[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PATCH v4 25/45] target/arm: Implement BFMOPA, BFMOPS
From: |
Richard Henderson |
Subject: |
[PATCH v4 25/45] target/arm: Implement BFMOPA, BFMOPS |
Date: |
Tue, 28 Jun 2022 09:50:57 +0530 |
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
target/arm/helper-sme.h | 2 ++
target/arm/sme.decode | 2 ++
target/arm/sme_helper.c | 52 ++++++++++++++++++++++++++++++++++++++
target/arm/translate-sme.c | 30 ++++++++++++++++++++++
4 files changed, 86 insertions(+)
diff --git a/target/arm/helper-sme.h b/target/arm/helper-sme.h
index f50d0fe1d6..1d68fb8c74 100644
--- a/target/arm/helper-sme.h
+++ b/target/arm/helper-sme.h
@@ -125,3 +125,5 @@ DEF_HELPER_FLAGS_7(sme_fmopa_s, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, ptr, ptr, i32)
DEF_HELPER_FLAGS_7(sme_fmopa_d, TCG_CALL_NO_RWG,
void, ptr, ptr, ptr, ptr, ptr, ptr, i32)
+DEF_HELPER_FLAGS_6(sme_bfmopa, TCG_CALL_NO_RWG,
+ void, ptr, ptr, ptr, ptr, ptr, i32)
diff --git a/target/arm/sme.decode b/target/arm/sme.decode
index ba4774d174..afd9c0dffd 100644
--- a/target/arm/sme.decode
+++ b/target/arm/sme.decode
@@ -73,3 +73,5 @@ ADDVA_d 11000000 11 01000 1 ... ... ..... 00 ...
@adda_64
FMOPA_s 10000000 100 ..... ... ... ..... . 00 .. @op_32
FMOPA_d 10000000 110 ..... ... ... ..... . 0 ... @op_64
+
+BFMOPA 10000001 100 ..... ... ... ..... . 00 .. @op_32
diff --git a/target/arm/sme_helper.c b/target/arm/sme_helper.c
index 31c53ad896..d2e1057124 100644
--- a/target/arm/sme_helper.c
+++ b/target/arm/sme_helper.c
@@ -961,3 +961,55 @@ void HELPER(sme_fmopa_d)(void *vza, void *vzn, void *vzm,
void *vpn,
}
}
}
+
+/*
+ * Alter PAIR as needed for controlling predicates being false,
+ * and for NEG on an enabled row element.
+ */
+static inline uint32_t f16mop_adj_pair(uint32_t pair, uint32_t pg, uint32_t
neg)
+{
+ pair ^= neg;
+ if (!(pg & 1)) {
+ pair &= 0xffff0000u;
+ }
+ if (!(pg & 4)) {
+ pair &= 0x0000ffffu;
+ }
+ return pair;
+}
+
+void HELPER(sme_bfmopa)(void *vza, void *vzn, void *vzm, void *vpn,
+ void *vpm, uint32_t desc)
+{
+ intptr_t row, col, oprsz = simd_maxsz(desc);
+ uint32_t neg = simd_data(desc) << 15;
+ uint16_t *pn = vpn, *pm = vpm;
+
+ for (row = 0; row < oprsz; ) {
+ uint16_t pa = pn[H2(row >> 4)];
+ do {
+ void *vza_row = vza + row * sizeof(ARMVectorReg);
+ uint32_t n = *(uint32_t *)(vzn + row);
+
+ n = f16mop_adj_pair(n, pa, neg);
+
+ for (col = 0; col < oprsz; ) {
+ uint16_t pb = pm[H2(col >> 4)];
+ do {
+ if ((pa & 0b0101) == 0b0101 || (pb & 0b0101) == 0b0101) {
+ uint32_t *a = vza_row + col;
+ uint32_t m = *(uint32_t *)(vzm + col);
+
+ m = f16mop_adj_pair(m, pb, neg);
+ *a = bfdotadd(*a, n, m);
+
+ col += 4;
+ pb >>= 4;
+ }
+ } while (col & 15);
+ }
+ row += 4;
+ pa >>= 4;
+ } while (row & 15);
+ }
+}
diff --git a/target/arm/translate-sme.c b/target/arm/translate-sme.c
index 1117a61f62..e537a14b6d 100644
--- a/target/arm/translate-sme.c
+++ b/target/arm/translate-sme.c
@@ -302,6 +302,33 @@ TRANS_FEAT(ADDVA_s, aa64_sme, do_adda, a, MO_32,
gen_helper_sme_addva_s)
TRANS_FEAT(ADDHA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addha_d)
TRANS_FEAT(ADDVA_d, aa64_sme_i16i64, do_adda, a, MO_64, gen_helper_sme_addva_d)
