Skip to content

[SYCL-PTX] Add _Float16 as native type #1848

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.

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

Merged
merged 2 commits into from
Jun 11, 2020
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions clang/lib/Basic/Targets/NVPTX.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -64,6 +64,8 @@ NVPTXTargetInfo::NVPTXTargetInfo(const llvm::Triple &Triple,
VLASupported = false;
AddrSpaceMap = &NVPTXAddrSpaceMap;
UseAddrSpaceMapMangling = true;
HasLegalHalfType = true;
HasFloat16 = true;

// Define available target features
// These must be defined in sorted order!
Expand Down
7 changes: 1 addition & 6 deletions clang/lib/Sema/Sema.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -1753,12 +1753,7 @@ void Sema::checkDeviceDecl(const ValueDecl *D, SourceLocation Loc) {
if (Ty->isDependentType())
return;

auto IsSYCLDeviceCuda = getLangOpts().SYCLIsDevice &&
Context.getTargetInfo().getTriple().isNVPTX();
if ((Ty->isFloat16Type() && !Context.getTargetInfo().hasFloat16Type() &&
// Disable check for SYCL CUDA BE until FP16 support is properly
// reported there (issue#1799)
!IsSYCLDeviceCuda) ||
if ((Ty->isFloat16Type() && !Context.getTargetInfo().hasFloat16Type()) ||
((Ty->isFloat128Type() ||
(Ty->isRealFloatingType() && Context.getTypeSize(Ty) == 128)) &&
!Context.getTargetInfo().hasFloat128Type()) ||
Expand Down
9 changes: 4 additions & 5 deletions clang/lib/Sema/SemaType.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -1521,12 +1521,11 @@ static QualType ConvertDeclSpecToType(TypeProcessingState &state) {
// CUDA host and device may have different _Float16 support, therefore
// do not diagnose _Float16 usage to avoid false alarm.
// ToDo: more precise diagnostics for CUDA.
auto IsSYCLDeviceCuda =
S.getLangOpts().SYCLIsDevice && S.Context.getTargetInfo().getTriple().isNVPTX();
if (!S.Context.getTargetInfo().hasFloat16Type() && !S.getLangOpts().CUDA &&
!(S.getLangOpts().OpenMP && S.getLangOpts().OpenMPIsDevice) && !IsSYCLDeviceCuda)
if (!S.Context.getTargetInfo().hasFloat16Type() &&
!S.getLangOpts().CUDA &&
!(S.getLangOpts().OpenMP && S.getLangOpts().OpenMPIsDevice))
S.Diag(DS.getTypeSpecTypeLoc(), diag::err_type_unsupported)
<< "_Float16";
<< "_Float16";
}
Result = Context.Float16Ty;
break;
Expand Down
22 changes: 22 additions & 0 deletions clang/test/CodeGenCXX/nvptx-float16.cpp
Original file line number Diff line number Diff line change
@@ -0,0 +1,22 @@
// REQUIRES: nvptx-registered-target
// RUN: %clang_cc1 -triple nvptx-nvidia-cuda -target-cpu sm_20 -S -o - %s | FileCheck %s -check-prefix=NOF16
// RUN: %clang_cc1 -triple nvptx-nvidia-cuda -target-cpu sm_60 -S -o - %s | FileCheck %s

// CHECK: .target sm_60
// NOF16: .target sm_20

void f() {
_Float16 x, y, z;
// CHECK: add.rn.f16
// NOF16: add.rn.f32
z = x + y;
// CHECK: sub.rn.f16
// NOF16: sub.rn.f32
z = x - y;
// CHECK: mul.rn.f16
// NOF16: mul.rn.f32
z = x * y;
// CHECK: div.rn.f32
// NOF16: div.rn.f32
z = x / y;
}