Author: yaxunl
Date: Tue Jul 18 07:46:03 2017
New Revision: 308313
URL: http://llvm.org/viewvc/llvm-project?rev=308313&view=rev
Log:
CodeGen: Insert addr space cast for automatic/temp var at right position
The uses of alloca may be in different blocks other than the block containing
the alloca.
Author: yaxunl
Date: Fri Feb 2 08:08:24 2018
New Revision: 324102
URL: http://llvm.org/viewvc/llvm-project?rev=324102&view=rev
Log:
[AMDGPU] Switch to the new addr space mapping by default
This requires corresponding llvm change.
Differential Revision: https://reviews.llvm.org/D40956
Modified:
Author: yaxunl
Date: Thu Mar 9 19:30:46 2017
New Revision: 297436
URL: http://llvm.org/viewvc/llvm-project?rev=297436&view=rev
Log:
[AMDGPU] Add builtin functions readlane ds_permute mov_dpp
Differential Revision: https://reviews.llvm.org/D30551
Modified:
cfe/trunk/include/clang/Basic/Built
yaxunl marked 3 inline comments as done.
yaxunl added a comment.
> But if you do need to support these conversions for some reason, the correct
> behavior is to ensure that null is mapped to null.
I only need to do this for constant folding, right? I found that the current
implementation is alr
Author: yaxunl
Date: Fri Dec 9 13:01:11 2016
New Revision: 289252
URL: http://llvm.org/viewvc/llvm-project?rev=289252&view=rev
Log:
Add support for non-zero null pointer for C and OpenCL
In amdgcn target, null pointers in global, constant, and generic address space
take value 0 but null pointer
Author: yaxunl
Date: Mon May 15 09:47:47 2017
New Revision: 303072
URL: http://llvm.org/viewvc/llvm-project?rev=303072&view=rev
Log:
[OpenCL] Emit function-scope variable in constant address space as static
variable
Differential Revision: https://reviews.llvm.org/D32977
Modified:
cfe/trunk/
Author: yaxunl
Date: Thu May 18 13:51:09 2017
New Revision: 303370
URL: http://llvm.org/viewvc/llvm-project?rev=303370&view=rev
Log:
CodeGen: Cast alloca to expected address space
Alloca always returns a pointer in alloca address space, which may
be different from the type defined by the language
Author: yaxunl
Date: Tue May 23 11:15:53 2017
New Revision: 303644
URL: http://llvm.org/viewvc/llvm-project?rev=303644&view=rev
Log:
[AMDGPU] Do not require opencl triple environment for OpenCL
A recent change requires opencl triple environment for compiling OpenCL
program, which causes regressio
yaxunl updated this revision to Diff 75203.
yaxunl marked 4 inline comments as done.
yaxunl added a comment.
Fix typo in test.
https://reviews.llvm.org/D25343
Files:
include/clang/Basic/Attr.td
include/clang/Basic/AttrDocs.td
lib/CodeGen/CGCall.cpp
lib/Headers/opencl-c.h
lib/Sema/Sema
yaxunl added a comment.
> Should we deprecate noduplicate then as convergent should cover both use
> cases for OpenCL I believe? As far as I understand noduplicate was added
> specifically for SPMD use cases...
noduplicate has different semantics than convergent. Although it is proposed
for SP
yaxunl created this revision.
yaxunl added reviewers: tstellarAMD, arsenm.
yaxunl added a subscriber: cfe-commits.
Herald added subscribers: tony-tye, wdng, kzhuravl.
Patch by Laurent Morichetti.
https://reviews.llvm.org/D25920
Files:
lib/Basic/Targets.cpp
Index: lib/Basic/Targets.cpp
=
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM. Thanks.
https://reviews.llvm.org/D25935
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listi
yaxunl created this revision.
yaxunl added a reviewer: Anastasia.
yaxunl added subscribers: cfe-commits, b-sumner.
https://reviews.llvm.org/D25954
Files:
lib/Headers/opencl-c.h
Index: lib/Headers/opencl-c.h
===
--- lib/Headers/op
Author: yaxunl
Date: Tue Oct 25 16:37:05 2016
New Revision: 285125
URL: http://llvm.org/viewvc/llvm-project?rev=285125&view=rev
Log:
[OpenCL] Add missing atom_xor for 64 bit to opencl-c.h
Differential Revision: https://reviews.llvm.org/D25954
Modified:
cfe/trunk/lib/Headers/opencl-c.h
Modif
This revision was automatically updated to reflect the committed changes.