+static bool do_outprod(DisasContext *s, arg_op *a, MemOp esz,
+ gen_helper_gvec_5 *fn)
+{
+ int svl = streaming_vec_reg_size(s);
+ uint32_t desc = simd_desc(svl, svl, a->sub);
+ TCGv_ptr za, zn, zm, pn, pm;
+
+ if (!sme_smza_enabled_check(s)) {
+ return true;
+ }
+
+ /* Sum XZR+zad to find ZAd. */
+ za = get_tile_rowcol(s, esz, 31, a->zad, false);
+ zn = vec_full_reg_ptr(s, a->zn);
+ zm = vec_full_reg_ptr(s, a->zm);
+ pn = pred_full_reg_ptr(s, a->pn);
+ pm = pred_full_reg_ptr(s, a->pm);
+
+ fn(za, zn, zm, pn, pm, tcg_constant_i32(desc));
+
+ tcg_temp_free_ptr(za);
+ tcg_temp_free_ptr(zn);
+ tcg_temp_free_ptr(pn);
+ tcg_temp_free_ptr(pm);
+ return true;
+}
+
static bool do_outprod_fpst(DisasContext *s, arg_op *a, MemOp esz,
gen_helper_gvec_5_ptr *fn)
{
@@ -333,3 +360,6 @@ static bool do_outprod_fpst(DisasContext *s, arg_op *a,
MemOp esz,
TRANS_FEAT(FMOPA_s, aa64_sme, do_outprod_fpst, a, MO_32,
gen_helper_sme_fmopa_s)
TRANS_FEAT(FMOPA_d, aa64_sme_f64f64, do_outprod_fpst, a, MO_64,
gen_helper_sme_fmopa_d)
+
+/* TODO: FEAT_EBF16 */
+TRANS_FEAT(BFMOPA, aa64_sme, do_outprod, a, MO_32, gen_helper_sme_bfmopa)
--
2.34.1
- [PATCH v4 13/45] target/arm: Mark LDFF1 and LDNF1 as non-streaming, (continued)
- [PATCH v4 13/45] target/arm: Mark LDFF1 and LDNF1 as non-streaming, Richard Henderson, 2022/06/28
- [PATCH v4 14/45] target/arm: Mark LD1RO as non-streaming, Richard Henderson, 2022/06/28
- [PATCH v4 15/45] target/arm: Add SME enablement checks, Richard Henderson, 2022/06/28
- [PATCH v4 16/45] target/arm: Handle SME in sve_access_check, Richard Henderson, 2022/06/28
- [PATCH v4 17/45] target/arm: Implement SME RDSVL, ADDSVL, ADDSPL, Richard Henderson, 2022/06/28
- [PATCH v4 19/45] target/arm: Implement SME MOVA, Richard Henderson, 2022/06/28
- [PATCH v4 18/45] target/arm: Implement SME ZERO, Richard Henderson, 2022/06/28
- [PATCH v4 20/45] target/arm: Implement SME LD1, ST1, Richard Henderson, 2022/06/28
- [PATCH v4 21/45] target/arm: Export unpredicated ld/st from translate-sve.c, Richard Henderson, 2022/06/28
- [PATCH v4 23/45] target/arm: Implement SME ADDHA, ADDVA, Richard Henderson, 2022/06/28
- [PATCH v4 25/45] target/arm: Implement BFMOPA, BFMOPS,
Richard Henderson <=
- [PATCH v4 22/45] target/arm: Implement SME LDR, STR, Richard Henderson, 2022/06/28
- [PATCH v4 24/45] target/arm: Implement FMOPA, FMOPS (non-widening), Richard Henderson, 2022/06/28
- [PATCH v4 26/45] target/arm: Implement FMOPA, FMOPS (widening), Richard Henderson, 2022/06/28
- [PATCH v4 27/45] target/arm: Implement SME integer outer product, Richard Henderson, 2022/06/28
- [PATCH v4 28/45] target/arm: Implement PSEL, Richard Henderson, 2022/06/28
- [PATCH v4 29/45] target/arm: Implement REVD, Richard Henderson, 2022/06/28
- [PATCH v4 30/45] target/arm: Implement SCLAMP, UCLAMP, Richard Henderson, 2022/06/28
- [PATCH v4 31/45] target/arm: Reset streaming sve state on exception boundaries, Richard Henderson, 2022/06/28
- [PATCH v4 32/45] target/arm: Enable SME for -cpu max, Richard Henderson, 2022/06/28
- [PATCH v4 33/45] linux-user/aarch64: Clear tpidr2_el0 if CLONE_SETTLS, Richard Henderson, 2022/06/28