qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Richard Henderson <richard.henderson@linaro.org>
To: qemu-devel@nongnu.org
Subject: [Qemu-devel] [PATCH v6 15/26] target/arm: Use vector infrastructure for aa64 zip/uzp/trn/xtn
Date: Tue, 21 Nov 2017 22:25:23 +0100	[thread overview]
Message-ID: <20171121212534.5177-16-richard.henderson@linaro.org> (raw)
In-Reply-To: <20171121212534.5177-1-richard.henderson@linaro.org>

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/translate-a64.c | 103 +++++++++++++++------------------------------
 1 file changed, 35 insertions(+), 68 deletions(-)

diff --git a/target/arm/translate-a64.c b/target/arm/translate-a64.c
index 55a4902fc2..8769b4505a 100644
--- a/target/arm/translate-a64.c
+++ b/target/arm/translate-a64.c
@@ -5576,11 +5576,7 @@ static void disas_simd_zip_trn(DisasContext *s, uint32_t insn)
     int opcode = extract32(insn, 12, 2);
     bool part = extract32(insn, 14, 1);
     bool is_q = extract32(insn, 30, 1);
-    int esize = 8 << size;
-    int i, ofs;
-    int datasize = is_q ? 128 : 64;
-    int elements = datasize / esize;
-    TCGv_i64 tcg_res, tcg_resl, tcg_resh;
+    GVecGen3Fn *gvec_fn;
 
     if (opcode == 0 || (size == 3 && !is_q)) {
         unallocated_encoding(s);
@@ -5591,60 +5587,24 @@ static void disas_simd_zip_trn(DisasContext *s, uint32_t insn)
         return;
     }
 
-    tcg_resl = tcg_const_i64(0);
-    tcg_resh = tcg_const_i64(0);
-    tcg_res = tcg_temp_new_i64();
-
-    for (i = 0; i < elements; i++) {
-        switch (opcode) {
-        case 1: /* UZP1/2 */
-        {
-            int midpoint = elements / 2;
-            if (i < midpoint) {
-                read_vec_element(s, tcg_res, rn, 2 * i + part, size);
-            } else {
-                read_vec_element(s, tcg_res, rm,
-                                 2 * (i - midpoint) + part, size);
-            }
-            break;
-        }
-        case 2: /* TRN1/2 */
-            if (i & 1) {
-                read_vec_element(s, tcg_res, rm, (i & ~1) + part, size);
-            } else {
-                read_vec_element(s, tcg_res, rn, (i & ~1) + part, size);
-            }
-            break;
-        case 3: /* ZIP1/2 */
-        {
-            int base = part * elements / 2;
-            if (i & 1) {
-                read_vec_element(s, tcg_res, rm, base + (i >> 1), size);
-            } else {
-                read_vec_element(s, tcg_res, rn, base + (i >> 1), size);
-            }
-            break;
-        }
-        default:
-            g_assert_not_reached();
-        }
-
-        ofs = i * esize;
-        if (ofs < 64) {
-            tcg_gen_shli_i64(tcg_res, tcg_res, ofs);
-            tcg_gen_or_i64(tcg_resl, tcg_resl, tcg_res);
-        } else {
-            tcg_gen_shli_i64(tcg_res, tcg_res, ofs - 64);
-            tcg_gen_or_i64(tcg_resh, tcg_resh, tcg_res);
-        }
+    switch (opcode) {
+    case 1: /* UZP1/2 */
+        gvec_fn = part ? tcg_gen_gvec_uzpo : tcg_gen_gvec_uzpe;
+        break;
+    case 2: /* TRN1/2 */
+        gvec_fn = part ? tcg_gen_gvec_trno : tcg_gen_gvec_trne;
+        break;
+    case 3: /* ZIP1/2 */
+        gvec_fn = part ? tcg_gen_gvec_ziph : tcg_gen_gvec_zipl;
+        break;
+    default:
+        g_assert_not_reached();
     }
 
-    tcg_temp_free_i64(tcg_res);
-
-    write_vec_element(s, tcg_resl, rd, 0, MO_64);
-    tcg_temp_free_i64(tcg_resl);
-    write_vec_element(s, tcg_resh, rd, 1, MO_64);
-    tcg_temp_free_i64(tcg_resh);
+    gvec_fn(size, vec_full_reg_offset(s, rd),
+            vec_full_reg_offset(s, rn),
+            vec_full_reg_offset(s, rm),
+            is_q ? 16 : 8, vec_full_reg_size(s));
 }
 
 static void do_minmaxop(DisasContext *s, TCGv_i32 tcg_elt1, TCGv_i32 tcg_elt2,
@@ -7922,6 +7882,22 @@ static void handle_2misc_narrow(DisasContext *s, bool scalar,
     int destelt = is_q ? 2 : 0;
     int passes = scalar ? 1 : 2;
 
