Skip to content

Commit cbed714

Browse files
authored
[AMDGPU][True16][MC] test update for v_add/sub_f16 in true16 (#118926)
This is a NFC change. Update mc test for v_add/sub_f16 in true16 format. MC source change was done by previous patch and automatically enabled by t16 pesudo
1 parent 342fa15 commit cbed714

23 files changed

+1263
-723
lines changed

llvm/lib/Target/AMDGPU/VOP2Instructions.td

Lines changed: 2 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -1818,10 +1818,8 @@ defm V_CVT_PK_RTZ_F16_F32 : VOP2_Real_FULL_with_name_gfx11_gfx12<0x02f,
18181818
"V_CVT_PKRTZ_F16_F32", "v_cvt_pk_rtz_f16_f32">;
18191819
defm V_PK_FMAC_F16 : VOP2_Real_e32_gfx11_gfx12<0x03c>;
18201820

1821-
defm V_ADD_F16_t16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x032, "v_add_f16">;
1822-
defm V_ADD_F16_fake16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x032, "v_add_f16">;
1823-
defm V_SUB_F16_t16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x033, "v_sub_f16">;
1824-
defm V_SUB_F16_fake16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x033, "v_sub_f16">;
1821+
defm V_ADD_F16 : VOP2_Real_FULL_t16_and_fake16_gfx11_gfx12<0x032, "v_add_f16">;
1822+
defm V_SUB_F16 : VOP2_Real_FULL_t16_and_fake16_gfx11_gfx12<0x033, "v_sub_f16">;
18251823
defm V_SUBREV_F16_t16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x034, "v_subrev_f16">;
18261824
defm V_SUBREV_F16_fake16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x034, "v_subrev_f16">;
18271825
defm V_MUL_F16_t16 : VOP2_Real_FULL_t16_gfx11_gfx12<0x035, "v_mul_f16">;

llvm/test/MC/AMDGPU/gfx11_asm_vop2.s

Lines changed: 90 additions & 60 deletions
Original file line numberDiff line numberDiff line change
@@ -124,50 +124,65 @@ v_add_co_ci_u32 v255, vcc, 0xaf123456, v255, vcc
124124
// W64: v_add_co_ci_u32_e32 v255, vcc, 0xaf123456, v255, vcc ; encoding: [0xff,0xfe,0xff,0x41,0x56,0x34,0x12,0xaf]
125125
// W32-ERR: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
126126

127-
v_add_f16 v5, v1, v2
128-
// GFX11: v_add_f16_e32 v5, v1, v2 ; encoding: [0x01,0x05,0x0a,0x64]
127+
v_add_f16 v5.l, v1.l, v2.l
128+
// GFX11: v_add_f16_e32 v5.l, v1.l, v2.l ; encoding: [0x01,0x05,0x0a,0x64]
129129

130-
v_add_f16 v5, v127, v2
131-
// GFX11: v_add_f16_e32 v5, v127, v2 ; encoding: [0x7f,0x05,0x0a,0x64]
130+
v_add_f16 v5.l, v127.l, v2.l
131+
// GFX11: v_add_f16_e32 v5.l, v127.l, v2.l ; encoding: [0x7f,0x05,0x0a,0x64]
132132

133-
v_add_f16 v5, s1, v2
134-
// GFX11: v_add_f16_e32 v5, s1, v2 ; encoding: [0x01,0x04,0x0a,0x64]
133+
v_add_f16 v5.l, s1, v2.l
134+
// GFX11: v_add_f16_e32 v5.l, s1, v2.l ; encoding: [0x01,0x04,0x0a,0x64]
135135

136-
v_add_f16 v5, s105, v2
137-
// GFX11: v_add_f16_e32 v5, s105, v2 ; encoding: [0x69,0x04,0x0a,0x64]
136+
v_add_f16 v5.l, s105, v2.l
137+
// GFX11: v_add_f16_e32 v5.l, s105, v2.l ; encoding: [0x69,0x04,0x0a,0x64]
138138

139-
v_add_f16 v5, vcc_lo, v2
140-
// GFX11: v_add_f16_e32 v5, vcc_lo, v2 ; encoding: [0x6a,0x04,0x0a,0x64]
139+
v_add_f16 v5.l, vcc_lo, v2.l
140+
// GFX11: v_add_f16_e32 v5.l, vcc_lo, v2.l ; encoding: [0x6a,0x04,0x0a,0x64]
141141

