[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PATCH v2 10/37] target/riscv: SIMD 8-bit Multiply Instructions
From: |
LIU Zhiwei |
Subject: |
[PATCH v2 10/37] target/riscv: SIMD 8-bit Multiply Instructions |
Date: |
Thu, 10 Jun 2021 15:58:41 +0800 |
There are 6 instructions, including 8-bit signed or unsigned multiply,
8-bit signed or unsigned crossed multiply, Q7 signed or signed crossed
saturating multiply.
Signed-off-by: LIU Zhiwei <zhiwei_liu@c-sky.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
---
target/riscv/helper.h | 7 ++
target/riscv/insn32.decode | 7 ++
target/riscv/insn_trans/trans_rvp.c.inc | 8 +++
target/riscv/packed_helper.c | 93 +++++++++++++++++++++++++
4 files changed, 115 insertions(+)
diff --git a/target/riscv/helper.h b/target/riscv/helper.h
index d13b84f165..4d0918b9a9 100644
--- a/target/riscv/helper.h
+++ b/target/riscv/helper.h
@@ -1225,3 +1225,10 @@ DEF_HELPER_3(umul16, i64, env, tl, tl)
DEF_HELPER_3(umulx16, i64, env, tl, tl)
DEF_HELPER_3(khm16, tl, env, tl, tl)
DEF_HELPER_3(khmx16, tl, env, tl, tl)
+
+DEF_HELPER_3(smul8, i64, env, tl, tl)
+DEF_HELPER_3(smulx8, i64, env, tl, tl)
+DEF_HELPER_3(umul8, i64, env, tl, tl)
+DEF_HELPER_3(umulx8, i64, env, tl, tl)
+DEF_HELPER_3(khm8, tl, env, tl, tl)
+DEF_HELPER_3(khmx8, tl, env, tl, tl)
diff --git a/target/riscv/insn32.decode b/target/riscv/insn32.decode
index cbee995229..05c3e67477 100644
--- a/target/riscv/insn32.decode
+++ b/target/riscv/insn32.decode
@@ -828,3 +828,10 @@ umul16 1011000 ..... ..... 000 ..... 1110111 @r
umulx16 1011001 ..... ..... 000 ..... 1110111 @r
khm16 1000011 ..... ..... 000 ..... 1110111 @r
khmx16 1001011 ..... ..... 000 ..... 1110111 @r
+
+smul8 1010100 ..... ..... 000 ..... 1110111 @r
+smulx8 1010101 ..... ..... 000 ..... 1110111 @r
+umul8 1011100 ..... ..... 000 ..... 1110111 @r
+umulx8 1011101 ..... ..... 000 ..... 1110111 @r
+khm8 1000111 ..... ..... 000 ..... 1110111 @r
+khmx8 1001111 ..... ..... 000 ..... 1110111 @r
diff --git a/target/riscv/insn_trans/trans_rvp.c.inc
b/target/riscv/insn_trans/trans_rvp.c.inc
index b93ba63dd8..2188de8505 100644
--- a/target/riscv/insn_trans/trans_rvp.c.inc
+++ b/target/riscv/insn_trans/trans_rvp.c.inc
@@ -286,3 +286,11 @@ GEN_RVP_R_D64_OOL(umul16);
GEN_RVP_R_D64_OOL(umulx16);
GEN_RVP_R_OOL(khm16);
GEN_RVP_R_OOL(khmx16);
+
+/* SIMD 8-bit Multiply Instructions */
+GEN_RVP_R_D64_OOL(smul8);
+GEN_RVP_R_D64_OOL(smulx8);
+GEN_RVP_R_D64_OOL(umul8);
+GEN_RVP_R_D64_OOL(umulx8);
+GEN_RVP_R_OOL(khm8);
+GEN_RVP_R_OOL(khmx8);
diff --git a/target/riscv/packed_helper.c b/target/riscv/packed_helper.c
index 13fed2c4d1..56baefeb8e 100644
--- a/target/riscv/packed_helper.c
+++ b/target/riscv/packed_helper.c
@@ -827,3 +827,96 @@ static inline void do_khmx16(CPURISCVState *env, void *vd,
void *va,
}
RVPR(khmx16, 2, 2);
+
+/* SIMD 8-bit Multiply Instructions */
+static inline void do_smul8(CPURISCVState *env, void *vd, void *va, void *vb)
+{
+ int16_t *d = vd;
+ int8_t *a = va, *b = vb;
+ d[H2(0)] = (int16_t)a[H1(0)] * b[H1(0)];
+ d[H2(1)] = (int16_t)a[H1(1)] * b[H1(1)];
+ d[H2(2)] = (int16_t)a[H1(2)] * b[H1(2)];
+ d[H2(3)] = (int16_t)a[H1(3)] * b[H1(3)];
+}
+
+RVPR64(smul8);
+
+static inline void do_smulx8(CPURISCVState *env, void *vd, void *va, void *vb)
+{
+ int16_t *d = vd;
+ int8_t *a = va, *b = vb;
+ d[H2(0)] = (int16_t)a[H1(0)] * b[H1(1)];
+ d[H2(1)] = (int16_t)a[H1(1)] * b[H1(0)];
+ d[H2(2)] = (int16_t)a[H1(2)] * b[H1(3)];
+ d[H2(3)] = (int16_t)a[H1(3)] * b[H1(2)];
+}
+
+RVPR64(smulx8);
+
+static inline void do_umul8(CPURISCVState *env, void *vd, void *va, void *vb)
+{
