qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Richard Henderson <richard.henderson@linaro.org>
To: qemu-devel@nongnu.org
Cc: qemu-arm@nongnu.org
Subject: [PATCH v2 03/29] target/arm: Use cmpsel in gen_ushl_vec
Date: Mon,  9 Sep 2024 09:22:13 -0700	[thread overview]
Message-ID: <20240909162240.647173-4-richard.henderson@linaro.org> (raw)
In-Reply-To: <20240909162240.647173-1-richard.henderson@linaro.org>

Instead of cmp+and or cmp+andc, use cmpsel.  This will
be better for hosts that use predicate registers for cmp.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/tcg/gengvec.c | 19 ++++++++-----------
 1 file changed, 8 insertions(+), 11 deletions(-)

diff --git a/target/arm/tcg/gengvec.c b/target/arm/tcg/gengvec.c
index 726a1383ae..3edbf3a262 100644
--- a/target/arm/tcg/gengvec.c
+++ b/target/arm/tcg/gengvec.c
@@ -1041,7 +1041,7 @@ static void gen_ushl_vec(unsigned vece, TCGv_vec dst,
     TCGv_vec rval = tcg_temp_new_vec_matching(dst);
     TCGv_vec lsh = tcg_temp_new_vec_matching(dst);
     TCGv_vec rsh = tcg_temp_new_vec_matching(dst);
-    TCGv_vec max;
+    TCGv_vec max, zero;
 
     tcg_gen_neg_vec(vece, rsh, shift);
     if (vece == MO_8) {
@@ -1061,23 +1061,20 @@ static void gen_ushl_vec(unsigned vece, TCGv_vec dst,
     tcg_gen_shrv_vec(vece, rval, src, rsh);
 
     /*
-     * The choice of LT (signed) and GEU (unsigned) are biased toward
+     * The choice of GE (signed) and GEU (unsigned) are biased toward
      * the instructions of the x86_64 host.  For MO_8, the whole byte
      * is significant so we must use an unsigned compare; otherwise we
      * have already masked to a byte and so a signed compare works.
      * Other tcg hosts have a full set of comparisons and do not care.
      */
+    zero = tcg_constant_vec_matching(dst, vece, 0);
     max = tcg_constant_vec_matching(dst, vece, 8 << vece);
     if (vece == MO_8) {
-        tcg_gen_cmp_vec(TCG_COND_GEU, vece, lsh, lsh, max);
-        tcg_gen_cmp_vec(TCG_COND_GEU, vece, rsh, rsh, max);
-        tcg_gen_andc_vec(vece, lval, lval, lsh);
-        tcg_gen_andc_vec(vece, rval, rval, rsh);
+        tcg_gen_cmpsel_vec(TCG_COND_GEU, vece, lval, lsh, max, zero, lval);
+        tcg_gen_cmpsel_vec(TCG_COND_GEU, vece, rval, rsh, max, zero, rval);
     } else {
-        tcg_gen_cmp_vec(TCG_COND_LT, vece, lsh, lsh, max);
-        tcg_gen_cmp_vec(TCG_COND_LT, vece, rsh, rsh, max);
-        tcg_gen_and_vec(vece, lval, lval, lsh);
-        tcg_gen_and_vec(vece, rval, rval, rsh);
+        tcg_gen_cmpsel_vec(TCG_COND_GE, vece, lval, lsh, max, zero, lval);
+        tcg_gen_cmpsel_vec(TCG_COND_GE, vece, rval, rsh, max, zero, rval);
     }
     tcg_gen_or_vec(vece, dst, lval, rval);
 }
@@ -1087,7 +1084,7 @@ void gen_gvec_ushl(unsigned vece, uint32_t rd_ofs, uint32_t rn_ofs,
 {
     static const TCGOpcode vecop_list[] = {
         INDEX_op_neg_vec, INDEX_op_shlv_vec,
-        INDEX_op_shrv_vec, INDEX_op_cmp_vec, 0
+        INDEX_op_shrv_vec, INDEX_op_cmpsel_vec, 0
     };
     static const GVecGen3 ops[4] = {
         { .fniv = gen_ushl_vec,
-- 
2.43.0



