[RISCV] Implement Intrinsics for XCVmac Extension in CV32E40P #83112
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
@llvm/pr-subscribers-backend-risc-v @llvm/pr-subscribers-llvm-ir Author: None (realqhc) ChangesImplement XCVmac intrinsics and CodeGen for CV32E40P according to the specification. This commit is part of a patch-set to upstream the vendor specific extensions of CV32E40P that need LLVM intrinsics to implement Clang builtins. Contributors: @CharKeaney, @ChunyuLiao, @jeremybennett, @lewis-revill, @NandniJamnadas, @PaoloS02, @serkm, @simonpcook, @xingmingjie. Full diff: https://github.com/llvm/llvm-project/pull/83112.diff 3 Files Affected:
diff --git a/llvm/include/llvm/IR/IntrinsicsRISCVXCV.td b/llvm/include/llvm/IR/IntrinsicsRISCVXCV.td
index f1590ad66e362b..af902c7f1ed56f 100644
--- a/llvm/include/llvm/IR/IntrinsicsRISCVXCV.td
+++ b/llvm/include/llvm/IR/IntrinsicsRISCVXCV.td
@@ -18,6 +18,18 @@ class ScalarCoreVBitManipGprIntrinsic
: DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty],
[IntrNoMem, IntrSpeculatable]>;
+class ScalarCoreVMacGprGprGprIntrinsic
+ : Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+ [IntrNoMem, IntrWillReturn, IntrSpeculatable]>;
+
+class ScalarCoreVMacGprGPRImmIntrinsic
+ : Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+ [IntrNoMem, IntrWillReturn, IntrSpeculatable, ImmArg<ArgIndex<2>>]>;
+
+class ScalarCoreVMacGprGprGprImmIntrinsic
+ : Intrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
+ [IntrNoMem, IntrWillReturn, IntrSpeculatable, ImmArg<ArgIndex<3>>]>;
+
let TargetPrefix = "riscv" in {
def int_riscv_cv_bitmanip_extract : ScalarCoreVBitManipGprGprIntrinsic;
def int_riscv_cv_bitmanip_extractu : ScalarCoreVBitManipGprGprIntrinsic;
@@ -34,4 +46,25 @@ let TargetPrefix = "riscv" in {
: DefaultAttrsIntrinsic<[llvm_i32_ty], [llvm_i32_ty, llvm_i32_ty, llvm_i32_ty],
[IntrNoMem, IntrWillReturn, IntrSpeculatable,
ImmArg<ArgIndex<1>>, ImmArg<ArgIndex<2>>]>;
+
+ def int_riscv_cv_mac_mac : ScalarCoreVMacGprGprGprIntrinsic;
+ def int_riscv_cv_mac_msu : ScalarCoreVMacGprGprGprIntrinsic;
+
+ def int_riscv_cv_mac_muluN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_mulhhuN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_mulsN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_mulhhsN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_muluRN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_mulhhuRN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_mulsRN : ScalarCoreVMacGprGPRImmIntrinsic;
+ def int_riscv_cv_mac_mulhhsRN : ScalarCoreVMacGprGPRImmIntrinsic;
+
+ def int_riscv_cv_mac_macuN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_machhuN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_macsN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_machhsN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_macuRN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_machhuRN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_macsRN : ScalarCoreVMacGprGprGprImmIntrinsic;
+ def int_riscv_cv_mac_machhsRN : ScalarCoreVMacGprGprGprImmIntrinsic;
} // TargetPrefix = "riscv"
diff --git a/llvm/lib/Target/RISCV/RISCVInstrInfoXCV.td b/llvm/lib/Target/RISCV/RISCVInstrInfoXCV.td
index 924e91e15c348f..b62906fcf8e3ca 100644
--- a/llvm/lib/Target/RISCV/RISCVInstrInfoXCV.td
+++ b/llvm/lib/Target/RISCV/RISCVInstrInfoXCV.td
@@ -704,3 +704,36 @@ let Predicates = [HasVendorXCVbitmanip, IsRV32] in {
(CV_BITREV GPR:$rs1, cv_tuimm2:$radix, cv_tuimm5:$pts)>;
def : Pat<(bitreverse (XLenVT GPR:$rs)), (CV_BITREV GPR:$rs, 0, 0)>;
}
+
+class PatCoreVMacGprGprGpr <string intr, string asm>
+ : Pat<(!cast<Intrinsic>("int_riscv_cv_mac_" # intr) GPR:$rs1, GPR:$rs2, GPR:$rd),
+ (!cast<RVInst>("CV_" # asm) GPR:$rd, GPR:$rs1, GPR:$rs2)>;
+class PatCoreVMacGprGprGprUimm5 <string intr, string asm>
+ : Pat<(!cast<Intrinsic>("int_riscv_cv_mac_" # intr) GPR:$rs1, GPR:$rs2, GPR:$rd, cv_tuimm5:$imm5),
+ (!cast<RVInst>("CV_" # asm) GPR:$rd, GPR:$rs1, GPR:$rs2, cv_tuimm5:$imm5)>;
+class PatCoreVMacGprGprUimm5 <string intr, string asm>
+ : Pat<(!cast<Intrinsic>("int_riscv_cv_mac_" # intr) GPR:$rs1, GPR:$rs2, cv_tuimm5:$imm5),
