summaryrefslogtreecommitdiff
path: root/lib/Target/R600
diff options
context:
space:
mode:
Diffstat (limited to 'lib/Target/R600')
-rw-r--r--lib/Target/R600/AMDGPUInstructions.td124
-rw-r--r--lib/Target/R600/R600Instructions.td42
-rw-r--r--lib/Target/R600/SIInstructions.td52
3 files changed, 104 insertions, 114 deletions
diff --git a/lib/Target/R600/AMDGPUInstructions.td b/lib/Target/R600/AMDGPUInstructions.td
index 4b37a53687d..8ab0946f47d 100644
--- a/lib/Target/R600/AMDGPUInstructions.td
+++ b/lib/Target/R600/AMDGPUInstructions.td
@@ -116,21 +116,21 @@ class CLAMP <RegisterClass rc> : AMDGPUShaderInst <
(outs rc:$dst),
(ins rc:$src0),
"CLAMP $dst, $src0",
- [(set rc:$dst, (int_AMDIL_clamp rc:$src0, (f32 FP_ZERO), (f32 FP_ONE)))]
+ [(set f32:$dst, (int_AMDIL_clamp f32:$src0, (f32 FP_ZERO), (f32 FP_ONE)))]
>;
class FABS <RegisterClass rc> : AMDGPUShaderInst <
(outs rc:$dst),
(ins rc:$src0),
"FABS $dst, $src0",
- [(set rc:$dst, (fabs rc:$src0))]
+ [(set f32:$dst, (fabs f32:$src0))]
>;
class FNEG <RegisterClass rc> : AMDGPUShaderInst <
(outs rc:$dst),
(ins rc:$src0),
"FNEG $dst, $src0",
- [(set rc:$dst, (fneg rc:$src0))]
+ [(set f32:$dst, (fneg f32:$src0))]
>;
} // usesCustomInserter = 1
@@ -141,8 +141,7 @@ multiclass RegisterLoadStore <RegisterClass dstClass, Operand addrClass,
(outs dstClass:$dst),
(ins addrClass:$addr, i32imm:$chan),
"RegisterLoad $dst, $addr",
- [(set (i32 dstClass:$dst), (AMDGPUregister_load addrPat:$addr,
- (i32 timm:$chan)))]
+ [(set i32:$dst, (AMDGPUregister_load addrPat:$addr, (i32 timm:$chan)))]
> {
let isRegisterLoad = 1;
}
@@ -151,7 +150,7 @@ multiclass RegisterLoadStore <RegisterClass dstClass, Operand addrClass,
(outs),
(ins dstClass:$val, addrClass:$addr, i32imm:$chan),
"RegisterStore $val, $addr",
- [(AMDGPUregister_store (i32 dstClass:$val), addrPat:$addr, (i32 timm:$chan))]
+ [(AMDGPUregister_store i32:$val, addrPat:$addr, (i32 timm:$chan))]
> {
let isRegisterStore = 1;
}
@@ -162,92 +161,83 @@ multiclass RegisterLoadStore <RegisterClass dstClass, Operand addrClass,
/* Generic helper patterns for intrinsics */
/* -------------------------------------- */
-class POW_Common <AMDGPUInst log_ieee, AMDGPUInst exp_ieee, AMDGPUInst mul,
- RegisterClass rc> : Pat <
- (fpow rc:$src0, rc:$src1),
- (exp_ieee (mul rc:$src1, (log_ieee rc:$src0)))
+class POW_Common <AMDGPUInst log_ieee, AMDGPUInst exp_ieee, AMDGPUInst mul>
+ : Pat <
+ (fpow f32:$src0, f32:$src1),
+ (exp_ieee (mul f32:$src1, (log_ieee f32:$src0)))
>;
/* Other helper patterns */
/* --------------------- */
/* Extract element pattern */
-class Extract_Element <ValueType sub_type, ValueType vec_type,
- RegisterClass vec_class, int sub_idx,
- SubRegIndex sub_reg>: Pat<
- (sub_type (vector_extract (vec_type vec_class:$src), sub_idx)),
- (EXTRACT_SUBREG vec_class:$src, sub_reg)
+class Extract_Element <ValueType sub_type, ValueType vec_type, int sub_idx,
+ SubRegIndex sub_reg>
+ : Pat<
+ (sub_type (vector_extract vec_type:$src, sub_idx)),
+ (EXTRACT_SUBREG $src, sub_reg)
>;
/* Insert element pattern */
class Insert_Element <ValueType elem_type, ValueType vec_type,
- RegisterClass elem_class, RegisterClass vec_class,
- int sub_idx, SubRegIndex sub_reg> : Pat <
-