Closed by commit rL285125: [OpenCL] Add missing atom_xor for 64 bit to
opencl-c.h (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D25954?vs=75740&id=75797#toc
Repository:
rL LLVM
https://review
This revision was automatically updated to reflect the committed changes.
Closed by commit rL285211: AMDGPU: Add missing ISA versions gfx7.x.x and 8.x.x.
(authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D25920?vs=75627&id=75912#toc
Repository:
rL LLVM
https://reviews.
Author: yaxunl
Date: Wed Oct 26 11:40:21 2016
New Revision: 285211
URL: http://llvm.org/viewvc/llvm-project?rev=285211&view=rev
Log:
AMDGPU: Add missing ISA versions gfx7.x.x and 8.x.x.
Patch by Laurent Morichetti.
Differential Revision: https://reviews.llvm.org/D25920
Modified:
cfe/trunk/l
yaxunl added inline comments.
Comment at: test/CodeGenOpenCL/convergent.cl:54
+// CHECK: tail call spir_func void @f()
+// CHECK-NOT: call spir_func void @non_convfun()
+// CHECK-NOT: call spir_func void @g()
Anastasia wrote:
> Did you mean to check @convfun() he
yaxunl added a comment.
LGTM. Thanks.
Repository:
rL LLVM
https://reviews.llvm.org/D26157
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
yaxunl created this revision.
yaxunl added reviewers: arsenm, tstellarAMD, rjmccall.
yaxunl added a subscriber: cfe-commits.
Herald added subscribers: tony-tye, nhaehnle, wdng, kzhuravl.
In amdgcn target, null pointers in global, constant, and generic address space
take value 0 but null pointers
Author: yaxunl
Date: Tue Nov 1 13:45:32 2016
New Revision: 285725
URL: http://llvm.org/viewvc/llvm-project?rev=285725&view=rev
Log:
[OpenCL] Mark group functions as convergent in opencl-c.h
Certain OpenCL builtin functions are supposed to be executed by all threads in
a work group or sub group.
This revision was automatically updated to reflect the committed changes.
Closed by commit rL285725: [OpenCL] Mark group functions as convergent in
opencl-c.h (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D25343?vs=75203&id=76612#toc
Repository:
rL LLVM
https://rev
yaxunl updated the summary for this revision.
yaxunl updated this revision to Diff 76792.
yaxunl added a comment.
Fixed translation of a pointer to bool. Added more tests.
https://reviews.llvm.org/D26196
Files:
lib/CodeGen/CGDecl.cpp
lib/CodeGen/CGExprConstant.cpp
lib/CodeGen/CGExprScalar
yaxunl added inline comments.
Comment at: lib/CodeGen/TargetInfo.cpp:7039
+ if (CGM.getTarget().getTriple().getArch() != llvm::Triple::amdgcn ||
+(AS != Ctx.getTargetAddressSpace(LangAS::opencl_local) && AS != 0))
+return C;
arsenm wrote:
> Shouldn't the
yaxunl added a comment.
One purpose of kernel arg info metadata is to be passed to OpenCL runtime to
fulfill clGetKernelArgInfo queries.
(https://www.khronos.org/registry/cl/sdk/1.2/docs/man/xhtml/clGetKernelArgInfo.html).
As such, the address space id needs to be at source level instead of IR
yaxunl added a comment.
The enum for address space is not necessarily bound to SPIR as long as the
runtime knows it.
Since most OpenCL runtime already uses that definition (if they use clang as
compiler), I don't see a reason to change that. I doubt this metadata can be
used for anything else
yaxunl created this revision.
yaxunl added a reviewer: Anastasia.
yaxunl added subscribers: cfe-commits, rsmith.
https://reviews.llvm.org/D26302
Files:
test/Headers/opencl-c-header.cl
Index: test/Headers/opencl-c-header.cl
===
--
yaxunl added a comment.
In https://reviews.llvm.org/D26196#586433, @rjmccall wrote:
> I think there are a number of conceptual questions here, chiefly among them
> whether an llvm::ConstantPointerNull in an address-space type should have the
> correct representation for the target. If the deci
yaxunl updated this revision to Diff 76932.
yaxunl added a comment.
