcf592a9993e6 commit r15-31-geeef1f69c5e77ecf13fdcf44df5bcf592a9993e6 Author: Aldy Hernandez Date: Wed Feb 21 09:34:29 2024 +0100 Remove GTY support for vrange and derived classes. Now that we have a vrange storage class to save

[gcc r15-29] Add tree versions of lower and upper bounds to vrange.

https://gcc.gnu.org/g:ba1a8e8eed963c0253c6e5550c8bccc264c5d469 commit r15-29-gba1a8e8eed963c0253c6e5550c8bccc264c5d469 Author: Aldy Hernandez Date: Mon Apr 22 13:34:48 2024 +0200 Add tree versions of lower and upper bounds to vrange. This patch adds vrange::lbound() and

[gcc r15-30] Move bitmask routines to vrange base class.

https://gcc.gnu.org/g:fd4cf7a092bb2ce21c0d8246c17c0b7f82de440c commit r15-30-gfd4cf7a092bb2ce21c0d8246c17c0b7f82de440c Author: Aldy Hernandez Date: Thu Feb 22 09:18:46 2024 +0100 Move bitmask routines to vrange base class. Any range can theoretically have a bitmask of set bits.

[gcc r15-28] Make some Value_Range's explicitly integer.

https://gcc.gnu.org/g:a46564e4876c9a863d9897d72963cc4f03689adc commit r15-28-ga46564e4876c9a863d9897d72963cc4f03689adc Author: Aldy Hernandez Date: Mon Apr 22 13:29:39 2024 +0200 Make some Value_Range's explicitly integer. Fix some Value_Range's that we know ahead of time will be

[gcc r15-27] Add a virtual vrange destructor.

https://gcc.gnu.org/g:a78dfb0fc83606e9b83b76575deb7e43300254fa commit r15-27-ga78dfb0fc83606e9b83b76575deb7e43300254fa Author: Aldy Hernandez Date: Wed Feb 21 09:33:19 2024 +0100 Add a virtual vrange destructor. Richi mentioned in PR113476 that it would be cleaner to move the

[gcc r15-26] Make vrange an abstract class.

https://gcc.gnu.org/g:21713edf52974d14238cdda865fc662eca58302a commit r15-26-g21713edf52974d14238cdda865fc662eca58302a Author: Aldy Hernandez Date: Wed Feb 21 20:37:14 2024 +0100 Make vrange an abstract class. Explicitly make vrange an abstract class. This involves fleshing out

1.76% performance loss in VRP due to inlining

Hi folks! In implementing prange (pointer ranges), I have found a 1.74% slowdown in VRP, even without any code path actually using the code. I have tracked this down to irange::get_bitmask() being compiled differently with and without the bare bones patch. With the patch, irange::get_bitmask()

[COMMITTED] [frange] Revert relation handling in LTGT_EXPR.

In trying to come up with a missing testcase for commit 979e0fbf53cd, I've realized the patch doesn't catch anything. A relation of VREL_EQ in foperator_ltgt::fold_range() is either both arguments the same (x LTGT_EXPR x), which we should never emit, or two arguments that are actually the same,

[COMMITTED] [irange] Fix typo in contains_zero_p.

In the conversion of iranges to wide_int (commit cb779afeff204f), I mistakenly made contains_zero_p() return TRUE for undefined ranges. This means the rest of the patch was adjusted for this stupidity. For example, we ended up doing the following, to make up for the fact that contains_zero_p was

[COMMITTED] [frange] Handle relations in LTGT_EXPR.

LTGT_EXPR hasn't been handling relations, especially with NANs as a possibility. This handles them while documenting how relations work in a world with NANs. Basically we need to special case VREL_EQ before calling frelop_early_resolve. Note that VREL_EQ on entry to a range-op entry is really

[PATCH] [frange] Relax floating point relational folding.

[Jakub/Andrew: I've been staring at this for far too long and could use another pair of eyes.] This patch implements a new frelop_early_resolve() that handles the NAN special cases instead of calling into the integer version which can break for some combinations. Relaxing FP conditional folding

[COMMITTED] [frange] Return false if nothing changed in union_nans().

