Skip to content

Commit d1652ff

Browse files
kmclaughlin-armkmclaughlin-arm
authored
[Clang][SME2] Add outer product and accumulate/subtract builtins (#71176)
Adds the following SME2 builtins: - svmop(a|s)_za32, - svbmop(a|s)_za32 See ARM-software/acle#217
1 parent 1726a59 commit d1652ff

File tree

4 files changed

+225
-0
lines changed

4 files changed

+225
-0
lines changed

‎clang/include/clang/Basic/arm_sme.td

Lines changed: 16 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -298,3 +298,19 @@ multiclass ZAAddSub<string n_suffix> {
298298

299299
defm SVADD : ZAAddSub<"add">;
300300
defm SVSUB : ZAAddSub<"sub">;
301+
302+
//
303+
// Outer product and accumulate/subtract
304+
//
305+
306+
let TargetGuard = "sme2" in {
307+
def SVSMOPA : Inst<"svmopa_za32[_{d}]_m", "viPPdd", "s", MergeNone, "aarch64_sme_smopa_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
308+
def SVUSMOPA : Inst<"svmopa_za32[_{d}]_m", "viPPdd", "Us", MergeNone, "aarch64_sme_umopa_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
309+
310+
def SVSMOPS : Inst<"svmops_za32[_{d}]_m", "viPPdd", "s", MergeNone, "aarch64_sme_smops_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
311+
def SVUSMOPS : Inst<"svmops_za32[_{d}]_m", "viPPdd", "Us", MergeNone, "aarch64_sme_umops_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
312+
313+
def SVBMOPA : Inst<"svbmopa_za32[_{d}]_m", "viPPdd", "iUi", MergeNone, "aarch64_sme_bmopa_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
314+
315+
def SVBMOPS : Inst<"svbmops_za32[_{d}]_m", "viPPdd", "iUi", MergeNone, "aarch64_sme_bmops_za32", [IsSharedZA, IsStreaming], [ImmCheck<0, ImmCheck0_3>]>;
316+
}