+ (!cast<RVInst>("CV_" # asm) GPR:$rs1, GPR:$rs2, cv_tuimm5:$imm5)>;
+
+let Predicates = [HasVendorXCVmac] in {
+ def : PatCoreVMacGprGprGpr<"mac", "MAC">;
+ def : PatCoreVMacGprGprGpr<"msu", "MSU">;
+
+ def : PatCoreVMacGprGprUimm5<"muluN", "MULUN">;
+ def : PatCoreVMacGprGprUimm5<"mulhhuN", "MULHHUN">;
+ def : PatCoreVMacGprGprUimm5<"mulsN", "MULSN">;
+ def : PatCoreVMacGprGprUimm5<"mulhhsN", "MULHHSN">;
+ def : PatCoreVMacGprGprUimm5<"muluRN", "MULURN">;
+ def : PatCoreVMacGprGprUimm5<"mulhhuRN", "MULHHURN">;
+ def : PatCoreVMacGprGprUimm5<"mulsRN", "MULSRN">;
+ def : PatCoreVMacGprGprUimm5<"mulhhsRN", "MULHHSRN">;
+
+ def : PatCoreVMacGprGprGprUimm5<"macuN", "MACUN">;
+ def : PatCoreVMacGprGprGprUimm5<"machhuN", "MACHHUN">;
+ def : PatCoreVMacGprGprGprUimm5<"macsN", "MACSN">;
+ def : PatCoreVMacGprGprGprUimm5<"machhsN", "MACHHSN">;
+ def : PatCoreVMacGprGprGprUimm5<"macuRN", "MACURN">;
+ def : PatCoreVMacGprGprGprUimm5<"machhuRN", "MACHHURN">;
+ def : PatCoreVMacGprGprGprUimm5<"macsRN", "MACSRN">;
+ def : PatCoreVMacGprGprGprUimm5<"machhsRN", "MACHHSRN">;
+}
diff --git a/llvm/test/CodeGen/RISCV/xcvmac.ll b/llvm/test/CodeGen/RISCV/xcvmac.ll
new file mode 100644
index 00000000000000..68efdf7210f7f5
--- /dev/null
+++ b/llvm/test/CodeGen/RISCV/xcvmac.ll
@@ -0,0 +1,211 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+m -mattr=+xcvmac -verify-machineinstrs < %s \
+; RUN: | FileCheck %s
+
+declare i32 @llvm.riscv.cv.mac.mac(i32, i32, i32)
+
+define i32 @test.mac(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.mac:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mac a2, a0, a1
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mac(i32 %a, i32 %b, i32 %c)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.msu(i32, i32, i32)
+
+define i32 @test.msu(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.msu:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.msu a2, a0, a1
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.msu(i32 %a, i32 %b, i32 %c)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.muluN(i32, i32, i32)
+
+define i32 @test.muluN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.muluN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulun a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.muluN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.mulhhuN(i32, i32, i32)
+
+define i32 @test.mulhhuN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.mulhhuN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulhhun a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mulhhuN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.mulsN(i32, i32, i32)
+
+define i32 @test.mulsN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.mulsN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulsn a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mulsN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.mulhhsN(i32, i32, i32)
+
+define i32 @test.mulhhsN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.mulhhsN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulhhsn a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mulhhsN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.muluRN(i32, i32, i32)
+
+define i32 @test.muluRN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.muluRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulurn a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.muluRN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.mulhhuRN(i32, i32, i32)
+
+define i32 @test.mulhhuRN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.mulhhuRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulhhurn a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mulhhuRN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.mulsRN(i32, i32, i32)
+
+define i32 @test.mulsRN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.mulsRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulsrn a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mulsRN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.mulhhsRN(i32, i32, i32)
+