Revised as Anastasia suggested.
https://reviews.llvm.org/D26302
Files:
test/Headers/opencl-c-header.cl
Index: test/Headers/opencl-c-header.cl
===
--- test/Head
yaxunl added a comment.
In https://reviews.llvm.org/D26302#587480, @Anastasia wrote:
> LGTM! Thanks! Do you know the runtime of this test now?
It takes about 1s in releaser build and 20s in debug build on my machine.
https://reviews.llvm.org/D26302
_
Author: yaxunl
Date: Mon Nov 7 09:55:51 2016
New Revision: 286114
URL: http://llvm.org/viewvc/llvm-project?rev=286114&view=rev
Log:
[OpenCL] Remove redundant test for OpenCL header file
Differential Revision: https://reviews.llvm.org/D26302
Modified:
cfe/trunk/test/Headers/opencl-c-header.cl
This revision was automatically updated to reflect the committed changes.
Closed by commit rL286114: [OpenCL] Remove redundant test for OpenCL header
file (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D26302?vs=76932&id=77039#toc
Repository:
rL LLVM
https://reviews
yaxunl added inline comments.
Comment at: test/SemaOpenCL/extensions.cl:28
+#ifdef FP64
+// expected-no-diagnostics
+#endif
expected-no-diagnostics applies to the whole file. better move to the beginning
of the file.
otherwise LGTM. Thanks!
https://reviews.ll
yaxunl retitled this revision from "AMDGPU: Translate null pointers in private
and local addr space" to "[WIP] Add support for non-zero null pointers".
yaxunl updated this revision to Diff 77097.
yaxunl added a comment.
This is work in progress. Revised by John's comments.
Refactored getNullPtr
yaxunl added a comment.
In https://reviews.llvm.org/D26157#589033, @pekka.jaaskelainen wrote:
> The uses for the OpenCL logical address spaces that I know of are:
>
> 1. to differentiate local kernel arguments as their memory allocation is
> different (per WG if parallel WGs)
> 2. alias analysis
yaxunl added inline comments.
Comment at: lib/CodeGen/CGExprConstant.cpp:1340
+ return C;
+return getNullPtr(PT, DestType);
}
rjmccall wrote:
> efriedma wrote:
> > Consider code like the following:
> >
> > int x = 0;
> > auto y1 =
yaxunl retitled this revision from "[WIP] Add support for non-zero null
pointers" to "Add support for non-zero null pointers".
yaxunl updated this revision to Diff 77210.
yaxunl added a comment.
Fixed list initialization of large struct which are mostly initialized by 0
through memset.
Added tes
yaxunl added inline comments.
Comment at: lib/CodeGen/CodeGenTypes.cpp:743
+auto NullPtr = CGM.getNullPtr(LLPT, T);
+return isa(NullPtr);
+ }
tony-tye wrote:
> Is this correct if the target does not represent a NULL pointer as the
> address with value 0
yaxunl updated this revision to Diff 77238.
yaxunl marked 2 inline comments as done.
yaxunl added a comment.
Added isNullPtrZero to TargetCodeGenInfo.
https://reviews.llvm.org/D26196
Files:
lib/CodeGen/CGDecl.cpp
lib/CodeGen/CGExprAgg.cpp
lib/CodeGen/CGExprConstant.cpp
lib/CodeGen/CGExp
yaxunl marked 3 inline comments as done.
yaxunl added inline comments.
Comment at: lib/CodeGen/CGExprConstant.cpp:1340
+ return C;
+return getNullPtr(PT, DestType);
}
yaxunl wrote:
> rjmccall wrote:
> > efriedma wrote:
> > > Consider code
yaxunl marked an inline comment as done.
yaxunl added inline comments.
Comment at: lib/CodeGen/CGExprConstant.cpp:1340
+ return C;
+return getNullPtr(PT, DestType);
}
efriedma wrote:
> yaxunl wrote:
> > yaxunl wrote:
> > > rjmccall wrote:
yaxunl marked 12 inline comments as done.
yaxunl added inline comments.
Comment at: lib/CodeGen/TargetInfo.cpp:7040
+ auto AS = PT->getAddressSpace();
+ return AS != Ctx.getTargetAddressSpace(LangAS::opencl_local) && AS != 0;
+}
rjmccall wrote:
> This check is de
yaxunl updated this revision to Diff 77525.
yaxunl marked an inline comment as done.
yaxunl added a comment.
