Skip to content

[PowerPC][CLANG] DMF VSX Vector float GER 2x (rank-2 update) #147383

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.

Sign up for GitHub

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

Jump to bottom
Open
wants to merge 3 commits into
base: users/lei137/mmafltbuiltin
Choose a base branch
from
Open

[PowerPC][CLANG] DMF VSX Vector float GER 2x (rank-2 update) #147383

wants to merge 3 commits into from

Conversation

lei137
Copy link
Contributor

@lei137 lei137 commented Jul 7, 2025

Add clang builtins for DMF VSX Vector floats:

void __builtin_mma_dmxvf16gerx2 (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2nn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2np (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2pn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2pp (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_pmdmxvf16gerx2 (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2nn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2np (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2pn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2pp (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);

void __builtin_mma_dmxvbf16gerx2 (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2nn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2np (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2pn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2pp (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_pmdmxvbf16gerx2 (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2nn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2np (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2pn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2pp (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);

@lei137 lei137 requested a review from maryammo July 7, 2025 19:54
@llvmbot llvmbot added clang Clang issues not falling into any other category backend:PowerPC clang:frontend Language frontend issues, e.g. anything involving "Sema" labels Jul 7, 2025
@llvmbot
Copy link
Member

llvmbot commented Jul 7, 2025 •
edited
Loading

@llvm/pr-subscribers-clang

@llvm/pr-subscribers-backend-powerpc

Author: Lei Huang (lei137)

Changes

Add clang builtins for DMF VSX Vector floats:

void __builtin_mma_dmxvf16gerx2 (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2nn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2np (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2pn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvf16gerx2pp (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_pmdmxvf16gerx2 (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2nn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2np (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2pn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvf16gerx2pp (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);

void __builtin_mma_dmxvbf16gerx2 (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2nn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2np (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2pn (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_dmxvbf16gerx2pp (__dmr1024 *, __vector_pair, vec_t);
void __builtin_mma_pmdmxvbf16gerx2 (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2nn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2np (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2pn (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);
void __builtin_mma_pmdmxvbf16gerx2pp (__dmr1024 *, __vector_pair, vec_t, uint8, uint4, uint2);

Patch is 22.00 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/147383.diff

3 Files Affected:

  • (modified) clang/include/clang/Basic/BuiltinsPPC.def (+9)
  • (added) clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c (+309)
  • (modified) clang/test/CodeGen/PowerPC/ppc-future-mma-builtin-err.c (+48)
diff --git a/clang/include/clang/Basic/BuiltinsPPC.def b/clang/include/clang/Basic/BuiltinsPPC.def
index 3fa9961a9602c..917167ff4a7ba 100644
--- a/clang/include/clang/Basic/BuiltinsPPC.def
+++ b/clang/include/clang/Basic/BuiltinsPPC.def
@@ -1095,6 +1095,15 @@ UNALIASED_CUSTOM_MMA_BUILTIN(mma_xvbf16ger2, "vW512*VV",
                              "mma,paired-vector-memops")
 UNALIASED_CUSTOM_MMA_BUILTIN(mma_pmxvbf16ger2, "vW512*VVi15i15i3",
                              "mma,paired-vector-memops")
+UNALIASED_CUSTOM_MMA_BUILTIN(mma_dmxvbf16gerx2, "vW1024*W256V",
+                             "mma,paired-vector-memops")
+UNALIASED_CUSTOM_MMA_BUILTIN(mma_pmdmxvbf16gerx2, "vW1024*W256Vi255i15i3",
+                             "mma,paired-vector-memops")
+UNALIASED_CUSTOM_MMA_BUILTIN(mma_dmxvf16gerx2, "vW1024*W256V",
+                             "mma,paired-vector-memops")
+UNALIASED_CUSTOM_MMA_BUILTIN(mma_pmdmxvf16gerx2, "vW1024*W256Vi255i15i3",
+                             "mma,paired-vector-memops")
+
 
