All of lore.kernel.org
 help / color / mirror / Atom feed
From: Peter Maydell <peter.maydell@linaro.org>
To: qemu-devel@nongnu.org
Subject: [PULL 104/114] target/arm: Implement SVE2 fp multiply-add long
Date: Tue, 25 May 2021 16:07:26 +0100	[thread overview]
Message-ID: <20210525150736.32695-10-peter.maydell@linaro.org> (raw)
In-Reply-To: <20210525150736.32695-1-peter.maydell@linaro.org>

From: Stephen Long <steplong@quicinc.com>

Implements both vectored and indexed FMLALB, FMLALT, FMLSLB, FMLSLT

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Stephen Long <steplong@quicinc.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20210525010358.152808-83-richard.henderson@linaro.org
Message-Id: <20200504171240.11220-1-steplong@quicinc.com>
[rth: Rearrange to use float16_to_float32_by_bits.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 target/arm/helper.h        |  5 +++
 target/arm/sve.decode      | 14 +++++++
 target/arm/translate-sve.c | 75 ++++++++++++++++++++++++++++++++++++++
 target/arm/vec_helper.c    | 47 ++++++++++++++++++++++++
 4 files changed, 141 insertions(+)

diff --git a/target/arm/helper.h b/target/arm/helper.h
index 2e212ae96be..92b81bbabe4 100644
--- a/target/arm/helper.h
+++ b/target/arm/helper.h
@@ -986,6 +986,11 @@ DEF_HELPER_FLAGS_4(sve2_sqrdmulh_idx_s, TCG_CALL_NO_RWG,
 DEF_HELPER_FLAGS_4(sve2_sqrdmulh_idx_d, TCG_CALL_NO_RWG,
                    void, ptr, ptr, ptr, i32)
 
+DEF_HELPER_FLAGS_6(sve2_fmlal_zzzw_s, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_6(sve2_fmlal_zzxw_s, TCG_CALL_NO_RWG,
+                   void, ptr, ptr, ptr, ptr, ptr, i32)
+
 DEF_HELPER_FLAGS_4(gvec_xar_d, TCG_CALL_NO_RWG, void, ptr, ptr, ptr, i32)
 
 #ifdef TARGET_AARCH64
diff --git a/target/arm/sve.decode b/target/arm/sve.decode
index ea98508cddc..78a2a31ab19 100644
--- a/target/arm/sve.decode
+++ b/target/arm/sve.decode
@@ -132,6 +132,8 @@
                 &rrrr_esz ra=%reg_movprfx
 
 # Four operand with unused vector element size
+@rda_rn_rm_e0   ........ ... rm:5 ... ... rn:5 rd:5 \
+                &rrrr_esz esz=0 ra=%reg_movprfx
 @rdn_ra_rm_e0   ........ ... rm:5 ... ... ra:5 rd:5 \
                 &rrrr_esz esz=0 rn=%reg_movprfx
 
@@ -1608,3 +1610,15 @@ FCVTLT_sd       01100100 11 0010 11 101 ... ..... .....  @rd_pg_rn_e0
 