When one operand is a known NAN, we always return TRUE from union_nans(), even if no change occurred. This patch fixes the oversight. gcc/ChangeLog: * value-range.cc (frange::union_nans): Return false if nothing changed. (range_tests_floats): New test. ---

[COMMITTED] [irange] Return FALSE if updated bitmask is unchanged [PR110753]

The mask/value pair we track in the irange is a bit fickle in that it can sometimes contradict the bitmask inherent in the range. This can happen when a series of calculations yield a combination such as: [3, 1000] MASK 0xfffe VALUE 0x0 The mask/value above implies that the lowest

Re: [PATCH] Read global value/mask in IPA.

On 7/31/23 18:47, Martin Jambor wrote: Hello, On Tue, Jul 18 2023, Aldy Hernandez wrote: On 7/17/23 15:14, Aldy Hernandez wrote: Instead of reading the known zero bits in IPA, read the value/mask pair which is available. There is a slight change of behavior here. I have removed the check

Re: Fix profile upate after vectorizer peeling

On 8/3/23 16:29, Jeff Law wrote: On 8/3/23 08:23, Jan Hubicka wrote: Jeff, an help would be appreciated here :) I will try to debug this.  One option would be to disable branch prediciton on vect_check for time being - it is not inlined anyway Not a lot of insight.  The backwards threader

Re: [PATCH] Read global value/mask in IPA.

PING * 2 On Tue, Jul 25, 2023 at 8:32 AM Aldy Hernandez wrote: > > Ping > > On Mon, Jul 17, 2023, 15:14 Aldy Hernandez wrote: >> >> Instead of reading the known zero bits in IPA, read the value/mask >> pair which is available. >> >> There is a slight change of behavior here. I have removed the

[COMMITTED] [range-ops] Remove special case for handling bitmasks in casts.

Now that we can generically handle bitmasks for unary operators, there's no need to special case them. gcc/ChangeLog: * range-op-mixed.h (class operator_cast): Add update_bitmask. * range-op.cc (operator_cast::update_bitmask): New. (operator_cast::fold_range): Call

[COMMITTED] [range-ops] Handle bitmasks for ABSU_EXPR.

gcc/ChangeLog: * range-op.cc (class operator_absu): Add update_bitmask. (operator_absu::update_bitmask): New. --- gcc/range-op.cc | 9 + 1 file changed, 9 insertions(+) diff --git a/gcc/range-op.cc b/gcc/range-op.cc index bfab53caea0..5653ca0d186 100644 ---

[COMMITTED] [range-ops] Handle bitmasks for ABS_EXPR.

gcc/ChangeLog: * range-op-mixed.h (class operator_abs): Add update_bitmask. * range-op.cc (operator_abs::update_bitmask): New. --- gcc/range-op-mixed.h | 2 ++ gcc/range-op.cc | 6 ++ 2 files changed, 8 insertions(+) diff --git a/gcc/range-op-mixed.h

[COMMITTED] [range-ops] Handle bitmasks for BIT_NOT_EXPR.

gcc/ChangeLog: * range-op-mixed.h (class operator_bitwise_not): Add update_bitmask. * range-op.cc (operator_bitwise_not::update_bitmask): New. --- gcc/range-op-mixed.h | 2 ++ gcc/range-op.cc | 7 +++ 2 files changed, 9 insertions(+) diff --git a/gcc/range-op-mixed.h

[COMMITTED] [range-ops] Handle bitmasks for unary operators.

It looks like we missed out on bitmasks for unary operators because we were using bit_value_binop exclusively. This patch hands off to bit_value_unop when appropriate, thus allowing us to handle ABS and BIT_NOT_EXPR, and others. Follow-up patches will add the tweaks for the range-ops entries

Re: [PATCH] range-op-float: Fix up -frounding-math frange_arithmetic +- handling [PR110755]

The frange bits look fine to me, so if you feel confident in the math logic, go right ahead :). Thanks. Aldy On 7/24/23 18:01, Jakub Jelinek wrote: Hi! IEEE754 says that x + (-x) and x - x result in +0 in all rounding modes but rounding towards negative infinity, in which case the result is

[PATCH] Initialize value in bit_value_unop.

bit_value_binop initializes VAL regardless of the final mask. It even has a comment to that effect: /* Ensure that VAL is initialized (to any value). */ However, bit_value_unop, which in theory shares the same API, does not. This causes range-ops to choke on uninitialized VALs for some

[COMMITTED] Make some functions in CCP static.

