[fix] v{max,min}{ph,sh} supports not {er} but {sae}

herumi · Oct 8, 2024 · d993342 · d993342
1 parent b6d2b90
commit d993342
Show file tree

Hide file tree

Showing 3 changed files with 19 additions and 12 deletions.
diff --git a/gen/gen_avx512.cpp b/gen/gen_avx512.cpp
@@ -835,18 +835,20 @@ void putFP16_1()
 	const struct Tbl {
 		uint8_t code;
 		const char *name;
+		int mode;
 	} tbl[] = {
-		{ 0x58, "add" },
-		{ 0x5C, "sub" },
-		{ 0x59, "mul" },
-		{ 0x5E, "div" },
-		{ 0x5F, "max" },
-		{ 0x5D, "min" },
+		{ 0x58, "add", 0 },
+		{ 0x5C, "sub", 0 },
+		{ 0x59, "mul", 0 },
+		{ 0x5E, "div", 0 },
+		{ 0x5F, "max", 1 },
+		{ 0x5D, "min", 1 },
 	};
+	const char *erTbl[] = { "ER", "SAE" };
 	for (size_t i = 0; i < NUM_OF_ARRAY(tbl); i++) {
 		const Tbl *p = &tbl[i];
-		printf("void v%sph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x%02X); }\n", p->name, p->code);
-		printf("void v%ssh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x%02X); }\n", p->name, p->code);
+		printf("void v%sph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_%s_Z | T_B16, 0x%02X); }\n", p->name, erTbl[p->mode], p->code);
+		printf("void v%ssh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_%s_X | T_N2, 0x%02X); }\n", p->name, erTbl[p->mode], p->code);
 	}
 }
 

diff --git a/test/make_nm.cpp b/test/make_nm.cpp
@@ -3487,6 +3487,10 @@ class Test {
 		put("vcvtpd2dq", YMM_KZ, _ZMM | ZMM_ER | M_1to8);
 #endif
 	}
+	void put512_fp16()
+	{
+		put("vaddph", ZMM, ZMM, ZMM_ER);
+	}
 	void putMin()
 	{
 #ifdef XBYAK64
@@ -3526,6 +3530,7 @@ class Test {
 		put512_AVX1();
 		separateFunc();
 		put512_cvt();
+		put512_fp16();
 #endif
 	}
 #endif

diff --git a/xbyak/xbyak_mnemonic.h b/xbyak/xbyak_mnemonic.h
@@ -2351,10 +2351,10 @@ void vinserti32x4(const Ymm& r1, const Ymm& r2, const Operand& op, uint8_t imm)
 void vinserti32x8(const Zmm& r1, const Zmm& r2, const Operand& op, uint8_t imm) {if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N32|T_66|T_0F3A|T_EW0|T_YMM|T_MUST_EVEX, 0x3A, imm); }
 void vinserti64x2(const Ymm& r1, const Ymm& r2, const Operand& op, uint8_t imm) {if (!(r1.getKind() == r2.getKind() && op.is(Operand::MEM | Operand::XMM))) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N16|T_66|T_0F3A|T_EW1|T_YMM|T_MUST_EVEX, 0x38, imm); }
 void vinserti64x4(const Zmm& r1, const Zmm& r2, const Operand& op, uint8_t imm) {if (!op.is(Operand::MEM | Operand::YMM)) XBYAK_THROW(ERR_BAD_COMBINATION) opVex(r1, &r2, op, T_N32|T_66|T_0F3A|T_EW1|T_YMM|T_MUST_EVEX, 0x3A, imm); }
-void vmaxph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5F); }
-void vmaxsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5F); }
-void vminph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_ER_Z | T_B16, 0x5D); }
-void vminsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_ER_X | T_N2, 0x5D); }
+void vmaxph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_SAE_Z | T_B16, 0x5F); }
+void vmaxsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_SAE_X | T_N2, 0x5F); }
+void vminph(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_EW0 | T_YMM | T_MUST_EVEX | T_SAE_Z | T_B16, 0x5D); }
+void vminsh(const Xmm& xmm, const Operand& op1, const Operand& op2 = Operand()) { opAVX_X_X_XM(xmm, op1, op2, T_MAP5 | T_F3 | T_EW0 | T_MUST_EVEX | T_SAE_X | T_N2, 0x5D); }
 void vmovdqa32(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_66|T_0F|T_EW0|T_YMM|T_ER_X|T_ER_Y|T_ER_Z|T_MUST_EVEX|T_M_K, 0x7F); }
 void vmovdqa32(const Xmm& x, const Operand& op) { opAVX_X_XM_IMM(x, op, T_66|T_0F|T_EW0|T_YMM|T_ER_X|T_ER_Y|T_ER_Z|T_MUST_EVEX, 0x6F); }
 void vmovdqa64(const Address& addr, const Xmm& x) { opAVX_X_XM_IMM(x, addr, T_66|T_0F|T_EW1|T_YMM|T_ER_X|T_ER_Y|T_ER_Z|T_MUST_EVEX|T_M_K, 0x7F); }