  parent reply	other threads:[~2024-09-09 16:26 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-09-09 16:22 [PATCH v2 00/29] target/arm: AdvSIMD decodetree conversion, part 4 Richard Henderson
2024-09-09 16:22 ` [PATCH v2 01/29] target/arm: Replace tcg_gen_dupi_vec with constants in gengvec.c Richard Henderson
2024-09-09 16:22 ` [PATCH v2 02/29] target/arm: Replace tcg_gen_dupi_vec with constants in translate-sve.c Richard Henderson
2024-09-09 16:22 ` Richard Henderson [this message]
2024-09-09 16:22 ` [PATCH v2 04/29] target/arm: Use cmpsel in gen_sshl_vec Richard Henderson
2024-09-09 16:22 ` [PATCH v2 05/29] target/arm: Use tcg_gen_extract2_i64 for EXT Richard Henderson
2024-09-09 22:22   ` Philippe Mathieu-Daudé
2024-09-09 16:22 ` [PATCH v2 06/29] target/arm: Convert EXT to decodetree Richard Henderson
2024-09-09 22:22   ` Philippe Mathieu-Daudé
2024-09-09 16:22 ` [PATCH v2 07/29] target/arm: Convert TBL, TBX " Richard Henderson
2024-09-09 22:23   ` Philippe Mathieu-Daudé
2024-09-09 16:22 ` [PATCH v2 08/29] target/arm: Convert UZP, TRN, ZIP " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 09/29] target/arm: Simplify do_reduction_op Richard Henderson
2024-09-09 16:22 ` [PATCH v2 10/29] target/arm: Convert ADDV, *ADDLV, *MAXV, *MINV to decodetree Richard Henderson
2024-09-09 16:22 ` [PATCH v2 11/29] target/arm: Convert FMAXNMV, FMINNMV, FMAXV, FMINV " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 12/29] target/arm: Convert FMOVI (scalar, immediate) " Richard Henderson
2024-09-09 22:24   ` Philippe Mathieu-Daudé
2024-09-10 12:27     ` Peter Maydell
2024-09-10 14:35       ` Philippe Mathieu-Daudé
2024-09-09 16:22 ` [PATCH v2 13/29] target/arm: Convert MOVI, FMOV, ORR, BIC (vector " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 14/29] target/arm: Introduce gen_gvec_sshr, gen_gvec_ushr Richard Henderson
2024-09-09 16:22 ` [PATCH v2 15/29] target/arm: Fix whitespace near gen_srshr64_i64 Richard Henderson
2024-09-10 14:37   ` Philippe Mathieu-Daudé
2024-09-09 16:22 ` [PATCH v2 16/29] target/arm: Convert handle_vec_simd_shri to decodetree Richard Henderson
2024-09-09 16:22 ` [PATCH v2 17/29] target/arm: Convert handle_vec_simd_shli " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 18/29] target/arm: Use {, s}extract in handle_vec_simd_wshli Richard Henderson
2024-09-09 16:22 ` [PATCH v2 19/29] target/arm: Convert SSHLL, USHLL to decodetree Richard Henderson
2024-09-09 16:22 ` [PATCH v2 20/29] target/arm: Push tcg_rnd into handle_shri_with_rndacc Richard Henderson
2024-09-10 14:37   ` Philippe Mathieu-Daudé
2024-09-09 16:22 ` [PATCH v2 21/29] target/arm: Split out subroutines of handle_shri_with_rndacc Richard Henderson
2024-09-09 16:22 ` [PATCH v2 22/29] target/arm: Convert SHRN, RSHRN to decodetree Richard Henderson
2024-09-09 16:22 ` [PATCH v2 23/29] target/arm: Convert handle_scalar_simd_shri " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 24/29] target/arm: Convert handle_scalar_simd_shli " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 25/29] target/arm: Convert VQSHL, VQSHLU to gvec Richard Henderson
2024-09-09 16:22 ` [PATCH v2 26/29] target/arm: Widen NeonGenNarrowEnvFn return to 64 bits Richard Henderson
2024-09-10 14:19   ` Peter Maydell
2024-09-09 16:22 ` [PATCH v2 27/29] target/arm: Convert SQSHL, UQSHL, SQSHLU (immediate) to decodetree Richard Henderson
2024-09-09 16:22 ` [PATCH v2 28/29] target/arm: Convert vector [US]QSHRN, [US]QRSHRN, SQSHRUN " Richard Henderson
2024-09-09 16:22 ` [PATCH v2 29/29] target/arm: Convert scalar " Richard Henderson
2024-09-10 15:30 ` [PATCH v2 00/29] target/arm: AdvSIMD decodetree conversion, part 4 Peter Maydell

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=20240909162240.647173-4-richard.henderson@linaro.org \
    --to=richard.henderson@linaro.org \
    --cc=qemu-arm@nongnu.org \
    --cc=qemu-devel@nongnu.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: link
Be 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 NNTP newsgroup(s).