- (vec_type (vector_insert (vec_type vec_class:$vec),
- (elem_type elem_class:$elem), sub_idx)),
- (INSERT_SUBREG vec_class:$vec, elem_class:$elem, sub_reg)
+ int sub_idx, SubRegIndex sub_reg>
+ : Pat <
+ (vector_insert vec_type:$vec, elem_type:$elem, sub_idx),
+ (INSERT_SUBREG $vec, $elem, sub_reg)
>;
// Vector Build pattern
-class Vector1_Build <ValueType vecType, RegisterClass vectorClass,
- ValueType elemType, RegisterClass elemClass> : Pat <
- (vecType (build_vector (elemType elemClass:$src))),
- (vecType elemClass:$src)
+class Vector1_Build <ValueType vecType, ValueType elemType> : Pat <
+ (vecType (build_vector elemType:$src)),
+ (INSERT_SUBREG (vecType (IMPLICIT_DEF)), $src, sub0)
>;
-class Vector2_Build <ValueType vecType, RegisterClass vectorClass,
- ValueType elemType, RegisterClass elemClass> : Pat <
- (vecType (build_vector (elemType elemClass:$sub0), (elemType elemClass:$sub1))),
+class Vector2_Build <ValueType vecType, ValueType elemType> : Pat <
+ (vecType (build_vector elemType:$sub0, elemType:$sub1)),
(INSERT_SUBREG (INSERT_SUBREG
- (vecType (IMPLICIT_DEF)), elemClass:$sub0, sub0), elemClass:$sub1, sub1)
+ (vecType (IMPLICIT_DEF)), $sub0, sub0), $sub1, sub1)
>;
-class Vector4_Build <ValueType vecType, RegisterClass vectorClass,
- ValueType elemType, RegisterClass elemClass> : Pat <
- (vecType (build_vector (elemType elemClass:$x), (elemType elemClass:$y),
- (elemType elemClass:$z), (elemType elemClass:$w))),
+class Vector4_Build <ValueType vecType, ValueType elemType> : Pat <
+ (vecType (build_vector elemType:$x, elemType:$y, elemType:$z, elemType:$w)),
(INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
- (vecType (IMPLICIT_DEF)), elemClass:$x, sub0), elemClass:$y, sub1),
- elemClass:$z, sub2), elemClass:$w, sub3)
+ (vecType (IMPLICIT_DEF)), $x, sub0), $y, sub1), $z, sub2), $w, sub3)
>;
-class Vector8_Build <ValueType vecType, RegisterClass vectorClass,
- ValueType elemType, RegisterClass elemClass> : Pat <
- (vecType (build_vector (elemType elemClass:$sub0), (elemType elemClass:$sub1),
- (elemType elemClass:$sub2), (elemType elemClass:$sub3),
- (elemType elemClass:$sub4), (elemType elemClass:$sub5),
- (elemType elemClass:$sub6), (elemType elemClass:$sub7))),
- (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
+class Vector8_Build <ValueType vecType, ValueType elemType> : Pat <
+ (vecType (build_vector elemType:$sub0, elemType:$sub1,
+ elemType:$sub2, elemType:$sub3,
+ elemType:$sub4, elemType:$sub5,
+ elemType:$sub6, elemType:$sub7)),
(INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
- (vecType (IMPLICIT_DEF)), elemClass:$sub0, sub0), elemClass:$sub1, sub1),
- elemClass:$sub2, sub2), elemClass:$sub3, sub3),
- elemClass:$sub4, sub4), elemClass:$sub5, sub5),
- elemClass:$sub6, sub6), elemClass:$sub7, sub7)
+ (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
+ (vecType (IMPLICIT_DEF)), $sub0, sub0), $sub1, sub1),
+ $sub2, sub2), $sub3, sub3),
+ $sub4, sub4), $sub5, sub5),
+ $sub6, sub6), $sub7, sub7)
>;
-class Vector16_Build <ValueType vecType, RegisterClass vectorClass,
- ValueType elemType, RegisterClass elemClass> : Pat <
- (vecType (build_vector (elemType elemClass:$sub0), (elemType elemClass:$sub1),