Changed APValue to differentiate with null pointer and pointers casted from
integer. Fixed casting integer to pointer.
Refactored isNullPtrZero to use just QualType as parameter.
Added t
yaxunl added a comment.
Hi John,
It seems the casting from a pointer to different address space is not affected
by this change. When a null pointer is casted to different address space, it is
casted the same way as an ordinary pointer, e.g. by addrspacecast.
I think this patch covers most of t
yaxunl added a comment.
My main concern is that by making the number of event be of type size_t instead
of uint we are not conforming to the spec, although I agree it is better for
the user.
Comment at: lib/CodeGen/CGBuiltin.cpp:2547
+ llvm::Value *NumEvents =
+
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM. Thanks!
https://reviews.llvm.org/D26507
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listi
yaxunl added inline comments.
Comment at: test/CodeGenOpenCL/builtins-amdgcn-error-f16-div-fixup.cl:8
+{
+ *out = __builtin_amdgcn_div_fixuph(a, b, c); // expected-error
{{'__builtin_amdgcn_div_fixuph' needs target feature 16-bit-insts}}
+}
The tests for checki
yaxunl added a comment.
I thought the number of events was also changed to size_t type. Now I
understand. Probably I was misled by the summary:
"Parameters representing number of events as well as sizes of enqueued block
arguments in enqueue_kernel function are specified as uint type. This prev
yaxunl added a comment.
>> It seems the casting from a pointer to different address space is not
>> affected by this change. When a null pointer is casted to different address
>> space, it is casted the same way as an ordinary pointer, e.g. by
>> addrspacecast.
>
> You mean, the code-generatio
yaxunl added a comment.
>> It seems the casting from a pointer to different address space is not
>> affected by this change. When a null pointer is casted to different address
>> space, it is casted the same way as an ordinary pointer, e.g. by
>> addrspacecast.
>
> You mean, the code-generatio
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM with the change for the summary. Thanks!
https://reviews.llvm.org/D26509
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://list
yaxunl updated this revision to Diff 64529.
yaxunl marked 7 inline comments as done.
yaxunl added a comment.
Revised by Anastasia's comments.
https://reviews.llvm.org/D21567
Files:
include/clang/AST/OperationKinds.def
include/clang/Basic/DiagnosticGroups.td
include/clang/Basic/DiagnosticS
yaxunl added inline comments.
Comment at: lib/Sema/SemaInit.cpp:6917-6929
@@ -6915,4 +6916,15 @@
<< SourceType;
- } else if (Entity.getKind() != InitializedEntity::EK_Variable) {
-llvm_unreachable("Invalid EntityKind!");
+ break;
+} else
yaxunl added inline comments.
Comment at: test/CodeGenOpenCL/opencl_types.cl:1
@@ -1,1 +1,2 @@
-// RUN: %clang_cc1 %s -emit-llvm -o - -O0 | FileCheck %s
+// RUN: %clang_cc1 %s -emit-llvm -o - -O0 | FileCheck %s
--check-prefix=CHECK-NORMAL
+// RUN: %clang_cc1 %s -triple "amdgcn--a
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM. Thanks!
Repository:
rL LLVM
https://reviews.llvm.org/D22523
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.or
This revision was automatically updated to reflect the committed changes.
Closed by commit rL276010: [OpenCL] Fixes bug of missing OCL version metadata
on the AMDGCN target (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22424?vs=64513&id=64545#toc
Repository:
rL LLV
Author: yaxunl
Date: Tue Jul 19 14:39:45 2016
New Revision: 276010
URL: http://llvm.org/viewvc/llvm-project?rev=276010&view=rev
Log:
[OpenCL] Fixes bug of missing OCL version metadata on the AMDGCN target
Added the opencl.ocl.version metadata to be emitted with amdgcn. Created a
static function
yaxunl added inline comments.
Comment at: lib/Sema/SemaInit.cpp:6945
@@ +6944,3 @@
+ // get the integer literal.
+ Init = cast(const_cast(
+Var->getInit()))->getSubExpr();
Anastasia wrote:
> What if global variable sampler is initiali
yaxunl updated this revision to Diff 64724.
yaxunl added a comment.