142-
v_add_f16 v5, vcc_hi, v2
143-
// GFX11: v_add_f16_e32 v5, vcc_hi, v2 ; encoding: [0x6b,0x04,0x0a,0x64]
142+
v_add_f16 v5.l, vcc_hi, v2.l
143+
// GFX11: v_add_f16_e32 v5.l, vcc_hi, v2.l ; encoding: [0x6b,0x04,0x0a,0x64]
144144

145-
v_add_f16 v5, ttmp15, v2
146-
// GFX11: v_add_f16_e32 v5, ttmp15, v2 ; encoding: [0x7b,0x04,0x0a,0x64]
145+
v_add_f16 v5.l, ttmp15, v2.l
146+
// GFX11: v_add_f16_e32 v5.l, ttmp15, v2.l ; encoding: [0x7b,0x04,0x0a,0x64]
147147

148-
v_add_f16 v5, m0, v2
149-
// GFX11: v_add_f16_e32 v5, m0, v2 ; encoding: [0x7d,0x04,0x0a,0x64]
148+
v_add_f16 v5.l, m0, v2.l
149+
// GFX11: v_add_f16_e32 v5.l, m0, v2.l ; encoding: [0x7d,0x04,0x0a,0x64]
150150

151-
v_add_f16 v5, exec_lo, v2
152-
// GFX11: v_add_f16_e32 v5, exec_lo, v2 ; encoding: [0x7e,0x04,0x0a,0x64]
151+
v_add_f16 v5.l, exec_lo, v2.l
152+
// GFX11: v_add_f16_e32 v5.l, exec_lo, v2.l ; encoding: [0x7e,0x04,0x0a,0x64]
153153

154-
v_add_f16 v5, exec_hi, v2
155-
// GFX11: v_add_f16_e32 v5, exec_hi, v2 ; encoding: [0x7f,0x04,0x0a,0x64]
154+
v_add_f16 v5.l, exec_hi, v2.l
155+
// GFX11: v_add_f16_e32 v5.l, exec_hi, v2.l ; encoding: [0x7f,0x04,0x0a,0x64]
156156

157-
v_add_f16 v5, null, v2
158-
// GFX11: v_add_f16_e32 v5, null, v2 ; encoding: [0x7c,0x04,0x0a,0x64]
157+
v_add_f16 v5.l, null, v2.l
158+
// GFX11: v_add_f16_e32 v5.l, null, v2.l ; encoding: [0x7c,0x04,0x0a,0x64]
159159

160-
v_add_f16 v5, -1, v2
161-
// GFX11: v_add_f16_e32 v5, -1, v2 ; encoding: [0xc1,0x04,0x0a,0x64]
160+
v_add_f16 v5.l, -1, v2.l
161+
// GFX11: v_add_f16_e32 v5.l, -1, v2.l ; encoding: [0xc1,0x04,0x0a,0x64]
162162

163-
v_add_f16 v5, 0.5, v2
164-
// GFX11: v_add_f16_e32 v5, 0.5, v2 ; encoding: [0xf0,0x04,0x0a,0x64]
163+
v_add_f16 v5.l, 0.5, v2.l
164+
// GFX11: v_add_f16_e32 v5.l, 0.5, v2.l ; encoding: [0xf0,0x04,0x0a,0x64]
165165

166-
v_add_f16 v5, src_scc, v2
167-
// GFX11: v_add_f16_e32 v5, src_scc, v2 ; encoding: [0xfd,0x04,0x0a,0x64]
166+
v_add_f16 v5.l, src_scc, v2.l
167+
// GFX11: v_add_f16_e32 v5.l, src_scc, v2.l ; encoding: [0xfd,0x04,0x0a,0x64]
168168

169-
v_add_f16 v127, 0xfe0b, v127
170-
// GFX11: v_add_f16_e32 v127, 0xfe0b, v127 ; encoding: [0xff,0xfe,0xfe,0x64,0x0b,0xfe,0x00,0x00]
169+
v_add_f16 v127.l, 0xfe0b, v127.l
170+
// GFX11: v_add_f16_e32 v127.l, 0xfe0b, v127.l ; encoding: [0xff,0xfe,0xfe,0x64,0x0b,0xfe,0x00,0x00]
171+
172+
v_add_f16 v5.l, v1.h, v2.l
173+
// GFX11: v_add_f16_e32 v5.l, v1.h, v2.l ; encoding: [0x81,0x05,0x0a,0x64]
174+
175+
v_add_f16 v5.l, v127.h, v2.l
176+
// GFX11: v_add_f16_e32 v5.l, v127.h, v2.l ; encoding: [0xff,0x05,0x0a,0x64]
177+
178+
v_add_f16 v127.l, 0.5, v127.l
179+
// GFX11: v_add_f16_e32 v127.l, 0.5, v127.l ; encoding: [0xf0,0xfe,0xfe,0x64]
180+
181+
v_add_f16 v5.h, src_scc, v2.h
182+
// GFX11: v_add_f16_e32 v5.h, src_scc, v2.h ; encoding: [0xfd,0x04,0x0b,0x65]
183+
184+
v_add_f16 v127.h, 0xfe0b, v127.h
185+
// GFX11: v_add_f16_e32 v127.h, 0xfe0b, v127.h ; encoding: [0xff,0xfe,0xff,0x65,0x0b,0xfe,0x00,0x00]
171186

