https://github.com/fileho updated https://github.com/llvm/llvm-project/pull/186866
From aa26cef64029a0587ed3fa632a84eef05fd92c71 Mon Sep 17 00:00:00 2001 From: Jiri Filek <[email protected]> Date: Mon, 16 Mar 2026 20:31:56 +0100 Subject: [PATCH 1/2] [CIR][AArch64] Implement vget_lane_bf16 and vgetq_lane_bf16 builtins --- clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp | 10 ++++++---- .../CodeGen/AArch64/bf16-getset-intrinsics.c | 18 ------------------ clang/test/CodeGen/AArch64/neon/bf16-getset.c | 16 ++++++++++++++++ 3 files changed, 22 insertions(+), 22 deletions(-) diff --git a/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp b/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp index d9d303cd07b92..902d583698047 100644 --- a/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp +++ b/clang/lib/CIR/CodeGen/CIRGenBuiltinAArch64.cpp @@ -2180,21 +2180,23 @@ CIRGenFunction::emitAArch64BuiltinExpr(unsigned builtinID, const CallExpr *expr, case NEON::BI__builtin_neon_vqdmlals_lane_s32: case NEON::BI__builtin_neon_vqdmlals_laneq_s32: case NEON::BI__builtin_neon_vqdmlsls_lane_s32: - case NEON::BI__builtin_neon_vqdmlsls_laneq_s32: - case NEON::BI__builtin_neon_vget_lane_bf16: + case NEON::BI__builtin_neon_vqdmlsls_laneq_s32: { cgm.errorNYI(expr->getSourceRange(), std::string("unimplemented AArch64 builtin call: ") + getContext().BuiltinInfo.getName(builtinID)); return mlir::Value{}; + } + case NEON::BI__builtin_neon_vget_lane_bf16: case NEON::BI__builtin_neon_vduph_lane_bf16: { return cir::VecExtractOp::create(builder, loc, ops[0], ops[1]); } - case NEON::BI__builtin_neon_vduph_lane_f16: - case NEON::BI__builtin_neon_vgetq_lane_bf16: + case NEON::BI__builtin_neon_vduph_lane_f16: { cgm.errorNYI(expr->getSourceRange(), std::string("unimplemented AArch64 builtin call: ") + getContext().BuiltinInfo.getName(builtinID)); return mlir::Value{}; + } + case NEON::BI__builtin_neon_vgetq_lane_bf16: case NEON::BI__builtin_neon_vduph_laneq_bf16: { return cir::VecExtractOp::create(builder, loc, ops[0], ops[1]); } diff --git a/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c b/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c index 69171902c7e69..19819d0d6c8c8 100644 --- a/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c +++ b/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c @@ -41,24 +41,6 @@ bfloat16x4_t test_vget_low_bf16(bfloat16x8_t a) { return vget_low_bf16(a); } -// CHECK-LABEL: @test_vget_lane_bf16( -// CHECK-NEXT: entry: -// CHECK-NEXT: [[VGET_LANE:%.*]] = extractelement <4 x bfloat> [[V:%.*]], i32 1 -// CHECK-NEXT: ret bfloat [[VGET_LANE]] -// -bfloat16_t test_vget_lane_bf16(bfloat16x4_t v) { - return vget_lane_bf16(v, 1); -} - -// CHECK-LABEL: @test_vgetq_lane_bf16( -// CHECK-NEXT: entry: -// CHECK-NEXT: [[VGETQ_LANE:%.