- (elemType elemClass:$sub2), (elemType elemClass:$sub3),
- (elemType elemClass:$sub4), (elemType elemClass:$sub5),
- (elemType elemClass:$sub6), (elemType elemClass:$sub7),
- (elemType elemClass:$sub8), (elemType elemClass:$sub9),
- (elemType elemClass:$sub10), (elemType elemClass:$sub11),
- (elemType elemClass:$sub12), (elemType elemClass:$sub13),
- (elemType elemClass:$sub14), (elemType elemClass:$sub15))),
- (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
- (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
- (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
+class Vector16_Build <ValueType vecType, ValueType elemType> : Pat <
+ (vecType (build_vector elemType:$sub0, elemType:$sub1,
+ elemType:$sub2, elemType:$sub3,
+ elemType:$sub4, elemType:$sub5,
+ elemType:$sub6, elemType:$sub7,
+ elemType:$sub8, elemType:$sub9,
+ elemType:$sub10, elemType:$sub11,
+ elemType:$sub12, elemType:$sub13,
+ elemType:$sub14, elemType:$sub15)),
(INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
- (vecType (IMPLICIT_DEF)), elemClass:$sub0, sub0), elemClass:$sub1, sub1),
- elemClass:$sub2, sub2), elemClass:$sub3, sub3),
- elemClass:$sub4, sub4), elemClass:$sub5, sub5),
- elemClass:$sub6, sub6), elemClass:$sub7, sub7),
- elemClass:$sub8, sub8), elemClass:$sub9, sub9),
- elemClass:$sub10, sub10), elemClass:$sub11, sub11),
- elemClass:$sub12, sub12), elemClass:$sub13, sub13),
- elemClass:$sub14, sub14), elemClass:$sub15, sub15)
+ (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
+ (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
+ (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG (INSERT_SUBREG
+ (vecType (IMPLICIT_DEF)), $sub0, sub0), $sub1, sub1),
+ $sub2, sub2), $sub3, sub3),
+ $sub4, sub4), $sub5, sub5),
+ $sub6, sub6), $sub7, sub7),
+ $sub8, sub8), $sub9, sub9),
+ $sub10, sub10), $sub11, sub11),
+ $sub12, sub12), $sub13, sub13),
+ $sub14, sub14), $sub15, sub15)
>;
// bitconvert pattern
diff --git a/lib/Target/R600/R600Instructions.td b/lib/Target/R600/R600Instructions.td
index 361fc9816b4..46e34991d03 100644
--- a/lib/Target/R600/R600Instructions.td
+++ b/lib/Target/R600/R600Instructions.td
@@ -1418,7 +1418,7 @@ let Predicates = [isR600] in {
def RECIP_UINT_r600 : RECIP_UINT_Common <0x78>;
defm DIV_r600 : DIV_Common<RECIP_IEEE_r600>;
- def : POW_Common <LOG_IEEE_r600, EXP_IEEE_r600, MUL, R600_Reg32>;
+ def : POW_Common <LOG_IEEE_r600, EXP_IEEE_r600, MUL>;
def TGSI_LIT_Z_r600 : TGSI_LIT_Z_Common<MUL_LIT_r600, LOG_CLAMPED_r600, EXP_IEEE_r600>;
def : Pat<(fsqrt R600_Reg32:$src),
@@ -1536,7 +1536,7 @@ def RECIPSQRT_IEEE_eg : RECIPSQRT_IEEE_Common<0x89>;
def SIN_eg : SIN_Common<0x8D>;
def COS_eg : COS_Common<0x8E>;
-def : POW_Common <LOG_IEEE_eg, EXP_IEEE_eg, MUL, R600_Reg32>;
+def : POW_Common <LOG_IEEE_eg, EXP_IEEE_eg, MUL>;
def : SIN_PAT <SIN_eg>;
def : COS_PAT <COS_eg>;
def : Pat<(fsqrt R600_Reg32:$src),
@@ -1921,7 +1921,7 @@ def SIN_cm : SIN_Common<0x8D>;
def COS_cm : COS_Common<0x8E>;
} // End isVector = 1
-def : POW_Common <LOG_IEEE_cm, EXP_IEEE_cm, MUL, R600_Reg32>;
+def : POW_Common <LOG_IEEE_cm, EXP_IEEE_cm, MUL>;
def : SIN_PAT <SIN_cm>;
def : COS_PAT <COS_cm>;
@@ -2335,28 +2335,28 @@ def : Pat <
(SETNE_DX10 R600_Reg32:$src0, R600_Reg32:$src1)