172187
v_add_f32 v5, v1, v2
173188
// GFX11: v_add_f32_e32 v5, v1, v2 ; encoding: [0x01,0x05,0x0a,0x06]
@@ -2092,50 +2107,65 @@ v_sub_co_ci_u32 v255, vcc, 0xaf123456, v255, vcc
20922107
// W64: v_sub_co_ci_u32_e32 v255, vcc, 0xaf123456, v255, vcc ; encoding: [0xff,0xfe,0xff,0x43,0x56,0x34,0x12,0xaf]
20932108
// W32-ERR: :[[@LINE-2]]:1: error: operands are not valid for this GPU or mode
20942109

2095-
v_sub_f16 v5, v1, v2
2096-
// GFX11: v_sub_f16_e32 v5, v1, v2 ; encoding: [0x01,0x05,0x0a,0x66]
2110+
v_sub_f16 v5.l, v1.l, v2.l
2111+
// GFX11: v_sub_f16_e32 v5.l, v1.l, v2.l ; encoding: [0x01,0x05,0x0a,0x66]
2112+
2113+
v_sub_f16 v5.l, v127.l, v2.l
2114+
// GFX11: v_sub_f16_e32 v5.l, v127.l, v2.l ; encoding: [0x7f,0x05,0x0a,0x66]
2115+
2116+
v_sub_f16 v5.l, s1, v2.l
2117+
// GFX11: v_sub_f16_e32 v5.l, s1, v2.l ; encoding: [0x01,0x04,0x0a,0x66]
2118+
2119+
v_sub_f16 v5.l, s105, v2.l
2120+
// GFX11: v_sub_f16_e32 v5.l, s105, v2.l ; encoding: [0x69,0x04,0x0a,0x66]
2121+
2122+
v_sub_f16 v5.l, vcc_lo, v2.l
2123+
// GFX11: v_sub_f16_e32 v5.l, vcc_lo, v2.l ; encoding: [0x6a,0x04,0x0a,0x66]
2124+
2125+
v_sub_f16 v5.l, vcc_hi, v2.l
2126+
// GFX11: v_sub_f16_e32 v5.l, vcc_hi, v2.l ; encoding: [0x6b,0x04,0x0a,0x66]
20972127

2098-
v_sub_f16 v5, v127, v2
2099-
// GFX11: v_sub_f16_e32 v5, v127, v2 ; encoding: [0x7f,0x05,0x0a,0x66]
2128+
v_sub_f16 v5.l, ttmp15, v2.l
2129+
// GFX11: v_sub_f16_e32 v5.l, ttmp15, v2.l ; encoding: [0x7b,0x04,0x0a,0x66]
21002130

2101-
v_sub_f16 v5, s1, v2
2102-
// GFX11: v_sub_f16_e32 v5, s1, v2 ; encoding: [0x01,0x04,0x0a,0x66]
2131+
v_sub_f16 v5.l, m0, v2.l
2132+
// GFX11: v_sub_f16_e32 v5.l, m0, v2.l ; encoding: [0x7d,0x04,0x0a,0x66]
21032133

2104-
v_sub_f16 v5, s105, v2
2105-
// GFX11: v_sub_f16_e32 v5, s105, v2 ; encoding: [0x69,0x04,0x0a,0x66]
2134+
v_sub_f16 v5.l, exec_lo, v2.l
2135+
// GFX11: v_sub_f16_e32 v5.l, exec_lo, v2.l ; encoding: [0x7e,0x04,0x0a,0x66]
21062136

2107-
v_sub_f16 v5, vcc_lo, v2
2108-
// GFX11: v_sub_f16_e32 v5, vcc_lo, v2 ; encoding: [0x6a,0x04,0x0a,0x66]
2137+
v_sub_f16 v5.l, exec_hi, v2.l
2138+
// GFX11: v_sub_f16_e32 v5.l, exec_hi, v2.l ; encoding: [0x7f,0x04,0x0a,0x66]
21092139