Committed as obvious. gcc/ChangeLog: * tree-ssa-ccp.cc (value_mask_to_min_max): Make static. (bit_value_mult_const): Same. (get_individual_bits): Same. --- gcc/tree-ssa-ccp.cc | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/gcc/tree-ssa-ccp.cc

Re: [PATCH] Read global value/mask in IPA.

Ping On Mon, Jul 17, 2023, 15:14 Aldy Hernandez wrote: > Instead of reading the known zero bits in IPA, read the value/mask > pair which is available. > > There is a slight change of behavior here. I have removed the check > for SSA_NAME, as the ranger can calculate the range and value/mask

Re: [PATCH] Read global value/mask in IPA.

On 7/17/23 15:14, Aldy Hernandez wrote: Instead of reading the known zero bits in IPA, read the value/mask pair which is available. There is a slight change of behavior here. I have removed the check for SSA_NAME, as the ranger can calculate the range and value/mask for INTEGER_CST. This

[PATCH] Read global value/mask in IPA.

Instead of reading the known zero bits in IPA, read the value/mask pair which is available. There is a slight change of behavior here. I have removed the check for SSA_NAME, as the ranger can calculate the range and value/mask for INTEGER_CST. This simplifies the code a bit, since there's no

[PATCH] Export value/mask known bits from IPA.

Currently IPA throws away the known 1 bits because VRP and irange have traditionally only had a way of tracking known 0s (set_nonzero_bits). With the ability to keep all the known bits in the irange, we can now save this between passes. OK? gcc/ChangeLog: * ipa-prop.cc

[PATCH] Export value/mask known bits from CCP.

Currently CCP throws away the known 1 bits because VRP and irange have traditionally only had a way of tracking known 0s (set_nonzero_bits). With the ability to keep all the known bits in the irange, we can now save this between passes. OK? gcc/ChangeLog: * tree-ssa-ccp.cc

[COMMITTED] Normalize irange_bitmask before union/intersect.

The bit twiddling in union/intersect for the value/mask pair must be normalized to have the unknown bits with a value of 0 in order to make the math simpler. Normalizing at construction slowed VRP by 1.5% so I opted to normalize before updating the bitmask in range-ops, since it was the only

[COMMITTED] Add global setter for value/mask pair for SSA names.

This patch provides a way to set the value/mask pair of known bits globally, similarly to how we can use set_nonzero_bits for known 0 bits. This can then be used by CCP and IPA to set value/mask info instead of throwing away the known 1 bits. In further clean-ups, I will see if it makes sense to

Re: [PATCH V4] Optimize '(X - N * M) / N' to 'X / N - M' if valid

On 7/14/23 15:37, Richard Biener wrote: On Fri, 14 Jul 2023, Aldy Hernandez wrote: I don't know what you're trying to accomplish here, as I haven't been following the PR, but adding all these helper functions to the ranger header file seems wrong, especially since there's only one use of

Re: [PATCH V4] Optimize '(X - N * M) / N' to 'X / N - M' if valid

I don't know what you're trying to accomplish here, as I haven't been following the PR, but adding all these helper functions to the ranger header file seems wrong, especially since there's only one use of them. I see you're tweaking the irange API, adding helper functions to range-op (which

Re: [COMMITTED] [range-op] Take known set bits into account in popcount [PR107053]

On 7/12/23 23:50, Jeff Law wrote: On 7/12/23 15:15, Aldy Hernandez via Gcc-patches wrote: This patch teaches popcount about known set bits which are now available in the irange. PR tree-optimization/107053 gcc/ChangeLog: * gimple-range-op.cc (cfn_popcount): Use known set bits