>;
-def : Extract_Element <f32, v4f32, R600_Reg128, 0, sub0>;
-def : Extract_Element <f32, v4f32, R600_Reg128, 1, sub1>;
-def : Extract_Element <f32, v4f32, R600_Reg128, 2, sub2>;
-def : Extract_Element <f32, v4f32, R600_Reg128, 3, sub3>;
+def : Extract_Element <f32, v4f32, 0, sub0>;
+def : Extract_Element <f32, v4f32, 1, sub1>;
+def : Extract_Element <f32, v4f32, 2, sub2>;
+def : Extract_Element <f32, v4f32, 3, sub3>;
-def : Insert_Element <f32, v4f32, R600_Reg32, R600_Reg128, 0, sub0>;
-def : Insert_Element <f32, v4f32, R600_Reg32, R600_Reg128, 1, sub1>;
-def : Insert_Element <f32, v4f32, R600_Reg32, R600_Reg128, 2, sub2>;
-def : Insert_Element <f32, v4f32, R600_Reg32, R600_Reg128, 3, sub3>;
+def : Insert_Element <f32, v4f32, 0, sub0>;
+def : Insert_Element <f32, v4f32, 1, sub1>;
+def : Insert_Element <f32, v4f32, 2, sub2>;
+def : Insert_Element <f32, v4f32, 3, sub3>;
-def : Extract_Element <i32, v4i32, R600_Reg128, 0, sub0>;
-def : Extract_Element <i32, v4i32, R600_Reg128, 1, sub1>;
-def : Extract_Element <i32, v4i32, R600_Reg128, 2, sub2>;
-def : Extract_Element <i32, v4i32, R600_Reg128, 3, sub3>;
+def : Extract_Element <i32, v4i32, 0, sub0>;
+def : Extract_Element <i32, v4i32, 1, sub1>;
+def : Extract_Element <i32, v4i32, 2, sub2>;
+def : Extract_Element <i32, v4i32, 3, sub3>;
-def : Insert_Element <i32, v4i32, R600_Reg32, R600_Reg128, 0, sub0>;
-def : Insert_Element <i32, v4i32, R600_Reg32, R600_Reg128, 1, sub1>;
-def : Insert_Element <i32, v4i32, R600_Reg32, R600_Reg128, 2, sub2>;
-def : Insert_Element <i32, v4i32, R600_Reg32, R600_Reg128, 3, sub3>;
+def : Insert_Element <i32, v4i32, 0, sub0>;
+def : Insert_Element <i32, v4i32, 1, sub1>;
+def : Insert_Element <i32, v4i32, 2, sub2>;
+def : Insert_Element <i32, v4i32, 3, sub3>;
-def : Vector4_Build <v4f32, R600_Reg128, f32, R600_Reg32>;
-def : Vector4_Build <v4i32, R600_Reg128, i32, R600_Reg32>;
+def : Vector4_Build <v4f32, f32>;
+def : Vector4_Build <v4i32, i32>;
// bitconvert patterns
diff --git a/lib/Target/R600/SIInstructions.td b/lib/Target/R600/SIInstructions.td
index a321ee1759b..bb253c7b2e0 100644
--- a/lib/Target/R600/SIInstructions.td
+++ b/lib/Target/R600/SIInstructions.td
@@ -1277,77 +1277,77 @@ defm : SamplePatterns<v16i32>;
foreach Index = 0-2 in {
def Extract_Element_v2i32_#Index : Extract_Element <
- i32, v2i32, VReg_64, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v2i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v2i32_#Index : Insert_Element <
- i32, v2i32, VReg_32, VReg_64, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v2i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Extract_Element_v2f32_#Index : Extract_Element <
- f32, v2f32, VReg_64, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v2f32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v2f32_#Index : Insert_Element <
- f32, v2f32, VReg_32, VReg_64, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v2f32, Index, !cast<SubRegIndex>(sub#Index)
>;
}
foreach Index = 0-3 in {
def Extract_Element_v4i32_#Index : Extract_Element <
- i32, v4i32, VReg_128, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v4i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v4i32_#Index : Insert_Element <