 // FIXME: Obviously incomplete.
 
diff --git a/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c b/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c
new file mode 100644
index 0000000000000..8fc9a68a5a613
--- /dev/null
+++ b/clang/test/CodeGen/PowerPC/builtins-dmf-vsx-vector-float.c
@@ -0,0 +1,309 @@
+// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 5
+//       Update then manual applied to commonize the checks for AIX and LoP.
+// RUN: %clang_cc1 -O3 -triple powerpc64le-unknown-unknown -target-cpu future \
+// RUN:            -emit-llvm %s -o - | FileCheck %s
+// RUN: %clang_cc1 -O3 -triple powerpc64-ibm-aix -target-cpu future \
+// RUN:             -emit-llvm %s -o - | FileCheck %s
+
+// CHECK-LABEL: void @test_dmxvbf16gerx2(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2:![0-9]+]]
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6:![0-9]+]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvbf16gerx2(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvbf16gerx2nn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvbf16gerx2nn(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvbf16gerx2np(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvbf16gerx2np(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvbf16gerx2pn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvbf16gerx2pn(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvbf16gerx2pp(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvbf16gerx2pp(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvbf16gerx2(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvbf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvbf16gerx2(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvbf16gerx2nn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvbf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvbf16gerx2nn(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvbf16gerx2np(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvbf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvbf16gerx2np(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvbf16gerx2pn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvbf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvbf16gerx2pn(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvbf16gerx2pp(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvbf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvbf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvbf16gerx2pp(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvf16gerx2(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2:![0-9]+]]
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6:![0-9]+]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvf16gerx2(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvf16gerx2nn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvf16gerx2nn(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvf16gerx2np(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvf16gerx2np(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvf16gerx2pn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvf16gerx2pn(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_dmxvf16gerx2pp(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.dmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]])
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_dmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_dmxvf16gerx2pp(&vdmr, vp, vc);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvf16gerx2(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP1:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2(<256 x i1> [[TMP0]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP1]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvf16gerx2(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvf16gerx2(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvf16gerx2nn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2nn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvf16gerx2nn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvf16gerx2nn(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvf16gerx2np(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2np(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvf16gerx2np(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvf16gerx2np(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvf16gerx2pn(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pn(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvf16gerx2pn(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvf16gerx2pn(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK-LABEL: void @test_pmdmxvf16gerx2pp(
