-
Notifications
You must be signed in to change notification settings - Fork 14.4k
[SiFive][RISCV] Remove VMConstraint from XSfvqmaccqoq and XSfvfwmaccqqq instructions. #142914
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
@llvm/pr-subscribers-mc @llvm/pr-subscribers-backend-risc-v Author: Craig Topper (topperc) ChangesThese instructions don't have a mask operand. The VMConstraint would cause an assertion if V0 is used as the destination and the last register isn't V0. Full diff: https://github.com/llvm/llvm-project/pull/142914.diff 4 Files Affected:
diff --git a/llvm/lib/Target/RISCV/RISCVInstrFormats.td b/llvm/lib/Target/RISCV/RISCVInstrFormats.td
index 8d56637d8b281..088a6923fadb1 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrFormats.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrFormats.td
@@ -99,6 +99,8 @@ def VMConstraint : RISCVVConstraint<0b100>;
//
// * vcompress: The destination vector register group cannot overlap the
// source vector register group or the source mask register
+def WidenVNoMask : RISCVVConstraint<!or(VS2Constraint.Value,
+ VS1Constraint.Value)>;
def WidenV : RISCVVConstraint<!or(VS2Constraint.Value,
VS1Constraint.Value,
VMConstraint.Value)>;
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td b/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
index 6981daa75d092..affe3a8476753 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoXSf.td
@@ -208,7 +208,7 @@ let Predicates = [HasVendorXSfvqmaccdod], DecoderNamespace = "XSfvector",
}
let Predicates = [HasVendorXSfvqmaccqoq], DecoderNamespace = "XSfvector",
- DestEEW = EEWSEWx4, RVVConstraint=WidenV in {
+ DestEEW = EEWSEWx4, RVVConstraint=WidenVNoMask in {
def VQMACCU_4x8x4 : CustomSiFiveVMACC<0b111100, OPMVV, "sf.vqmaccu.4x8x4">;
def VQMACC_4x8x4 : CustomSiFiveVMACC<0b111101, OPMVV, "sf.vqmacc.4x8x4">;
def VQMACCUS_4x8x4 : CustomSiFiveVMACC<0b111110, OPMVV, "sf.vqmaccus.4x8x4">;
@@ -216,7 +216,7 @@ let Predicates = [HasVendorXSfvqmaccqoq], DecoderNamespace = "XSfvector",
}
let Predicates = [HasVendorXSfvfwmaccqqq], DecoderNamespace = "XSfvector",
- DestEEW = EEWSEWx2, RVVConstraint=WidenV in {
+ DestEEW = EEWSEWx2, RVVConstraint=WidenVNoMask in {
def VFWMACC_4x4x4 : CustomSiFiveVMACC<0b111100, OPFVV, "sf.vfwmacc.4x4x4">;
}
diff --git a/llvm/test/MC/RISCV/rvv/xsfvfwmacc.s b/llvm/test/MC/RISCV/rvv/xsfvfwmacc.s
index a9843c350fc80..5168c42da70d9 100644
--- a/llvm/test/MC/RISCV/rvv/xsfvfwmacc.s
+++ b/llvm/test/MC/RISCV/rvv/xsfvfwmacc.s
@@ -13,3 +13,9 @@ sf.vfwmacc.4x4x4 v8, v4, v20
# CHECK-ENCODING: [0x5b,0x14,0x42,0xf3]
# CHECK-ERROR: instruction requires the following: 'XSfvfwmaccqqq' (SiFive Matrix Multiply Accumulate Instruction and 4-by-4))
# CHECK-UNKNOWN: f342145b <unknown>
+
+sf.vfwmacc.4x4x4 v0, v4, v20
+# CHECK-INST: sf.vfwmacc.4x4x4 v0, v4, v20
+# CHECK-ENCODING: [0x5b,0x10,0x42,0xf3]
+# CHECK-ERROR: instruction requires the following: 'XSfvfwmaccqqq' (SiFive Matrix Multiply Accumulate Instruction and 4-by-4))
+# CHECK-UNKNOWN: f342105b <unknown>
diff --git a/llvm/test/MC/RISCV/rvv/xsfvqmacc.s b/llvm/test/MC/RISCV/rvv/xsfvqmacc.s
index 81703c847d74e..d7d3cba41cbc6 100644
--- a/llvm/test/MC/RISCV/rvv/xsfvqmacc.s
+++ b/llvm/test/MC/RISCV/rvv/xsfvqmacc.s
@@ -55,3 +55,9 @@ sf.vqmaccsu.4x8x4 v8, v4, v20
# CHECK-ENCODING: [0x5b,0x24,0x42,0xff]
# CHECK-ERROR: instruction requires the following: 'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))
# CHECK-UNKNOWN: ff42245b <unknown>
+
+sf.vqmaccu.4x8x4 v0, v4, v20
+# CHECK-INST: sf.vqmaccu.4x8x4 v0, v4, v20
+# CHECK-ENCODING: [0x5b,0x20,0x42,0xf3]
+# CHECK-ERROR: instruction requires the following: 'XSfvqmaccqoq' (SiFive Int8 Matrix Multiplication Instructions (4-by-8 and 8-by-4))
+# CHECK-UNKNOWN: f342205b <unknown>
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM.
…qq instructions. (llvm#142914) These instructions don't have a mask operand. The VMConstraint would cause an assertion if V0 is used as the destination and the last register isn't V0.
…qq instructions. (llvm#142914) These instructions don't have a mask operand. The VMConstraint would cause an assertion if V0 is used as the destination and the last register isn't V0.
These instructions don't have a mask operand. The VMConstraint would cause an assertion if V0 is used as the destination and the last register isn't V0.