 ### SVE2 floating-point convert to integer
 FLOGB           01100101 00 011 esz:2 0101 pg:3 rn:5 rd:5  &rpr_esz
+
+### SVE2 floating-point multiply-add long (vectors)
+FMLALB_zzzw     01100100 10 1 ..... 10 0 00 0 ..... .....  @rda_rn_rm_e0
+FMLALT_zzzw     01100100 10 1 ..... 10 0 00 1 ..... .....  @rda_rn_rm_e0
+FMLSLB_zzzw     01100100 10 1 ..... 10 1 00 0 ..... .....  @rda_rn_rm_e0
+FMLSLT_zzzw     01100100 10 1 ..... 10 1 00 1 ..... .....  @rda_rn_rm_e0
+
+### SVE2 floating-point multiply-add long (indexed)
+FMLALB_zzxw     01100100 10 1 ..... 0100.0 ..... .....     @rrxr_3a esz=2
+FMLALT_zzxw     01100100 10 1 ..... 0100.1 ..... .....     @rrxr_3a esz=2
+FMLSLB_zzxw     01100100 10 1 ..... 0110.0 ..... .....     @rrxr_3a esz=2
+FMLSLT_zzxw     01100100 10 1 ..... 0110.1 ..... .....     @rrxr_3a esz=2
diff --git a/target/arm/translate-sve.c b/target/arm/translate-sve.c
index 92c0620bc8e..428ae018a35 100644
--- a/target/arm/translate-sve.c
+++ b/target/arm/translate-sve.c
@@ -8535,3 +8535,78 @@ static bool trans_FLOGB(DisasContext *s, arg_rpr_esz *a)
     }
     return true;
 }
+
+static bool do_FMLAL_zzzw(DisasContext *s, arg_rrrr_esz *a, bool sub, bool sel)
+{
+    if (!dc_isar_feature(aa64_sve2, s)) {
+        return false;
+    }
+    if (sve_access_check(s)) {
+        unsigned vsz = vec_full_reg_size(s);
+        tcg_gen_gvec_4_ptr(vec_full_reg_offset(s, a->rd),
+                           vec_full_reg_offset(s, a->rn),
+                           vec_full_reg_offset(s, a->rm),
+                           vec_full_reg_offset(s, a->ra),
+                           cpu_env, vsz, vsz, (sel << 1) | sub,
+                           gen_helper_sve2_fmlal_zzzw_s);
+    }
+    return true;
+}
+
+static bool trans_FMLALB_zzzw(DisasContext *s, arg_rrrr_esz *a)
+{
+    return do_FMLAL_zzzw(s, a, false, false);
+}
+
+static bool trans_FMLALT_zzzw(DisasContext *s, arg_rrrr_esz *a)
+{
+    return do_FMLAL_zzzw(s, a, false, true);
+}
+
+static bool trans_FMLSLB_zzzw(DisasContext *s, arg_rrrr_esz *a)
+{
+    return do_FMLAL_zzzw(s, a, true, false);
+}
+
+static bool trans_FMLSLT_zzzw(DisasContext *s, arg_rrrr_esz *a)
+{
+    return do_FMLAL_zzzw(s, a, true, true);
+}
+
+static bool do_FMLAL_zzxw(DisasContext *s, arg_rrxr_esz *a, bool sub, bool sel)
+{
+    if (!dc_isar_feature(aa64_sve2, s)) {
+        return false;
+    }
+    if (sve_access_check(s)) {
+        unsigned vsz = vec_full_reg_size(s);
+        tcg_gen_gvec_4_ptr(vec_full_reg_offset(s, a->rd),
+                           vec_full_reg_offset(s, a->rn),
+                           vec_full_reg_offset(s, a->rm),
+                           vec_full_reg_offset(s, a->ra),
+                           cpu_env, vsz, vsz,
+                           (a->index << 2) | (sel << 1) | sub,
+                           gen_helper_sve2_fmlal_zzxw_s);
+    }
+    return true;
+}
+
+static bool trans_FMLALB_zzxw(DisasContext *s, arg_rrxr_esz *a)
+{
+    return do_FMLAL_zzxw(s, a, false, false);
+}
+
+static bool trans_FMLALT_zzxw(DisasContext *s, arg_rrxr_esz *a)
+{
+    return do_FMLAL_zzxw(s, a, false, true);
+}
+
+static bool trans_FMLSLB_zzxw(DisasContext *s, arg_rrxr_esz *a)
+{
+    return do_FMLAL_zzxw(s, a, true, false);
+}
+
+static bool trans_FMLSLT_zzxw(DisasContext *s, arg_rrxr_esz *a)
+{
+    return do_FMLAL_zzxw(s, a, true, true);
+}
diff --git a/target/arm/vec_helper.c b/target/arm/vec_helper.c
index f5af45375df..19c4ba1bdf5 100644
--- a/target/arm/vec_helper.c
+++ b/target/arm/vec_helper.c
@@ -1668,6 +1668,27 @@ void HELPER(gvec_fmlal_a64)(void *vd, void *vn, void *vm,
              get_flush_inputs_to_zero(&env->vfp.fp_status_f16));
 }
 
+void HELPER(sve2_fmlal_zzzw_s)(void *vd, void *vn, void *vm, void *va,
+                               void *venv, uint32_t desc)
+{
+    intptr_t i, oprsz = simd_oprsz(desc);
+    uint16_t negn = extract32(desc, SIMD_DATA_SHIFT, 1) << 15;
+    intptr_t sel = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(float16);
+    CPUARMState *env = venv;
+    float_status *status = &env->vfp.fp_status;
+    bool fz16 = get_flush_inputs_to_zero(&env->vfp.fp_status_f16);
+
+    for (i = 0; i < oprsz; i += sizeof(float32)) {
+        float16 nn_16 = *(float16 *)(vn + H1_2(i + sel)) ^ negn;
+        float16 mm_16 = *(float16 *)(vm + H1_2(i + sel));
+        float32 nn = float16_to_float32_by_bits(nn_16, fz16);
+        float32 mm = float16_to_float32_by_bits(mm_16, fz16);
+        float32 aa = *(float32 *)(va + H1_4(i));
+
+        *(float32 *)(vd + H1_4(i)) = float32_muladd(nn, mm, aa, 0, status);
+    }
+}
+
 static void do_fmlal_idx(float32 *d, void *vn, void *vm, float_status *fpst,
                          uint32_t desc, bool fz16)
 {
@@ -1712,6 +1733,32 @@ void HELPER(gvec_fmlal_idx_a64)(void *vd, void *vn, void *vm,
                  get_flush_inputs_to_zero(&env->vfp.fp_status_f16));
 }
 