Added more diagnostic tests.
https://reviews.llvm.org/D21567
Files:
include/clang/AST/OperationKinds.def
include/clang/Basic/DiagnosticGroups.td
include/clang/Basic/DiagnosticSemaKinds.td
lib/AST/ASTContext.cpp
lib/AS
Author: yaxunl
Date: Wed Jul 20 14:21:11 2016
New Revision: 276161
URL: http://llvm.org/viewvc/llvm-project?rev=276161&view=rev
Log:
[OpenCL] AMDGCN target will generate images in constant address space
Allows AMDGCN target to generate images (such as %opencl.image2d_t) in constant
address space
This revision was automatically updated to reflect the committed changes.
Closed by commit rL276161: [OpenCL] AMDGCN target will generate images in
constant address space (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22523?vs=64706&id=64729#toc
Repository:
rL LLVM
yaxunl updated this revision to Diff 64927.
yaxunl marked 4 inline comments as done.
yaxunl added a comment.
Update the test to accommodate changes in branch tip.
https://reviews.llvm.org/D21567
Files:
include/clang/AST/OperationKinds.def
include/clang/Basic/DiagnosticGroups.td
include/cl
yaxunl added a comment.
Need to add a test to Misc/amdgcn.languageOptsOpenCL.cl and
SemaOpenCL/extension-version.cl
Repository:
rL LLVM
https://reviews.llvm.org/D22637
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org
yaxunl added inline comments.
Comment at: include/clang/Basic/OpenCLExtensions.def:73
@@ -72,2 +72,3 @@
OPENCLEXT_INTERNAL(cl_khr_terminate_context, 200, ~0U)
+OPENCLEXT_INTERNAL(cl_khr_mipmap_image, 200, ~0U)
Better to follow the alphabetical order.
Reposito
yaxunl added inline comments.
Comment at: test/Misc/amdgcn.languageOptsOpenCL.cl:190
@@ +189,3 @@
+// expected-warning@-1{{unsupported OpenCL extension 'cl_khr_mipmap_image' -
ignoring}}
+#endif
+
we also need to check the else case for `__OPENCL_C_VERSION__ < 20
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM. Thanks!
Repository:
rL LLVM
https://reviews.llvm.org/D22637
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.or
yaxunl added inline comments.
Comment at: test/Misc/amdgcn.languageOptsOpenCL.cl:188
@@ +187,3 @@
+#endif
+// expected-warning@+6{{unsupported OpenCL extension 'cl_khr_mipmap_image' -
ignoring}}
+#else
Anastasia wrote:
> Can you move this error message down inste
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM. Thanks!
Repository:
rL LLVM
https://reviews.llvm.org/D22767
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.or
yaxunl created this revision.
yaxunl added reviewers: tstellarAMD, arsenm, nhaustov, Anastasia, rsmith.
yaxunl added a subscriber: cfe-commits.
Herald added a subscriber: kzhuravl.
Adjust target features for amdgcn target when -cl-denorms-are-zero is set.
Denormal support is controlled by feature
yaxunl updated this revision to Diff 65588.
yaxunl marked an inline comment as done.
yaxunl added a comment.
Add more sema tests. Eliminate redundant diagnostic messages.
https://reviews.llvm.org/D21567
Files:
include/clang/AST/OperationKinds.def
include/clang/Basic/DiagnosticGroups.td
in
yaxunl marked 9 inline comments as done.
Comment at: lib/CodeGen/CGOpenCLRuntime.cpp:84
@@ +83,3 @@
+SamplerTy = llvm::PointerType::get(llvm::StructType::create(
+ CGM.getLLVMContext(), "__opencl_sampler_t"),
+ CGM.getContext().getTargetAddressSpace(
yaxunl updated this revision to Diff 65796.
yaxunl marked 5 inline comments as done.
yaxunl added a comment.
Revised by Anastasia's comments. Added missing codegen test.
https://reviews.llvm.org/D21567
Files:
include/clang/AST/OperationKinds.def
include/clang/Basic/DiagnosticGroups.td
inc
yaxunl updated the summary for this revision.
yaxunl updated this revision to Diff 65928.
yaxunl added a comment.
Use 'opencl.sampler_t' as opque type name for sampler in LLVM IR.
https://reviews.llvm.org/D21567
Files:
include/clang/AST/OperationKinds.def
include/clang/Basic/DiagnosticGroup
yaxunl marked 2 inline comments as done.