*]] = extractelement <8 x bfloat> [[V:%.*]], i32 7 -// CHECK-NEXT: ret bfloat [[VGETQ_LANE]] -// -bfloat16_t test_vgetq_lane_bf16(bfloat16x8_t v) { - return vgetq_lane_bf16(v, 7); -} - // CHECK-LABEL: @test_vset_lane_bf16( // CHECK-NEXT: entry: // CHECK-NEXT: [[VSET_LANE:%.*]] = insertelement <4 x bfloat> [[V:%.*]], bfloat [[A:%.*]], i32 1 diff --git a/clang/test/CodeGen/AArch64/neon/bf16-getset.c b/clang/test/CodeGen/AArch64/neon/bf16-getset.c index eeb9209fe97a5..563110c72ba21 100644 --- a/clang/test/CodeGen/AArch64/neon/bf16-getset.c +++ b/clang/test/CodeGen/AArch64/neon/bf16-getset.c @@ -120,3 +120,19 @@ bfloat16x8_t test_vdupq_laneq_bf16(bfloat16x8_t v) { // LLVM-NEXT: ret <8 x bfloat> [[VECINIT7_I]] return vdupq_n_bf16(v); } + +// ALL-LABEL: @test_vget_lane_bf16( +bfloat16_t test_vget_lane_bf16(bfloat16x4_t v) { + // CIR: cir.vec.extract %{{.*}}[%{{.*}} : !s32i] : !cir.vector<4 x !cir.bf16> + // LLVM: [[VGET_LANE:%.*]] = extractelement <4 x bfloat> %{{.*}}, i32 1 + // LLVM: ret bfloat [[VGET_LANE]] + return vget_lane_bf16(v, 1); +} + +// ALL-LABEL: @test_vgetq_lane_bf16( +bfloat16_t test_vgetq_lane_bf16(bfloat16x8_t v) { + // CIR: cir.vec.extract %{{.*}}[%{{.*}} : !s32i] : !cir.vector<8 x !cir.bf16> + // LLVM: [[VGETQ_LANE:%.*]] = extractelement <8 x bfloat> %{{.*}}, i32 7 + // LLVM: ret bfloat [[VGETQ_LANE]] + return vgetq_lane_bf16(v, 7); +} From 0e78e81670dc89b782f53aebb14062fe78232cbe Mon Sep 17 00:00:00 2001 From: Jiri Filek <[email protected]> Date: Sun, 29 Mar 2026 19:03:14 +0200 Subject: [PATCH 2/2] Add vget_high_bf16 and vget_lane_bf16 tests --- .../CodeGen/AArch64/bf16-getset-intrinsics.c | 18 ----------- clang/test/CodeGen/AArch64/neon/bf16-getset.c | 32 +++++++++++++++++-- 2 files changed, 30 insertions(+), 20 deletions(-) diff --git a/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c b/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c index 19819d0d6c8c8..89b332d14d588 100644 --- a/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c +++ b/clang/test/CodeGen/AArch64/bf16-getset-intrinsics.c @@ -23,24 +23,6 @@ bfloat16x8_t test_vcombine_bf16(bfloat16x4_t low, bfloat16x4_t high) { return vcombine_bf16(low, high); } -// CHECK-LABEL: @test_vget_high_bf16( -// CHECK-NEXT: entry: -// CHECK-NEXT: [[SHUFFLE_I:%.*]] = shufflevector <8 x bfloat> [[A:%.*]], <8 x bfloat> [[A]], <4 x i32> <i32 4, i32 5, i32 6, i32 7> -// CHECK-NEXT: ret <4 x bfloat> [[SHUFFLE_I]] -// -bfloat16x4_t test_vget_high_bf16(bfloat16x8_t a) { - return vget_high_bf16(a); -} - -// CHECK-LABEL: @test_vget_low_bf16( -// CHECK-NEXT: entry: -// CHECK-NEXT: [[SHUFFLE_I:%.