+ uint16_t *d = vd;
+ uint8_t *a = va, *b = vb;
+ d[H2(0)] = (uint16_t)a[H1(0)] * b[H1(0)];
+ d[H2(1)] = (uint16_t)a[H1(1)] * b[H1(1)];
+ d[H2(2)] = (uint16_t)a[H1(2)] * b[H1(2)];
+ d[H2(3)] = (uint16_t)a[H1(3)] * b[H1(3)];
+}
+
+RVPR64(umul8);
+
+static inline void do_umulx8(CPURISCVState *env, void *vd, void *va, void *vb)
+{
+ uint16_t *d = vd;
+ uint8_t *a = va, *b = vb;
+ d[H2(0)] = (uint16_t)a[H1(0)] * b[H1(1)];
+ d[H2(1)] = (uint16_t)a[H1(1)] * b[H1(0)];
+ d[H2(2)] = (uint16_t)a[H1(2)] * b[H1(3)];
+ d[H2(3)] = (uint16_t)a[H1(3)] * b[H1(2)];
+}
+
+RVPR64(umulx8);
+
+static inline void do_khm8(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int8_t *d = vd, *a = va, *b = vb;
+
+ if (a[i] == INT8_MIN && b[i] == INT8_MIN) {
+ env->vxsat = 1;
+ d[i] = INT8_MAX;
+ } else {
+ d[i] = (int16_t)a[i] * b[i] >> 7;
+ }
+}
+
+RVPR(khm8, 1, 1);
+
+static inline void do_khmx8(CPURISCVState *env, void *vd, void *va,
+ void *vb, uint8_t i)
+{
+ int8_t *d = vd, *a = va, *b = vb;
+ /*
+ * t[x] = ra.B[x] s* rb.B[y];
+ * rt.B[x] = SAT.Q7(t[x] s>> 7);
+ *
+ * (RV32: (x,y)=(3,2),(2,3),
+ * (1,0),(0,1),
+ * (RV64: (x,y)=(7,6),(6,7),(5,4),(4,5),
+ * (3,2),(2,3),(1,0),(0,1))
+ */
+ if (a[H1(i)] == INT8_MIN && b[H1(i + 1)] == INT8_MIN) {
+ env->vxsat = 1;
+ d[H1(i)] = INT8_MAX;
+ } else {
+ d[H1(i)] = (int16_t)a[H1(i)] * b[H1(i + 1)] >> 7;
+ }
+ if (a[H1(i + 1)] == INT8_MIN && b[H1(i)] == INT8_MIN) {
+ env->vxsat = 1;
+ d[H1(i + 1)] = INT8_MAX;
+ } else {
+ d[H1(i + 1)] = (int16_t)a[H1(i + 1)] * b[H1(i)] >> 7;
+ }
+}
+
+RVPR(khmx8, 2, 1);
--
2.25.1
- [PATCH v2 00/37] target/riscv: support packed extension v0.9.4, LIU Zhiwei, 2021/06/10
- [PATCH v2 01/37] target/riscv: implementation-defined constant parameters, LIU Zhiwei, 2021/06/10
- [PATCH v2 02/37] target/riscv: Make the vector helper functions public, LIU Zhiwei, 2021/06/10
- [PATCH v2 03/37] target/riscv: 16-bit Addition & Subtraction Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 04/37] target/riscv: 8-bit Addition & Subtraction Instruction, LIU Zhiwei, 2021/06/10
- [PATCH v2 05/37] target/riscv: SIMD 16-bit Shift Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 06/37] target/riscv: SIMD 8-bit Shift Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 07/37] target/riscv: SIMD 16-bit Compare Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 08/37] target/riscv: SIMD 8-bit Compare Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 09/37] target/riscv: SIMD 16-bit Multiply Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 10/37] target/riscv: SIMD 8-bit Multiply Instructions,
LIU Zhiwei <=
- [PATCH v2 11/37] target/riscv: SIMD 16-bit Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 12/37] target/riscv: SIMD 8-bit Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 13/37] target/riscv: 8-bit Unpacking Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 14/37] target/riscv: 16-bit Packing Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 15/37] target/riscv: Signed MSW 32x32 Multiply and Add Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 16/37] target/riscv: Signed MSW 32x16 Multiply and Add Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 17/37] target/riscv: Signed 16-bit Multiply 32-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 18/37] target/riscv: Signed 16-bit Multiply 64-bit Add/Subtract Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 19/37] target/riscv: Partial-SIMD Miscellaneous Instructions, LIU Zhiwei, 2021/06/10
- [PATCH v2 20/37] target/riscv: 8-bit Multiply with 32-bit Add Instructions, LIU Zhiwei, 2021/06/10