Comment at: lib/Sema/SemaInit.cpp:6961
@@ +6960,3 @@
+// the initializer.
+if (!Init->isConstantInitializer(S.Context, false))
+ break;
Anastasia wrote:
> I think you don't need this check any more
yaxunl created this revision.
yaxunl added a reviewer: Anastasia.
yaxunl added subscribers: cfe-commits, tstellarAMD.
The size of image type is reported incorrectly as size of a pointer to address
space 0, which causes error when casting image type to pointers by
__builtin_astype.
The fix is to
Author: yaxunl
Date: Thu Jul 28 14:26:30 2016
New Revision: 277024
URL: http://llvm.org/viewvc/llvm-project?rev=277024&view=rev
Log:
[OpenCL] Generate opaque type for sampler_t and function call for the
initializer
Currently Clang use int32 to represent sampler_t, which have been a source of
is
This revision was automatically updated to reflect the committed changes.
yaxunl marked an inline comment as done.
Closed by commit rL277024: [OpenCL] Generate opaque type for sampler_t and
function call for the… (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D21567?vs=
yaxunl created this revision.
yaxunl added reviewers: Anastasia, nhaustov.
yaxunl added subscribers: cfe-commits, tstellarAMD.
Let the driver pass the option to frontend. Do not set precision metadata for
division instructions when this option is set. Set function attribute
"correctly-rounded-di
Author: yaxunl
Date: Fri Jul 29 12:50:10 2016
New Revision: 277179
URL: http://llvm.org/viewvc/llvm-project?rev=277179&view=rev
Log:
[OpenCL] Added CLK_ABGR definition for get_image_channel_order return value
Added CLK_ABGR definition for get_image_channel_order return value inside
opencl-c.h fi
This revision was automatically updated to reflect the committed changes.
Closed by commit rL277179: [OpenCL] Added CLK_ABGR definition for
get_image_channel_order return value (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22767?vs=65375&id=66149#toc
Repository:
rL
Author: yaxunl
Date: Fri Jul 29 12:52:34 2016
New Revision: 277181
URL: http://llvm.org/viewvc/llvm-project?rev=277181&view=rev
Log:
[OpenCL] Add extension cl_khr_mipmap_image to clang
Adding extension cl_khr_mipmap_image to clang's OpenCL Extensions and initiated
inside AMDGPU Target.
Patch by
This revision was automatically updated to reflect the committed changes.
Closed by commit rL277181: [OpenCL] Add extension cl_khr_mipmap_image to clang
(authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22637?vs=65773&id=66150#toc
Repository:
rL LLVM
https://reviews.l
yaxunl added a comment.
ASTContext::getTypeInfoImpl in lib/AST/ASTContext.cpp needs to be updated for
the real size of ndrange_t.
Comment at: lib/CodeGen/CGOpenCLRuntime.cpp:43
@@ +42,3 @@
+
+ return llvm::StructType::create(EleTypes, "ndrange_t");
+}
struct n
yaxunl accepted this revision.
yaxunl added a comment.
This revision is now accepted and ready to land.
LGTM. Thanks!
https://reviews.llvm.org/D23120
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
http://lists.llvm.org/cgi-bin/mailman/listi
yaxunl added inline comments.
Comment at: lib/CodeGen/CGOpenCLRuntime.cpp:43
@@ +42,3 @@
+
+ return llvm::StructType::create(EleTypes, "ndrange_t");
+}
yaxunl wrote:
> struct name should be "struct.ndrange_t" to allow library code to access it.
Sorry, should be "
Author: yaxunl
Date: Wed Aug 3 15:38:06 2016
New Revision: 277647
URL: http://llvm.org/viewvc/llvm-project?rev=277647&view=rev
Log:
[OpenCL] Fix size of image type
The size of image type is reported incorrectly as size of a pointer to address
space 0, which causes error when casting image type
This revision was automatically updated to reflect the committed changes.
Closed by commit rL277647: [OpenCL] Fix size of image type (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22927?vs=65964&id=66705#toc
Repository:
rL LLVM
https://reviews.llvm.org/D22927
Files
yaxunl added inline comments.