+// CHECK-NEXT:  [[ENTRY:.*:]]
+// CHECK-NEXT:    [[TMP0:%.*]] = load <1024 x i1>, ptr [[VDMRP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    [[TMP1:%.*]] = load <256 x i1>, ptr [[VPP:%.*]], align 32, !tbaa [[TBAA2]]
+// CHECK-NEXT:    [[TMP2:%.*]] = tail call <1024 x i1> @llvm.ppc.mma.pmdmxvf16gerx2pp(<1024 x i1> [[TMP0]], <256 x i1> [[TMP1]], <16 x i8> [[VC:%.*]], i32 0, i32 0, i32 0)
+// CHECK-NEXT:    store <1024 x i1> [[TMP2]], ptr [[RESP:%.*]], align 128, !tbaa [[TBAA6]]
+// CHECK-NEXT:    ret void
+//
+void test_pmdmxvf16gerx2pp(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc, unsigned char *resp) {
+  __dmr1024 vdmr = *((__dmr1024 *)vdmrp);
+  __vector_pair vp = *((__vector_pair *)vpp);
+  __builtin_mma_pmdmxvf16gerx2pp(&vdmr, vp, vc, 0, 0, 0);
+  *((__dmr1024 *)resp) = vdmr;
+}
+
+// CHECK: [[TBAA2]] = !{[[META3:![0-9]+]], [[META3]], i64 0}
+// CHECK: [[META3]] = !{!"__vector_pair", [[META4:![0-9]+]], i64 0}
+// CHECK: [[META4]] = !{!"omnipotent char", [[META5:![0-9]+]], i64 0}
+// CHECK: [[META5]] = !{!"Simple C/C++ TBAA"}
+// CHECK: [[TBAA6]] = !{[[META7:![0-9]+]], [[META7]], i64 0}
+// CHECK: [[META7]] = !{!"__dmr1024", [[META4]], i64 0}
diff --git a/clang/test/CodeGen/PowerPC/ppc-future-mma-builtin-err.c b/clang/test/CodeGen/PowerPC/ppc-future-mma-builtin-err.c
index 1b8d345ac7ec7..b70dd739652b8 100644
--- a/clang/test/CodeGen/PowerPC/ppc-future-mma-builtin-err.c
+++ b/clang/test/CodeGen/PowerPC/ppc-future-mma-builtin-err.c
@@ -18,4 +18,52 @@ void test_mma(unsigned char *vdmrp, unsigned char *vpp, vector unsigned char vc)
 // CHECK: error: '__builtin_mma_pmdmxvi8gerx4pp' needs target feature mma,paired-vector-memops
 // CHECK: error: '__builtin_mma_dmxvi8gerx4spp' needs target feature mma,paired-vector-memops
 // CHECK: error: '__builtin_mma_pmdmxvi8gerx4spp' needs target feature mma,paired-vector-memops
+
+  // DMF VSX Vector bfloat16 GER 2x builtins.
+
+  __builtin_mma_dmxvbf16gerx2(&vdmr, vp, vc);
+  __builtin_mma_dmxvbf16gerx2nn(&vdmr, vp, vc);
+  __builtin_mma_dmxvbf16gerx2np(&vdmr, vp, vc);
+  __builtin_mma_dmxvbf16gerx2pn(&vdmr, vp, vc);
+  __builtin_mma_dmxvbf16gerx2pp(&vdmr, vp, vc);
+  __builtin_mma_pmdmxvbf16gerx2(&vdmr, vp, vc, 0, 0, 0);
+  __builtin_mma_pmdmxvbf16gerx2nn(&vdmr, vp, vc, 0, 0, 0);
+  __builtin_mma_pmdmxvbf16gerx2np(&vdmr, vp, vc, 0, 0, 0);
+  __builtin_mma_pmdmxvbf16gerx2pn(&vdmr, vp, vc, 0, 0, 0);
+  __builtin_mma_pmdmxvbf16gerx2pp(&vdmr, vp, vc, 0, 0, 0);
+
+// CHECK: error: '__builtin_mma_dmxvbf16gerx2' needs target feature mma,paired-vector-memops
+// CHECK: error: '__builtin_mma_dmxvbf16gerx2nn' needs target feature mma,paired-vector-memops
+// CHECK: error: '__builtin_mma_dmxvbf16gerx2np' needs target feature mma,paired-vector-memops
+// CHECK: error: '__builtin_mma_dmxvbf16gerx2pn' needs target feature mma,paired-vector-memops
+// CHECK: error: '__builtin_mma_dmxvbf16gerx2pp' needs target feature mma,paired-vector-memops...
[truncated]

@lei137 lei137 changed the title RFC02658:CLANG: DMF VSX Vector float GER 2x (rank-2 update) PowerPC:CLANG: DMF VSX Vector float GER 2x (rank-2 update) Jul 7, 2025
@lei137 lei137 changed the title PowerPC:CLANG: DMF VSX Vector float GER 2x (rank-2 update) [PowerPC][CLANG] DMF VSX Vector float GER 2x (rank-2 update) Jul 7, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@maryammo maryammo Awaiting requested review from maryammo

@RolandF77 RolandF77 Awaiting requested review from RolandF77

@amy-kwan amy-kwan Awaiting requested review from amy-kwan

@diggerlin diggerlin Awaiting requested review from diggerlin

Assignees
No one assigned
Labels
backend:PowerPC clang:frontend Language frontend issues, e.g. anything involving "Sema" clang Clang issues not falling into any other category
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@lei137 @llvmbot