[COMMITTED] [range-op] Take known set bits into account in popcount [PR107053]

This patch teaches popcount about known set bits which are now available in the irange. PR tree-optimization/107053 gcc/ChangeLog: * gimple-range-op.cc (cfn_popcount): Use known set bits. gcc/testsuite/ChangeLog: * gcc.dg/tree-ssa/pr107053.c: New test. ---

[COMMITTED] [range-op] Take known mask into account for bitwise ands [PR107043]

PR tree-optimization/107043 gcc/ChangeLog: * range-op.cc (operator_bitwise_and::op1_range): Update bitmask. gcc/testsuite/ChangeLog: * gcc.dg/tree-ssa/pr107043.c: New test. --- gcc/range-op.cc | 8

[COMMITTED] [range-op] Enable value/mask propagation in range-op.

Throw the switch in range-ops to make full use of the value/mask information instead of only the nonzero bits. This will cause most of the operators implemented in range-ops to use the value/mask information calculated by CCP's bit_value_binop() function which range-ops uses. This opens up more

[COMMITTED] Implement value/mask tracking for irange.

Integer ranges (irange) currently track known 0 bits. We've wanted to track known 1 bits for some time, and instead of tracking known 0 and known 1's separately, it has been suggested we track a value/mask pair similarly to what we do for CCP and RTL. This patch implements such a thing. With

[COMMITTED] The caller to irange::intersect (wide_int, wide_int) must normalize the range.

Per the function comment, the caller to intersect(wide_int, wide_int) must handle the mask. This means it must also normalize the range if anything changed. gcc/ChangeLog: * value-range.cc (irange::intersect): Leave normalization to caller. --- gcc/value-range.cc | 7 +-- 1

[COMMITTED] A singleton irange has all known bits.

gcc/ChangeLog: * value-range.cc (irange::get_bitmask_from_range): Return all the known bits for a singleton. (irange::set_range_from_bitmask): Set a range of a singleton when all bits are known. --- gcc/value-range.cc | 19 ++- 1 file changed, 18

[COMMITTED] Tidy up the range normalization code.

There's a few spots where a range is being altered in-place, but we fail to call normalize the range. This patch makes sure we always call normalize_kind(), and that normalize_kind in turn, calls verify_range to make sure verything is canonical. gcc/ChangeLog: * value-range.cc

[COMMITTED] Move maybe_set_nonzero_bits() to its only user.

gcc/ChangeLog: * tree-vrp.cc (maybe_set_nonzero_bits): Move from here... * tree-ssa-dom.cc (maybe_set_nonzero_bits): ...to here. * tree-vrp.h (maybe_set_nonzero_bits): Remove. --- gcc/tree-ssa-dom.cc | 65 + gcc/tree-vrp.cc

Re: [PATCH] Implement ipa_vr hashing.

ng*2 > > > > On Wed, Jun 14, 2023, 14:11 Aldy Hernandez wrote: > >> > >> PING > >> > >> On Sat, Jun 10, 2023 at 10:30 PM Aldy Hernandez wrote: > >> > > >> > > >> > > >> > On 5/29/23 16:51, Martin Jam

Re: [PATCH] Implement ipa_vr hashing.

5/29/23 16:51, Martin Jambor wrote: >> > > Hi, >> > > >> > > On Mon, May 22 2023, Aldy Hernandez via Gcc-patches wrote: >> > >> Implement hashing for ipa_vr. When all is said and done, all these >> > >> patches incurr a 7.64% slowdo

Re: [PATCH] Implement ipa_vr hashing.

Ping*2 On Wed, Jun 14, 2023, 14:11 Aldy Hernandez wrote: > PING > > On Sat, Jun 10, 2023 at 10:30 PM Aldy Hernandez wrote: > > > > > > > > On 5/29/23 16:51, Martin Jambor wrote: > > > Hi, > > > > > > On Mon, May 22 2023, Aldy Hernande

  1   2   3   4   5   6   7   8   9   10   >