Skip to content

Commit

Permalink
[Clang][SVE2.1] Add builtins for 2-way svdot (vectors, indexed)
Browse files Browse the repository at this point in the history 
As described in: ARM-software/acle#257

Patch by: David Sherwood <david.sherwood@arm.com>

Reviewed By: dtemirbulatov

Differential Revision: https://reviews.llvm.org/D151439
  • Loading branch information
@CarolineConcatto
CarolineConcatto committed Oct 19, 2023
1 parent 03d1c99 commit 9f93a99
Show file tree
Hide file tree
Showing 3 changed files with 123 additions and 0 deletions.
7 changes: 7 additions & 0 deletions clang/include/clang/Basic/arm_sve.td
View file
Expand Up @@ -1952,6 +1952,13 @@ def SVSTNT1B_VNUM_X4 : MInst<"svstnt1_vnum[_{2}_x4]", "v}pl4", "cUc", [IsStructS
def SVSTNT1H_VNUM_X4 : MInst<"svstnt1_vnum[_{2}_x4]", "v}pl4", "sUshb", [IsStructStore], MemEltTyDefault, "aarch64_sve_stnt1_pn_x4">;
def SVSTNT1W_VNUM_X4 : MInst<"svstnt1_vnum[_{2}_x4]", "v}pl4", "iUif", [IsStructStore], MemEltTyDefault, "aarch64_sve_stnt1_pn_x4">;
def SVSTNT1D_VNUM_X4 : MInst<"svstnt1_vnum[_{2}_x4]", "v}pl4", "lUld", [IsStructStore], MemEltTyDefault, "aarch64_sve_stnt1_pn_x4">;

def SVDOT_X2_S : SInst<"svdot[_{d}_{2}_{3}]", "ddhh", "i", MergeNone, "aarch64_sve_sdot_x2", [], []>;
def SVDOT_X2_U : SInst<"svdot[_{d}_{2}_{3}]", "ddhh", "Ui", MergeNone, "aarch64_sve_udot_x2", [], []>;
def SVDOT_X2_F : SInst<"svdot[_{d}_{2}_{3}]", "ddhh", "f", MergeNone, "aarch64_sve_fdot_x2", [], []>;
def SVDOT_LANE_X2_S : SInst<"svdot_lane[_{d}_{2}_{3}]", "ddhhi", "i", MergeNone, "aarch64_sve_sdot_lane_x2", [], [ImmCheck<3, ImmCheck0_3>]>;
def SVDOT_LANE_X2_U : SInst<"svdot_lane[_{d}_{2}_{3}]", "ddhhi", "Ui", MergeNone, "aarch64_sve_udot_lane_x2", [], [ImmCheck<3, ImmCheck0_3>]>;
def SVDOT_LANE_X2_F : SInst<"svdot_lane[_{d}_{2}_{3}]", "ddhhi", "f", MergeNone, "aarch64_sve_fdot_lane_x2", [], [ImmCheck<3, ImmCheck0_3>]>;
}

let TargetGuard = "sve2p1" in {
Expand Down
107 changes: 107 additions & 0 deletions clang/test/CodeGen/aarch64-sve2p1-intrinsics/acle_sve2p1_dot.c
View file
@@ -0,0 +1,107 @@
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// REQUIRES: aarch64-registered-target
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve2p1 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
#include <arm_sve.h>

#ifdef SVE_OVERLOADED_FORMS
// A simple used,unused... macro, long enough to represent any SVE builtin.
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3) A1##A3
#else
#define SVE_ACLE_FUNC(A1,A2,A3) A1##A2##A3
#endif

// CHECK-LABEL: @test_svdot_s32_x2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z17test_svdot_s32_x2u11__SVInt32_tu11__SVInt16_tu11__SVInt16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svint32_t test_svdot_s32_x2(svint32_t op1, svint16_t op2, svint16_t op3)
{
return SVE_ACLE_FUNC(svdot,_s32_s16_s16,)(op1, op2, op3);
}

// CHECK-LABEL: @test_svdot_u32_x2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.udot.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z17test_svdot_u32_x2u12__SVUint32_tu12__SVUint16_tu12__SVUint16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.udot.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svuint32_t test_svdot_u32_x2(svuint32_t op1, svuint16_t op2, svuint16_t op3)
{
return SVE_ACLE_FUNC(svdot,_u32_u16_u16,)(op1, op2, op3);
}

// CHECK-LABEL: @test_svdot_f32_x2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.fdot.x2.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[OP3:%.*]])
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z17test_svdot_f32_x2u13__SVFloat32_tu13__SVFloat16_tu13__SVFloat16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.fdot.x2.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[OP3:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
svfloat32_t test_svdot_f32_x2(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
{
return SVE_ACLE_FUNC(svdot,_f32_f16_f16,)(op1, op2, op3);
}



// CHECK-LABEL: @test_svdot_lane_s32_x2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.lane.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 3)
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z22test_svdot_lane_s32_x2u11__SVInt32_tu11__SVInt16_tu11__SVInt16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.lane.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 3)
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svint32_t test_svdot_lane_s32_x2(svint32_t op1, svint16_t op2, svint16_t op3)
{
return SVE_ACLE_FUNC(svdot_lane,_s32_s16_s16,)(op1, op2, op3, 3);
}

// CHECK-LABEL: @test_svdot_lane_u32_x2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.udot.lane.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 3)
// CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z22test_svdot_lane_u32_x2u12__SVUint32_tu12__SVUint16_tu12__SVUint16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.aarch64.sve.udot.lane.x2.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[OP3:%.*]], i32 3)
// CPP-CHECK-NEXT: ret <vscale x 4 x i32> [[TMP0]]
//
svuint32_t test_svdot_lane_u32_x2(svuint32_t op1, svuint16_t op2, svuint16_t op3)
{
return SVE_ACLE_FUNC(svdot_lane,_u32_u16_u16,)(op1, op2, op3, 3);
}

// CHECK-LABEL: @test_svdot_lane_f32_x2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.fdot.lane.x2.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[OP3:%.*]], i32 3)
// CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z22test_svdot_lane_f32_x2u13__SVFloat32_tu13__SVFloat16_tu13__SVFloat16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x float> @llvm.aarch64.sve.fdot.lane.x2.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[OP3:%.*]], i32 3)
// CPP-CHECK-NEXT: ret <vscale x 4 x float> [[TMP0]]
//
svfloat32_t test_svdot_lane_f32_x2(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
{
return SVE_ACLE_FUNC(svdot_lane,_f32_f16_f16,)(op1, op2, op3, 3);
}
9 changes: 9 additions & 0 deletions clang/test/Sema/aarch64-sve2p1-intrinsics/acle_sve2p1_imm.cpp
View file
Expand Up @@ -107,3 +107,12 @@ void test_cntp(svcount_t c) {
svcntp_c14(c, 3); // expected-error {{argument should be a multiple of 2}}
}

void test_svdot_lane_2way(svint32_t s32, svuint32_t u32, svint16_t s16, svuint16_t u16,
svfloat32_t f32, svfloat16_t f16) {
svdot_lane_s32_s16_s16(s32, s16, s16, 1); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
svdot_lane_u32_u16_u16(u32, u16, u16, 1); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
svdot_lane_f32_f16_f16(f32, f16, f16, 1); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
svdot_lane_s32_s16_s16(s32, s16, s16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
svdot_lane_u32_u16_u16(u32, u16, u16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
svdot_lane_f32_f16_f16(f32, f16, f16, 4); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
}

0 comments on commit 9f93a99

Please sign in to comment.