Comment at: lib/CodeGen/CGOpenCLRuntime.cpp:43
@@ +42,3 @@
+
+ return llvm::StructType::create(EleTypes, "ndrange_t");
+}
Anastasia wrote:
> yaxunl wrote:
> > yaxunl wrote:
> > > struct name should be "struct.ndrange_t" to allow libr
Author: yaxunl
Date: Thu Aug 4 14:30:54 2016
New Revision: 277754
URL: http://llvm.org/viewvc/llvm-project?rev=277754&view=rev
Log:
[OpenCL] Remove extra native_ functions from opencl-c.h
There should be no native_ builtin functions with double type arguments.
Patch by Aaron En Ye Shi.
Differe
Author: yaxunl
Date: Thu Aug 4 14:35:17 2016
New Revision: 277756
URL: http://llvm.org/viewvc/llvm-project?rev=277756&view=rev
Log:
[OpenCL] Add the lit test for image size which was omitted by r277647.
Added:
cfe/trunk/test/CodeGenOpenCL/cast_image.cl
Added: cfe/trunk/test/CodeGenOpenCL/ca
yaxunl added a comment.
In https://reviews.llvm.org/D23086#507055, @majnemer wrote:
> This approach seems wrong to me.
>
> Instead, why not just make `ndrange_t` a typedef of a real struct in
> `Sema::Initialize`?
ndrange_t is an OpenCL builtin type and the spec does not define its layout.
Th
yaxunl added a comment.
How about assuming ndrange_t is a struct type defined by user and identify it
by struct type name in Clang? This gives user freedom of implementing it
differently than SPIR. In opencl-c.h define it as a struct type as SPIR
required.
Repository:
rL LLVM
https://revie
yaxunl added a comment.
In https://reviews.llvm.org/D23086#507203, @yaxunl wrote:
> How about assuming ndrange_t is a struct type defined by user and identify it
> by struct type name in Clang? This gives user freedom of implementing it
> differently than SPIR. In opencl-c.h define it as a stru
yaxunl created this revision.
yaxunl added reviewers: nhaustov, Anastasia.
yaxunl added subscribers: cfe-commits, tstellarAMD.
Herald added a reviewer: tstellarAMD.
Herald added a subscriber: kzhuravl.
https://reviews.llvm.org/D23322
Files:
include/clang/Basic/OpenCLExtensions.def
lib/Basic/T
Author: yaxunl
Date: Tue Aug 9 14:43:38 2016
New Revision: 278151
URL: http://llvm.org/viewvc/llvm-project?rev=278151&view=rev
Log:
[OpenCL][AMDGPU] Add support for -cl-denorms-are-zero
Adjust target features for amdgcn target when -cl-denorms-are-zero is set.
Denormal support is controlled by
This revision was automatically updated to reflect the committed changes.
Closed by commit rL278151: [OpenCL][AMDGPU] Add support for
-cl-denorms-are-zero (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22815?vs=65533&id=67397#toc
Repository:
rL LLVM
https://reviews
Author: yaxunl
Date: Tue Aug 9 15:10:18 2016
New Revision: 278155
URL: http://llvm.org/viewvc/llvm-project?rev=278155&view=rev
Log:
[OpenCL] Handle -cl-fp32-correctly-rounded-divide-sqrt
Let the driver pass the option to frontend. Do not set precision metadata for
division instructions when thi
This revision was automatically updated to reflect the committed changes.
Closed by commit rL278155: [OpenCL] Handle
-cl-fp32-correctly-rounded-divide-sqrt (authored by yaxunl).
Changed prior to commit:
https://reviews.llvm.org/D22940?vs=66002&id=67406#toc
Repository:
rL LLVM
https://review
yaxunl created this revision.
yaxunl added reviewers: nhaustov, Anastasia.
yaxunl added subscribers: cfe-commits, tstellarAMD.
Pointers of certain GPUs in AMDGCN target in private address space is 32 bit
but pointers in other address spaces are 64 bit. size_t type should be defined
as 64 bit for
yaxunl updated this revision to Diff 67687.
yaxunl added a comment.
Revised by Matt's comments.
https://reviews.llvm.org/D23361
Files:
include/clang/Basic/TargetInfo.h
lib/Basic/TargetInfo.cpp
lib/Basic/Targets.cpp
lib/CodeGen/CodeGenModule.cpp
test/CodeGenOpenCL/size_t.cl
Index: tes
801 - 900 of 1680 matches
Mail list logo