- i32, v4i32, VReg_32, VReg_128, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v4i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Extract_Element_v4f32_#Index : Extract_Element <
- f32, v4f32, VReg_128, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v4f32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v4f32_#Index : Insert_Element <
- f32, v4f32, VReg_32, VReg_128, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v4f32, Index, !cast<SubRegIndex>(sub#Index)
>;
}
foreach Index = 0-7 in {
def Extract_Element_v8i32_#Index : Extract_Element <
- i32, v8i32, VReg_256, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v8i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v8i32_#Index : Insert_Element <
- i32, v8i32, VReg_32, VReg_256, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v8i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Extract_Element_v8f32_#Index : Extract_Element <
- f32, v8f32, VReg_256, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v8f32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v8f32_#Index : Insert_Element <
- f32, v8f32, VReg_32, VReg_256, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v8f32, Index, !cast<SubRegIndex>(sub#Index)
>;
}
foreach Index = 0-15 in {
def Extract_Element_v16i32_#Index : Extract_Element <
- i32, v16i32, VReg_512, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v16i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v16i32_#Index : Insert_Element <
- i32, v16i32, VReg_32, VReg_512, Index, !cast<SubRegIndex>(sub#Index)
+ i32, v16i32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Extract_Element_v16f32_#Index : Extract_Element <
- f32, v16f32, VReg_512, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v16f32, Index, !cast<SubRegIndex>(sub#Index)
>;
def Insert_Element_v16f32_#Index : Insert_Element <
- f32, v16f32, VReg_32, VReg_512, Index, !cast<SubRegIndex>(sub#Index)
+ f32, v16f32, Index, !cast<SubRegIndex>(sub#Index)
>;
}
-def : Vector1_Build <v1i32, VReg_32, i32, VReg_32>;
-def : Vector2_Build <v2i32, VReg_64, i32, VReg_32>;
-def : Vector2_Build <v2f32, VReg_64, f32, VReg_32>;
-def : Vector4_Build <v4i32, VReg_128, i32, VReg_32>;
-def : Vector4_Build <v4f32, VReg_128, f32, VReg_32>;
-def : Vector8_Build <v8i32, VReg_256, i32, VReg_32>;
-def : Vector8_Build <v8f32, VReg_256, f32, VReg_32>;
-def : Vector16_Build <v16i32, VReg_512, i32, VReg_32>;
-def : Vector16_Build <v16f32, VReg_512, f32, VReg_32>;
+def : Vector1_Build <v1i32, i32>;
+def : Vector2_Build <v2i32, i32>;
+def : Vector2_Build <v2f32, f32>;
+def : Vector4_Build <v4i32, i32>;
+def : Vector4_Build <v4f32, f32>;
+def : Vector8_Build <v8i32, i32>;
+def : Vector8_Build <v8f32, f32>;
+def : Vector16_Build <v16i32, i32>;
+def : Vector16_Build <v16f32, f32>;
def : BitConvert <i32, f32, SReg_32>;
def : BitConvert <i32, f32, VReg_32>;
@@ -1431,7 +1431,7 @@ def : Pat <
/********** ================== **********/
/* llvm.AMDGPU.pow */
-def : POW_Common <V_LOG_F32_e32, V_EXP_F32_e32, V_MUL_LEGACY_F32_e32, VReg_32>;
+def : POW_Common <V_LOG_F32_e32, V_EXP_F32_e32, V_MUL_LEGACY_F32_e32>;
def : Pat <
(int_AMDGPU_div f32:$src0, f32:$src1),
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-03 19:55:25 +0000