public inbox for gcc-cvs@sourceware.org help / color / mirror / Atom feed
From: Michael Meissner <meissner@gcc.gnu.org> To: gcc-cvs@gcc.gnu.org Subject: [gcc(refs/users/meissner/heads/work119)] Revert patches Date: Fri, 21 Apr 2023 23:07:59 +0000 (GMT) [thread overview] Message-ID: <20230421230759.F40F63858D20@sourceware.org> (raw) https://gcc.gnu.org/g:70ee48b31d4b97b697212187efafe8fa9ed71559 commit 70ee48b31d4b97b697212187efafe8fa9ed71559 Author: Michael Meissner <meissner@linux.ibm.com> Date: Fri Apr 21 19:07:56 2023 -0400 Revert patches Diff: --- gcc/ChangeLog.meissner | 58 +++++++++++++++++++++++++++++++++++++++++++++--- gcc/config/rs6000/vsx.md | 38 +++++++------------------------ 2 files changed, 63 insertions(+), 33 deletions(-) diff --git a/gcc/ChangeLog.meissner b/gcc/ChangeLog.meissner index c157e73db69..61634f875b1 100644 --- a/gcc/ChangeLog.meissner +++ b/gcc/ChangeLog.meissner @@ -1,8 +1,60 @@ -==================== Branch work119, patch #52 was reverted ==================== +==================== Branch work119, patch #52 ==================== -==================== Branch work119, patch #51 was reverted ==================== +Improve vec_extract of V4SF with variable element number. -==================== Branch work119, patch #50 was reverted ==================== +This patch adds a combine insn that merges loading up a vec_extract of V4SFmode +where the element number is variable combined with a conversion to DFmode. + +I also modified the insn for vec_extract of V4SFmode where the element number is +variable to split before register allocation. + +2023-04-21 Michael Meissner <meissner@linux.ibm.com> + +gcc/ + + * config/rs6000/vsx.md (vsx_extract_v4sf_var_load): Allow split before + register allocation. + (vsx_extract_v4sf_var_load_to_df): New insn. + +gcc/testsuite/ + + * gcc.target/powerpc/vec-extract-mem-float-2.c: New test. + +==================== Branch work119, patch #51 ==================== + +Combine vec_extract of V4SF with DF convert. + +This patch adds a combine insn that merges loading up a vec_extract of V4SFmode +where the element number is constant combined with a conversion to DFmode. + +In addition, I changed the vec_extract of V4SFmode where the element number is +constant without conversion to do the split before register allocation. I also +simplified the alternatives. + +2023-04-21 Michael Meissner <meissner@linux.ibm.com> + +gcc/ + + * config/rs6000/vsx.md (vsx_extract_v4sf_load): Allow split before + register allocation. + (vsx_extract_v4sf_to_df_load): New insn. + +gcc/testsuite/ + + * gcc.target/powerpc/vec-extract-mem-float-1.c: New test. + +==================== Branch work119, patch #50 ==================== + +Allow vec_extract support functions to be called before reload. + +2023-04-21 Michael Meissner <meissner@linux.ibm.com> + +gcc/ + + * config/rs6000/rs6000.cc (get_vector_offset): Allow being called before + register allocation. + (adjust_vec_address_pcrel): Likewise. + (rs6000_adjust_vec_address): Likewise. ==================== Branch work119, patch #49 was reverted ==================== diff --git a/gcc/config/rs6000/vsx.md b/gcc/config/rs6000/vsx.md index 17e56ab1ce4..417aff5e24b 100644 --- a/gcc/config/rs6000/vsx.md +++ b/gcc/config/rs6000/vsx.md @@ -3549,45 +3549,23 @@ [(set_attr "length" "8") (set_attr "type" "fp")]) -;; V4SF extract from memory convert to DFmode with constant element number. If -;; the element number is 0, we don't need a temporary register. (define_insn_and_split "*vsx_extract_v4sf_load" - [(set (match_operand:SF 0 "register_operand" "=wa,wa,?r,?r") + [(set (match_operand:SF 0 "register_operand" "=f,v,v,?r") (vec_select:SF - (match_operand:V4SF 1 "memory_operand" "m,Q,m,Q") - (parallel - [(match_operand:QI 2 "const_0_to_3_operand" "O,n,O,n")]))) - (clobber (match_scratch:P 3 "=X,&b,X,&b"))] + (match_operand:V4SF 1 "memory_operand" "m,Z,m,m") + (parallel [(match_operand:QI 2 "const_0_to_3_operand" "n,n,n,n")]))) + (clobber (match_scratch:P 3 "=&b,&b,&b,&b"))] "VECTOR_MEM_VSX_P (V4SFmode)" "#" - "&& 1" + "&& reload_completed" [(set (match_dup 0) (match_dup 4))] { operands[4] = rs6000_adjust_vec_address (operands[0], operands[1], operands[2], operands[3], SFmode); } - [(set_attr "type" "fpload,fpload,load,load") - (set_attr "length" "4,8,4,8")]) - -;; V4SF extract from memory and convert to DFmode with constant element number. -(define_insn_and_split "*vsx_extract_v4sf_load_to_df" - [(set (match_operand:DF 0 "register_operand" "=wa,wa") - (float_extend:DF - (vec_select:SF - (match_operand:V4SF 1 "memory_operand" "m,Q") - (parallel [(match_operand:QI 2 "const_0_to_3_operand" "O,n")])))) - (clobber (match_scratch:P 3 "=X,&b"))] - "VECTOR_MEM_VSX_P (V4SFmode)" - "#" - "&& 1" - [(set (match_dup 0) - (float_extend:DF (match_dup 4)))] -{ - operands[4] = rs6000_adjust_vec_address (operands[0], operands[1], operands[2], - operands[3], SFmode); -} - [(set_attr "type" "fpload") - (set_attr "length" "4,8")]) + [(set_attr "type" "fpload,fpload,fpload,load") + (set_attr "length" "8") + (set_attr "isa" "*,p7v,p9v,*")]) ;; Variable V4SF extract from a register (define_insn_and_split "vsx_extract_v4sf_var"
next reply other threads:[~2023-04-21 23:07 UTC|newest] Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top 2023-04-21 23:07 Michael Meissner [this message] -- strict thread matches above, loose matches on Subject: below -- 2023-05-01 17:16 Michael Meissner 2023-04-29 1:28 Michael Meissner 2023-04-28 22:57 Michael Meissner 2023-04-28 21:39 Michael Meissner 2023-04-28 21:30 Michael Meissner 2023-04-28 19:07 Michael Meissner 2023-04-28 18:28 Michael Meissner 2023-04-28 18:11 Michael Meissner 2023-04-28 17:33 Michael Meissner 2023-04-28 6:29 Michael Meissner 2023-04-28 3:57 Michael Meissner 2023-04-27 20:50 Michael Meissner 2023-04-27 19:33 Michael Meissner 2023-04-27 19:11 Michael Meissner 2023-04-27 2:58 Michael Meissner 2023-04-27 2:51 Michael Meissner 2023-04-27 2:48 Michael Meissner 2023-04-26 23:55 Michael Meissner 2023-04-26 15:37 Michael Meissner 2023-04-26 15:35 Michael Meissner 2023-04-26 5:05 Michael Meissner 2023-04-25 22:46 Michael Meissner 2023-04-25 15:48 Michael Meissner 2023-04-25 6:40 Michael Meissner 2023-04-25 6:28 Michael Meissner 2023-04-25 2:15 Michael Meissner 2023-04-25 1:58 Michael Meissner 2023-04-25 1:51 Michael Meissner 2023-04-24 22:57 Michael Meissner 2023-04-22 5:50 Michael Meissner 2023-04-21 23:05 Michael Meissner 2023-04-21 20:19 Michael Meissner 2023-04-21 19:50 Michael Meissner 2023-04-21 19:36 Michael Meissner 2023-04-21 19:03 Michael Meissner 2023-04-21 18:10 Michael Meissner 2023-04-21 15:37 Michael Meissner 2023-04-21 15:36 Michael Meissner 2023-04-21 15:27 Michael Meissner 2023-04-21 15:09 Michael Meissner 2023-04-21 3:57 Michael Meissner 2023-04-21 3:45 Michael Meissner 2023-04-21 3:09 Michael Meissner 2023-04-20 23:41 Michael Meissner 2023-04-19 22:11 Michael Meissner 2023-04-19 21:47 Michael Meissner 2023-04-19 19:28 Michael Meissner 2023-04-19 19:21 Michael Meissner 2023-04-19 19:14 Michael Meissner 2023-04-19 16:36 Michael Meissner 2023-04-19 16:22 Michael Meissner 2023-04-18 22:13 Michael Meissner 2023-04-18 5:37 Michael Meissner 2023-04-17 22:28 Michael Meissner
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20230421230759.F40F63858D20@sourceware.org \ --to=meissner@gcc.gnu.org \ --cc=gcc-cvs@gcc.gnu.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).