+void HELPER(sve2_fmlal_zzxw_s)(void *vd, void *vn, void *vm, void *va,
+                               void *venv, uint32_t desc)
+{
+    intptr_t i, j, oprsz = simd_oprsz(desc);
+    uint16_t negn = extract32(desc, SIMD_DATA_SHIFT, 1) << 15;
+    intptr_t sel = extract32(desc, SIMD_DATA_SHIFT + 1, 1) * sizeof(float16);
+    intptr_t idx = extract32(desc, SIMD_DATA_SHIFT + 2, 3) * sizeof(float16);
+    CPUARMState *env = venv;
+    float_status *status = &env->vfp.fp_status;
+    bool fz16 = get_flush_inputs_to_zero(&env->vfp.fp_status_f16);
+
+    for (i = 0; i < oprsz; i += 16) {
+        float16 mm_16 = *(float16 *)(vm + i + idx);
+        float32 mm = float16_to_float32_by_bits(mm_16, fz16);
+
+        for (j = 0; j < 16; j += sizeof(float32)) {
+            float16 nn_16 = *(float16 *)(vn + H1_2(i + j + sel)) ^ negn;
+            float32 nn = float16_to_float32_by_bits(nn_16, fz16);
+            float32 aa = *(float32 *)(va + H1_4(i + j));
+
+            *(float32 *)(vd + H1_4(i + j)) =
+                float32_muladd(nn, mm, aa, 0, status);
+        }
+    }
+}
+
 void HELPER(gvec_sshl_b)(void *vd, void *vn, void *vm, uint32_t desc)
 {
     intptr_t i, opr_sz = simd_oprsz(desc);
-- 
2.20.1



  parent reply	other threads:[~2021-05-25 16:52 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-05-25 15:07 [PULL 095/114] target/arm: Implement SVE2 FCVTLT Peter Maydell
2021-05-25 15:07 ` [PULL 096/114] target/arm: Implement SVE2 FCVTXNT, FCVTX Peter Maydell
2021-05-25 15:07 ` [PULL 097/114] target/arm: Implement SVE2 FLOGB Peter Maydell
2021-05-25 15:07 ` [PULL 098/114] target/arm: Share table of sve load functions Peter Maydell
2021-05-25 15:07 ` [PULL 099/114] target/arm: Tidy do_ldrq Peter Maydell
2021-05-25 15:07 ` [PULL 100/114] target/arm: Implement SVE2 LD1RO Peter Maydell
2021-05-25 15:07 ` [PULL 101/114] target/arm: Implement 128-bit ZIP, UZP, TRN Peter Maydell
2021-05-25 15:07 ` [PULL 102/114] target/arm: Implement SVE2 bitwise shift immediate Peter Maydell
2021-05-25 15:07 ` [PULL 103/114] target/arm: Move endian adjustment macros to vec_internal.h Peter Maydell
2021-05-25 15:07 ` Peter Maydell [this message]
2021-05-25 15:07 ` [PULL 105/114] target/arm: Implement aarch64 SUDOT, USDOT Peter Maydell
2021-05-25 15:07 ` [PULL 106/114] target/arm: Split out do_neon_ddda_fpst Peter Maydell
2021-05-25 15:07 ` [PULL 107/114] target/arm: Remove unused fpst from VDOT_scalar Peter Maydell
2021-05-25 15:07 ` [PULL 108/114] target/arm: Fix decode for VDOT (indexed) Peter Maydell
2021-05-25 15:07 ` [PULL 109/114] target/arm: Split out do_neon_ddda Peter Maydell
2021-05-25 15:07 ` [PULL 110/114] target/arm: Split decode of VSDOT and VUDOT Peter Maydell
2021-05-25 15:07 ` [PULL 111/114] target/arm: Implement aarch32 VSUDOT, VUSDOT Peter Maydell
2021-05-25 15:07 ` [PULL 112/114] target/arm: Implement integer matrix multiply accumulate Peter Maydell
2021-05-25 15:07 ` [PULL 113/114] linux-user/aarch64: Enable hwcap bits for sve2 and related extensions Peter Maydell
2021-05-25 15:07 ` [PULL 114/114] target/arm: Enable SVE2 " 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=20210525150736.32695-10-peter.maydell@linaro.org \
    --to=peter.maydell@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.