*]] = shufflevector <8 x bfloat> [[A:%.*]], <8 x bfloat> [[A]], <4 x i32> <i32 0, i32 1, i32 2, i32 3> -// CHECK-NEXT: ret <4 x bfloat> [[SHUFFLE_I]] -// -bfloat16x4_t test_vget_low_bf16(bfloat16x8_t a) { - return vget_low_bf16(a); -} - // CHECK-LABEL: @test_vset_lane_bf16( // CHECK-NEXT: entry: // CHECK-NEXT: [[VSET_LANE:%.*]] = insertelement <4 x bfloat> [[V:%.*]], bfloat [[A:%.*]], i32 1 diff --git a/clang/test/CodeGen/AArch64/neon/bf16-getset.c b/clang/test/CodeGen/AArch64/neon/bf16-getset.c index 563110c72ba21..f860bfb7e40aa 100644 --- a/clang/test/CodeGen/AArch64/neon/bf16-getset.c +++ b/clang/test/CodeGen/AArch64/neon/bf16-getset.c @@ -121,10 +121,36 @@ bfloat16x8_t test_vdupq_laneq_bf16(bfloat16x8_t v) { return vdupq_n_bf16(v); } +//===------------------------------------------------------===// +// 2.14.1.4 Split vectors +//===------------------------------------------------------===// + +// ALL-LABEL: @test_vget_high_bf16( +bfloat16x4_t test_vget_high_bf16(bfloat16x8_t a) { + // CIR: cir.call @vget_high_bf16({{%.*}}) : (!cir.vector<8 x !cir.bf16> + + // LLVM-SAME: <8 x bfloat> {{.*}} [[A:%.*]]) + // LLVM: [[SHUFFLE_I:%.*]] = shufflevector <8 x bfloat> [[A:%.*]], <8 x bfloat> [[A]], <4 x i32> <i32 4, i32 5, i32 6, i32 7> + // LLVM: ret <4 x bfloat> [[SHUFFLE_I]] + return vget_high_bf16(a); +} + +// ALL-LABEL: @test_vget_low_bf16( +bfloat16x4_t test_vget_low_bf16(bfloat16x8_t a) { + // CIR: cir.call @vget_low_bf16({{%.*}}) : (!cir.vector<8 x !cir.bf16> + + // LLVM-SAME: <8 x bfloat> {{.*}} [[A:%.*]]) + // LLVM: [[SHUFFLE_I:%.*]] = shufflevector <8 x bfloat> [[A:%.*]], <8 x bfloat> [[A]], <4 x i32> <i32 0, i32 1, i32 2, i32 3> + // LLVM: ret <4 x bfloat> [[SHUFFLE_I]] + return vget_low_bf16(a); +} + // ALL-LABEL: @test_vget_lane_bf16( bfloat16_t test_vget_lane_bf16(bfloat16x4_t v) { // CIR: cir.vec.extract %{{.*}}[%{{.*}} : !s32i] : !cir.vector<4 x !cir.bf16> - // LLVM: [[VGET_LANE:%.*]] = extractelement <4 x bfloat> %{{.*}}, i32 1 + + // LLVM-SAME: <4 x bfloat> {{.*}} [[V:%.*]]) + // LLVM: [[VGET_LANE:%.*]] = extractelement <4 x bfloat> [[V]], i32 1 // LLVM: ret bfloat [[VGET_LANE]] return vget_lane_bf16(v, 1); } @@ -132,7 +158,9 @@ bfloat16_t test_vget_lane_bf16(bfloat16x4_t v) { // ALL-LABEL: @test_vgetq_lane_bf16( bfloat16_t test_vgetq_lane_bf16(bfloat16x8_t v) { // CIR: cir.vec.extract %{{.*}}[%{{.*}} : !s32i] : !cir.vector<8 x !cir.bf16> - // LLVM: [[VGETQ_LANE:%.*]] = extractelement <8 x bfloat> %{{.*}}, i32 7 + + // LLVM-SAME: <8 x bfloat> {{.*}} [[V:%.*]]) + // LLVM: [[VGETQ_LANE:%.*]] = extractelement <8 x bfloat> [[V]], i32 7 // LLVM: ret bfloat [[VGETQ_LANE]] return vgetq_lane_bf16(v, 7); } _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