‎clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_bmop.c

Lines changed: 94 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,94 @@
1+
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2+
3+
// REQUIRES: aarch64-registered-target
4+
5+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
6+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
7+
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
8+
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
9+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
10+
11+
#include <arm_sme_draft_spec_subject_to_change.h>
12+
13+
#ifdef SVE_OVERLOADED_FORMS
14+
// A simple used,unused... macro, long enough to represent any SVE builtin.
15+
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
16+
#else
17+
#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
18+
#endif
19+
20+
// BMOPA
21+
22+
// CHECK-LABEL: @test_svbmopa_u32(
23+
// CHECK-NEXT: entry:
24+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
25+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
26+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.bmopa.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
27+
// CHECK-NEXT: ret void
28+
//
29+
// CPP-CHECK-LABEL: @_Z16test_svbmopa_u32u10__SVBool_tS_u12__SVUint32_tS0_(
30+
// CPP-CHECK-NEXT: entry:
31+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
32+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
33+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.bmopa.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
34+
// CPP-CHECK-NEXT: ret void
35+
//
36+
void test_svbmopa_u32(svbool_t pn, svbool_t pm, svuint32_t zn, svuint32_t zm) __arm_streaming __arm_shared_za {
37+
SVE_ACLE_FUNC(svbmopa_za32,_u32,_m,)(3, pn, pm, zn, zm);
38+
}
39+
40+
// CHECK-LABEL: @test_svbmopa_s32(
41+
// CHECK-NEXT: entry:
42+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
43+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
44+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.bmopa.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
45+
// CHECK-NEXT: ret void
46+
//
47+
// CPP-CHECK-LABEL: @_Z16test_svbmopa_s32u10__SVBool_tS_u11__SVInt32_tS0_(
48+
// CPP-CHECK-NEXT: entry:
49+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
50+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
51+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.bmopa.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
52+
// CPP-CHECK-NEXT: ret void
53+
//
54+
void test_svbmopa_s32(svbool_t pn, svbool_t pm, svint32_t zn, svint32_t zm) __arm_streaming __arm_shared_za {
55+
SVE_ACLE_FUNC(svbmopa_za32,_s32,_m,)(3, pn, pm, zn, zm);
56+
}
57+
58+
// BMOPS
59+
60+
// CHECK-LABEL: @test_svbmops_u32(
61+
// CHECK-NEXT: entry:
62+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
63+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
64+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.bmops.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
65+
// CHECK-NEXT: ret void
66+
//
67+
// CPP-CHECK-LABEL: @_Z16test_svbmops_u32u10__SVBool_tS_u12__SVUint32_tS0_(
68+
// CPP-CHECK-NEXT: entry:
69+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
70+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
71+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.bmops.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
72+
// CPP-CHECK-NEXT: ret void
73+
//
74+
void test_svbmops_u32(svbool_t pn, svbool_t pm, svuint32_t zn, svuint32_t zm) __arm_streaming __arm_shared_za {
75+
SVE_ACLE_FUNC(svbmops_za32,_u32,_m,)(3, pn, pm, zn, zm);
76+
}
77+
78+
// CHECK-LABEL: @test_svbmops_s32(
79+
// CHECK-NEXT: entry:
80+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
81+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
82+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.bmops.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
83+
// CHECK-NEXT: ret void
84+
//
85+
// CPP-CHECK-LABEL: @_Z16test_svbmops_s32u10__SVBool_tS_u11__SVInt32_tS0_(
86+
// CPP-CHECK-NEXT: entry:
87+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PN:%.*]])
88+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PM:%.*]])
89+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.bmops.za32.nxv4i32(i32 3, <vscale x 4 x i1> [[TMP0]], <vscale x 4 x i1> [[TMP1]], <vscale x 4 x i32> [[ZN:%.*]], <vscale x 4 x i32> [[ZM:%.*]])
90+
// CPP-CHECK-NEXT: ret void
91+
//
92+
void test_svbmops_s32(svbool_t pn, svbool_t pm, svint32_t zn, svint32_t zm) __arm_streaming __arm_shared_za {
93+
SVE_ACLE_FUNC(svbmops_za32,_s32,_m,)(3, pn, pm, zn, zm);
94+
}