2110-
v_sub_f16 v5, vcc_hi, v2
2111-
// GFX11: v_sub_f16_e32 v5, vcc_hi, v2 ; encoding: [0x6b,0x04,0x0a,0x66]
2140+
v_sub_f16 v5.l, null, v2.l
2141+
// GFX11: v_sub_f16_e32 v5.l, null, v2.l ; encoding: [0x7c,0x04,0x0a,0x66]
21122142

2113-
v_sub_f16 v5, ttmp15, v2
2114-
// GFX11: v_sub_f16_e32 v5, ttmp15, v2 ; encoding: [0x7b,0x04,0x0a,0x66]
2143+
v_sub_f16 v5.l, -1, v2.l
2144+
// GFX11: v_sub_f16_e32 v5.l, -1, v2.l ; encoding: [0xc1,0x04,0x0a,0x66]
21152145

2116-
v_sub_f16 v5, m0, v2
2117-
// GFX11: v_sub_f16_e32 v5, m0, v2 ; encoding: [0x7d,0x04,0x0a,0x66]
2146+
v_sub_f16 v5.l, 0.5, v2.l
2147+
// GFX11: v_sub_f16_e32 v5.l, 0.5, v2.l ; encoding: [0xf0,0x04,0x0a,0x66]
21182148

2119-
v_sub_f16 v5, exec_lo, v2
2120-
// GFX11: v_sub_f16_e32 v5, exec_lo, v2 ; encoding: [0x7e,0x04,0x0a,0x66]
2149+
v_sub_f16 v5.l, src_scc, v2.l
2150+
// GFX11: v_sub_f16_e32 v5.l, src_scc, v2.l ; encoding: [0xfd,0x04,0x0a,0x66]
21212151

2122-
v_sub_f16 v5, exec_hi, v2
2123-
// GFX11: v_sub_f16_e32 v5, exec_hi, v2 ; encoding: [0x7f,0x04,0x0a,0x66]
2152+
v_sub_f16 v127.l, 0xfe0b, v127.l
2153+
// GFX11: v_sub_f16_e32 v127.l, 0xfe0b, v127.l ; encoding: [0xff,0xfe,0xfe,0x66,0x0b,0xfe,0x00,0x00]
21242154

2125-
v_sub_f16 v5, null, v2
2126-
// GFX11: v_sub_f16_e32 v5, null, v2 ; encoding: [0x7c,0x04,0x0a,0x66]
2155+
v_sub_f16 v5.l, v1.h, v2.l
2156+
// GFX11: v_sub_f16_e32 v5.l, v1.h, v2.l ; encoding: [0x81,0x05,0x0a,0x66]
21272157

2128-
v_sub_f16 v5, -1, v2
2129-
// GFX11: v_sub_f16_e32 v5, -1, v2 ; encoding: [0xc1,0x04,0x0a,0x66]
2158+
v_sub_f16 v5.l, v127.h, v2.l
2159+
// GFX11: v_sub_f16_e32 v5.l, v127.h, v2.l ; encoding: [0xff,0x05,0x0a,0x66]
21302160

2131-
v_sub_f16 v5, 0.5, v2
2132-
// GFX11: v_sub_f16_e32 v5, 0.5, v2 ; encoding: [0xf0,0x04,0x0a,0x66]
2161+
v_sub_f16 v127.l, 0.5, v127.l
2162+
// GFX11: v_sub_f16_e32 v127.l, 0.5, v127.l ; encoding: [0xf0,0xfe,0xfe,0x66]
21332163

2134-
v_sub_f16 v5, src_scc, v2
2135-
// GFX11: v_sub_f16_e32 v5, src_scc, v2 ; encoding: [0xfd,0x04,0x0a,0x66]
2164+
v_sub_f16 v5.h, src_scc, v2.h
2165+
// GFX11: v_sub_f16_e32 v5.h, src_scc, v2.h ; encoding: [0xfd,0x04,0x0b,0x67]
21362166

2137-
v_sub_f16 v127, 0xfe0b, v127
2138-
// GFX11: v_sub_f16_e32 v127, 0xfe0b, v127 ; encoding: [0xff,0xfe,0xfe,0x66,0x0b,0xfe,0x00,0x00]
2167+
v_sub_f16 v127.h, 0xfe0b, v127.h
2168+
// GFX11: v_sub_f16_e32 v127.h, 0xfe0b, v127.h ; encoding: [0xff,0xfe,0xff,0x67,0x0b,0xfe,0x00,0x00]
21392169

21402170
v_sub_f32 v5, v1, v2
21412171
// GFX11: v_sub_f32_e32 v5, v1, v2 ; encoding: [0x01,0x05,0x0a,0x08]

0 commit comments

Comments
 (0)