+define i32 @test.mulhhsRN(i32 %a, i32 %b) {
+; CHECK-LABEL: test.mulhhsRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.mulhhsrn a0, a0, a1, 5
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.mulhhsRN(i32 %a, i32 %b, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.macuN(i32, i32, i32, i32)
+
+define i32 @test.macuN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.macuN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.macun a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.macuN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.machhuN(i32, i32, i32, i32)
+
+define i32 @test.machhuN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.machhuN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.machhun a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.machhuN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.macsN(i32, i32, i32, i32)
+
+define i32 @test.macsN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.macsN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.macsn a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.macsN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.machhsN(i32, i32, i32, i32)
+
+define i32 @test.machhsN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.machhsN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.machhsn a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.machhsN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.macuRN(i32, i32, i32, i32)
+
+define i32 @test.macuRN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.macuRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.macurn a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.macuRN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.machhuRN(i32, i32, i32, i32)
+
+define i32 @test.machhuRN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.machhuRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.machhurn a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.machhuRN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.macsRN(i32, i32, i32, i32)
+
+define i32 @test.macsRN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.macsRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.macsrn a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.macsRN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
+
+declare i32 @llvm.riscv.cv.mac.machhsRN(i32, i32, i32, i32)
+
+define i32 @test.machhsRN(i32 %a, i32 %b, i32 %c) {
+; CHECK-LABEL: test.machhsRN:
+; CHECK: # %bb.0:
+; CHECK-NEXT: cv.machhsrn a2, a0, a1, 5
+; CHECK-NEXT: mv a0, a2
+; CHECK-NEXT: ret
+ %1 = call i32 @llvm.riscv.cv.mac.machhsRN(i32 %a, i32 %b, i32 %c, i32 5)
+ ret i32 %1
+}
|
|
|
realqhc
force-pushed
the
mac-intrinsics-upstream
branch
from
d462def to
0215040
Compare
11 tasks
realqhc
force-pushed
the
mac-intrinsics-upstream
branch
from
0215040 to
de54c9b
Compare
realqhc
force-pushed
the
mac-intrinsics-upstream
branch
from
de54c9b to
096936a
Compare
Implement XCVmac intrinsics and CodeGen for CV32E40P according to the specification. This commit is part of a patch-set to upstream the vendor specific extensions of CV32E40P that need LLVM intrinsics to implement Clang builtins. Contributors: @CharKeaney, @ChunyuLiao, @jeremybennett, @lewis-revill, @NandniJamnadas, @PaoloS02, @serkm, @simonpcook, @xingmingjie.
realqhc
force-pushed
the
mac-intrinsics-upstream
branch
from
096936a to
c0d5e45
Compare
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/66/builds/1495 Here is the relevant piece of the build log for the reference: |
|
LLVM Buildbot has detected a new failure on builder Full details are available at: https://lab.llvm.org/buildbot/#/builders/63/builds/514 Here is the relevant piece of the build log for the reference: |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Implement XCVmac intrinsics for CV32E40P according to the specification.
This commit is part of a patch-set to upstream the vendor specific extensions of CV32E40P that need LLVM intrinsics to implement Clang builtins.
Contributors: @CharKeaney, @ChunyuLiao, @jeremybennett, @lewis-revill, @NandniJamnadas, @PaoloS02, @serkm, @simonpcook, @xingmingjie.