+    if (opcode == 0x12 && !u) { /* XTN, XTN2 */
+        tcg_debug_assert(!scalar);
+        if (is_q) { /* XTN2 */
+            tcg_gen_gvec_uzpe(size, vec_reg_offset(s, rd, 1, MO_64),
+                              vec_reg_offset(s, rn, 0, MO_64),
+                              vec_reg_offset(s, rn, 1, MO_64),
+                              8, vec_full_reg_size(s) - 8);
+        } else {
+            tcg_gen_gvec_uzpe(size, vec_reg_offset(s, rd, 0, MO_64),
+                              vec_reg_offset(s, rn, 0, MO_64),
+                              vec_reg_offset(s, rn, 1, MO_64),
+                              8, vec_full_reg_size(s));
+        }
+        return;
+    }
+
     if (scalar) {
         tcg_res[1] = tcg_const_i32(0);
     }
@@ -7939,23 +7915,14 @@ static void handle_2misc_narrow(DisasContext *s, bool scalar,
         tcg_res[pass] = tcg_temp_new_i32();
 
         switch (opcode) {
-        case 0x12: /* XTN, SQXTUN */
+        case 0x12: /* , SQXTUN */
         {
-            static NeonGenNarrowFn * const xtnfns[3] = {
-                gen_helper_neon_narrow_u8,
-                gen_helper_neon_narrow_u16,
-                tcg_gen_extrl_i64_i32,
-            };
             static NeonGenNarrowEnvFn * const sqxtunfns[3] = {
                 gen_helper_neon_unarrow_sat8,
                 gen_helper_neon_unarrow_sat16,
                 gen_helper_neon_unarrow_sat32,
             };
-            if (u) {
-                genenvfn = sqxtunfns[size];
-            } else {
-                genfn = xtnfns[size];
-            }
+            genenvfn = sqxtunfns[size];
             break;
         }
         case 0x14: /* SQXTN, UQXTN */
-- 
2.13.6

  parent reply	other threads:[~2017-11-21 21:28 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-11-21 21:25 [Qemu-devel] [PATCH v6 00/26] tcg: generic vector operations Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 01/26] tcg: Remove TCGV_UNUSED* and TCGV_IS_UNUSED* Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 02/26] tcg: Dynamically allocate TCGOps Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 03/26] tcg: Generalize TCGOp parameters Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 04/26] tcg: Add types and basic operations for host vectors Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 05/26] tcg: Add generic vector expanders Richard Henderson
2017-12-06 10:21   ` Kirill Batuzov
2017-12-08 21:35     ` Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 06/26] tcg: Allow multiple word entries into the constant pool Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 07/26] tcg: Add tcg_signed_cond Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 08/26] target/arm: Align vector registers Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 09/26] target/arm: Use vector infrastructure for aa64 add/sub/logic Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 10/26] target/arm: Use vector infrastructure for aa64 mov/not/neg Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 11/26] target/arm: Use vector infrastructure for aa64 dup/movi Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 12/26] tcg/i386: Add vector operations Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 13/26] tcg: Add tcg_expand_vec_op and tcg-target.opc.h Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 14/26] tcg: Add generic vector ops for interleave Richard Henderson
2017-11-21 21:25 ` Richard Henderson [this message]
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 16/26] tcg: Add generic vector ops for constant shifts Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 17/26] target/arm: Use vector infrastructure for aa64 " Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 18/26] tcg: Add generic vector ops for comparisons Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 19/26] target/arm: Use vector infrastructure for aa64 compares Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 20/26] tcg/i386: Add vector operations/expansions for shift/cmp/interleave Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 21/26] tcg: Add generic vector ops for multiplication Richard Henderson
2017-12-05 11:33   ` Kirill Batuzov
2017-12-08 21:36     ` Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 22/26] target/arm: Use vector infrastructure for aa64 multiplies Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 23/26] tcg: Add generic vector ops for extension Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 24/26] target/arm: Use vector infrastructure for aa64 widening shifts Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 25/26] tcg/i386: Add vector operations/expansions for mul/extend Richard Henderson
2017-11-21 21:25 ` [Qemu-devel] [PATCH v6 26/26] tcg/aarch64: Add vector operations Richard Henderson
2017-11-21 22:10 ` [Qemu-devel] [PATCH v6 00/26] tcg: generic " no-reply
2017-11-21 22:19 ` no-reply
2017-11-21 22:23 ` no-reply
2017-11-21 22:44 ` no-reply
2017-11-27 16:09 ` Timothy Pearson

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=20171121212534.5177-16-richard.henderson@linaro.org \
    --to=richard.henderson@linaro.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).