‎clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_mop.c

Lines changed: 94 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,94 @@
1+
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
2+
3+
// REQUIRES: aarch64-registered-target
4+
5+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
6+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
7+
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s
8+
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK
9+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sme2 -target-feature +sve -S -disable-O0-optnone -Werror -Wall -o /dev/null %s
10+
11+
#include <arm_sme_draft_spec_subject_to_change.h>
12+
13+
#ifdef SVE_OVERLOADED_FORMS
14+
// A simple used,unused... macro, long enough to represent any SVE builtin.
15+
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
16+
#else
17+
#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
18+
#endif
19+
20+
// MOPA
21+
22+
// CHECK-LABEL: @test_svmopa_s16(
23+
// CHECK-NEXT: entry:
24+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
25+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
26+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.smopa.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
27+
// CHECK-NEXT: ret void
28+
//
29+
// CPP-CHECK-LABEL: @_Z15test_svmopa_s16u10__SVBool_tS_u11__SVInt16_tS0_(
30+
// CPP-CHECK-NEXT: entry:
31+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
32+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
33+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.smopa.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
34+
// CPP-CHECK-NEXT: ret void
35+
//
36+
void test_svmopa_s16(svbool_t pn, svbool_t pm, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
37+
SVE_ACLE_FUNC(svmopa_za32,_s16,_m,)(3, pn, pm, zn, zm);
38+
}
39+
40+
// CHECK-LABEL: @test_svmopa_u16(
41+
// CHECK-NEXT: entry:
42+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
43+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
44+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.umopa.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
45+
// CHECK-NEXT: ret void
46+
//
47+
// CPP-CHECK-LABEL: @_Z15test_svmopa_u16u10__SVBool_tS_u12__SVUint16_tS0_(
48+
// CPP-CHECK-NEXT: entry:
49+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
50+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
51+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.umopa.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
52+
// CPP-CHECK-NEXT: ret void
53+
//
54+
void test_svmopa_u16(svbool_t pn, svbool_t pm, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
55+
SVE_ACLE_FUNC(svmopa_za32,_u16,_m,)(3, pn, pm, zn, zm);
56+
}
57+
58+
// MOPS
59+
60+
// CHECK-LABEL: @test_svmops_s16(
61+
// CHECK-NEXT: entry:
62+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
63+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
64+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.smops.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
65+
// CHECK-NEXT: ret void
66+
//
67+
// CPP-CHECK-LABEL: @_Z15test_svmops_s16u10__SVBool_tS_u11__SVInt16_tS0_(
68+
// CPP-CHECK-NEXT: entry:
69+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
70+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
71+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.smops.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
72+
// CPP-CHECK-NEXT: ret void
73+
//
74+
void test_svmops_s16(svbool_t pn, svbool_t pm, svint16_t zn, svint16_t zm) __arm_streaming __arm_shared_za {
75+
SVE_ACLE_FUNC(svmops_za32,_s16,_m,)(3, pn, pm, zn, zm);
76+
}
77+
78+
// CHECK-LABEL: @test_svmops_u16(
79+
// CHECK-NEXT: entry:
80+
// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
81+
// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
82+
// CHECK-NEXT: tail call void @llvm.aarch64.sme.umops.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
83+
// CHECK-NEXT: ret void
84+
//
85+
// CPP-CHECK-LABEL: @_Z15test_svmops_u16u10__SVBool_tS_u12__SVUint16_tS0_(
86+
// CPP-CHECK-NEXT: entry:
87+
// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PN:%.*]])
88+
// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PM:%.*]])
89+
// CPP-CHECK-NEXT: tail call void @llvm.aarch64.sme.umops.za32.nxv8i16(i32 3, <vscale x 8 x i1> [[TMP0]], <vscale x 8 x i1> [[TMP1]], <vscale x 8 x i16> [[ZN:%.*]], <vscale x 8 x i16> [[ZM:%.*]])
90+
// CPP-CHECK-NEXT: ret void
91+
//
92+
void test_svmops_u16(svbool_t pn, svbool_t pm, svuint16_t zn, svuint16_t zm) __arm_streaming __arm_shared_za {
93+
SVE_ACLE_FUNC(svmops_za32,_u16,_m,)(3, pn, pm, zn, zm);
94+
}

‎clang/test/Sema/aarch64-sme2-intrinsics/acle_sme2_imm.cpp

Lines changed: 21 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,21 @@
1+
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu \
2+
// RUN: -target-feature +sve2 -target-feature +sme2 -target-feature +sve -fsyntax-only -verify %s
3+
4+
// REQUIRES: aarch64-registered-target
5+
6+
#include <arm_sme_draft_spec_subject_to_change.h>
7+
8+
void test_outer_product(svbool_t pred, svint16_t s16, svuint16_t u16, svint32_t s32, svuint32_t u32) __arm_streaming __arm_shared_za {
9+
// Test Tile Range
10+
svmopa_za32_u16_m(4, pred, pred, u16, u16); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
11+
svmopa_za32_s16_m(4, pred, pred, s16, s16); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
12+
13+
svmops_za32_u16_m(4, pred, pred, u16, u16); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
14+
svmops_za32_s16_m(4, pred, pred, s16, s16); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
15+
16+
svbmopa_za32_u32_m(4, pred, pred, u32, u32); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
17+
svbmopa_za32_s32_m(4, pred, pred, s32, s32); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
18+
19+
svbmops_za32_u32_m(4, pred, pred, u32, u32); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
20+
svbmops_za32_s32_m(4, pred, pred, s32, s32); // expected-error {{argument value 4 is outside the valid range [0, 3]}}
21+
}

0 commit comments

Comments
 (0)