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)] Update ChangeLog.* Date: Wed, 19 Apr 2023 22:09:30 +0000 (GMT) [thread overview] Message-ID: <20230419220930.C14153858D33@sourceware.org> (raw) https://gcc.gnu.org/g:1ea2a49ccd0b6fec6d5e6d166447ab94795a2b89 commit 1ea2a49ccd0b6fec6d5e6d166447ab94795a2b89 Author: Michael Meissner <meissner@linux.ibm.com> Date: Wed Apr 19 18:09:27 2023 -0400 Update ChangeLog.* Diff: --- gcc/ChangeLog.meissner | 91 ++++++++++++++++++++++++++++---------------------- 1 file changed, 52 insertions(+), 39 deletions(-) diff --git a/gcc/ChangeLog.meissner b/gcc/ChangeLog.meissner index 8fff1cb0c00..bf4b77a0dc9 100644 --- a/gcc/ChangeLog.meissner +++ b/gcc/ChangeLog.meissner @@ -1,72 +1,66 @@ -==================== Branch work119, patch #23 ==================== +==================== Branch work119, patch #45 was reverted ==================== -Optimize V8HI/V16QI vec_extract from memory with constant element number - -This patch adds combiner insns to fold in conversion to DImode from vec_extract -of a V8HI or V16QI variable with constant element number. With this patch, GCC -will directly emit LHA, LHZ, or LBZ without needing an instruction to do the -sign or zero extension. +Fold conversion to float into V4SI vsx_extract from memory. +This patch folds conversion to floating point of vsx_extract from memory of V4SI +elements where the element number is constant. This code optimizes things so it +will load the integer with LFIWAX or LFIWZX directly into a vector register +rather than loading it into a GPR and doing a direct move operation. 2023-04-18 Michael Meissner <meissner@linux.ibm.com> gcc/ - * config/rs6000/vsx.md (vsx_extract_<mode>_load_to_udi): New insn. - (sx_extract_v8hi_load_to_sd): New insn. + * config/rs6000/vsx.md (SIGN_ZERO_EXTEND): New mode attribute. + (vsx_extract_v4si_load_to_<uns><mode>): New insn. gcc/testsuite/ - * gcc.target/powerpc/vec-extract-mem-char-1.c: New file. - * gcc.target/powerpc/vec-extract-mem-short-1.c: New file. + * gcc.target/powerpc/vec-extract-mem-int-2.c: New file. -==================== Branch work119, patch #22 ==================== +==================== Branch work119, patch #44 was reverted ==================== -Add float/double conversions fro V4SI vec_extract +==================== Branch work119, patch #43 ==================== -This patch adds combiner insns to fold in conversion to float, double, or the -IEEE 128-bit types (both signed and unsigned) of V4SI vec_extract with a -constant element. With this patch, GCC will load the SImode value directly into -the vector register with LFIWZX or LFIWAX instead of doing a LWZ and then moving -the value over with a direct move before the floating point conversion. +Fold sign or zero convert into vsx_extract from memory. + +This patch folds sign or zero convert operations into vsx_extract from memory +where the element number is constant. 2023-04-18 Michael Meissner <meissner@linux.ibm.com> gcc/ - * config/rs6000/vsx.md (SIGN_ZERO): New code attribute. - (FL_CONSTRAINT): New code attribute. - (vsx_extract_v4si_load_to_<uns><mode): New insn. + * config/rs6000/vsx.md (VSX_EXTRACT_ISIGN): New mode attribute. + (vsx_extract_<mode>_load_to_udi): New insn. + (vsx_extract_<mode>_load_to_sdi): New insn. + (vsx_extract_v8hi_load_to_<su>si): New insn. gcc/testsuite/ - * gcc.target/powerpc/vec-extract-mem-int-3.c: New file. - -==================== Branch work119, patch #21 ==================== + * gcc.target/powerpc/vec-extract-mem-char-1.c: New file. + * gcc.target/powerpc/vec-extract-mem-int-1.c: New file. + * gcc.target/powerpc/vec-extract-mem-short-1.c: New file. -Add sign/zero conversions for V4SI vec_extract +==================== Branch work119, patch #42 ==================== -This patch adds combiner insns to fold in sign and zero extension of vec_extract -of V4SI with a constant element when expanding to DImode. +Allow vec_extract to load vector registers. 2023-04-18 Michael Meissner <meissner@linux.ibm.com> gcc/ - * config/rs6000/vsx.md (vsx_extract_v4si_load_to_<su>di): New insn. - -gcc/testsuite/ - - * gcc.target/powerpc/vec-extract-mem-int-2.c: New test. + * config/rs6000/vsx.md (VSX_EX_ISA): New mode attribute + (vsx_extract_<mode>_load): Allow vec_extract of integer types with a + constant element number to load into vector registers. Allow splitting + before register allocation. -==================== Branch work119, patch #21 ==================== +==================== Branch work119, patch #41 ==================== -Enhance vec_extract from int memory with constant element numbers. +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. -Without this patch, the compiler would load the value into a GPR register and -then do a direct move if it needs the value in a vector register. 2023-04-18 Michael Meissner <meissner@linux.ibm.com> @@ -74,9 +68,28 @@ gcc/ * config/rs6000/vsx.md (vsx_extract_v4sf_to_df_load): New insn. -gcc/testsuite/ +==================== Branch work119, patch #40 ==================== + +Allow vec_extract support functions to be called before reload. + +2023-04-19 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. - * gcc.target/powerpc/vec-extract-mem-int-1.c: New test. +==================== Branch work119, patch #33 was reverted ==================== + +==================== Branch work119, patch #32 was reverted ==================== + +==================== Branch work119, patch #31 was reverted ==================== + +==================== Branch work119, patch #30 was reverted ==================== + +==================== Branch work119, patch #23 was reverted ==================== ==================== Branch work119, patch #22 ==================== @@ -88,7 +101,7 @@ In doing other work, I noticed that there was an insn: Which did not have an iterator. I removed the useless <mode>. -2023-04-17 Michael Meissner <meissner@linux.ibm.com> +2023-04-18 Michael Meissner <meissner@linux.ibm.com> gcc/
next reply other threads:[~2023-04-19 22:09 UTC|newest] Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top 2023-04-19 22:09 Michael Meissner [this message] -- strict thread matches above, loose matches on Subject: below -- 2023-04-24 23:01 Michael Meissner 2023-04-22 0:58 Michael Meissner 2023-04-21 23:33 Michael Meissner 2023-04-21 23:08 Michael Meissner 2023-04-21 23:06 Michael Meissner 2023-04-21 22:02 Michael Meissner 2023-04-21 15:23 Michael Meissner 2023-04-21 4:23 Michael Meissner 2023-04-21 2:57 Michael Meissner 2023-04-20 20:48 Michael Meissner 2023-04-20 19:42 Michael Meissner 2023-04-18 22:16 Michael Meissner 2023-04-18 5:45 Michael Meissner 2023-04-17 20:02 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=20230419220930.C14153858D33@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).