@@ -1558,6 +1558,13 @@ void clang::InitializePreprocessor(Preprocessor &PP,
const PCHContainerReader &PCHContainerRdr,
const FrontendOptions &FEOpts,
const CodeGenOptions &C
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/137665
From 72bc06251047bb92a5bc6a068670b047c078a3c3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Mon, 28 Apr 2025 17:05:46 +0200
Subject: [PATCH 1/4] Pre-commit test: [Pr
jmmartinez wrote:
Ping !
https://github.com/llvm/llvm-project/pull/137665
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez approved this pull request.
Thanks for the fix !
https://github.com/llvm/llvm-project/pull/140490
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/139697
From e60d2b04d1cb7873e4034cb639406b1c329d1c45 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Tue, 13 May 2025 10:32:41 +0200
Subject: [PATCH] [HIP] Fix return type in
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/139891
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
jmmartinez wrote:
> This seems reasonable. Can you update the description to explain the
> motivation for the change, though? Is it a bug you are trying to fix, or a
> specific scenario you think should be optimized? Thanks
I've updated the description with an explanation of the situation that
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/137665
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/139892
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/139697
Rate limit · GitHub
body {
background-color: #f6f8fa;
color: #24292e;
font-family: -apple-system,BlinkMacSystemFont,Segoe
UI,Helvetica,Arial,san
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/139892
ambibuity -> ambiguity
From 5509f15dde030de212660a394d9b98530eda3681 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Wed, 14 May 2025 13:40:42 +0200
Subject: [PATCH]
jmmartinez wrote:
Added a tests and created a separate PR to commit the tests before the fix in
https://github.com/llvm/llvm-project/pull/139891
https://github.com/llvm/llvm-project/pull/139697
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
h
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/139891
Tests related to https://github.com/llvm/llvm-project/pull/139697
From 57abd4e1a54aedaeeaaceca394f298d7f7abc4eb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Wed, 1
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/137665
Rate limit · GitHub
body {
background-color: #f6f8fa;
color: #24292e;
font-family: -apple-system,BlinkMacSystemFont,Segoe
UI,Helvetica,Arial,san
jmmartinez wrote:
I fixed the `Modules/initializers.cpp` test by moving the preprocessor
directives outside of the module contents (the module contents should be
already preprocessed and not contain preproc directives).
https://github.com/llvm/llvm-project/pull/137665
_
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/139697
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/139697
Before, some functions like `isgreater(double, double)` would return a `double`
instead of a `bool`.
I've stumbled upon this bug while trying to adapt
[`External/CUDA/cmath.cu`](https://github.com/llvm/llvm
jmmartinez wrote:
> Now the question is -- how do we test it?
I'm working on adding some tests for the cuda_wrapper headers:
* [ ] First in https://github.com/llvm/llvm-test-suite/tree/main/External/CUDA
(there are some already related to other cuda_wrapper headers)
* [ ] and likely something v
jmmartinez wrote:
> module maps don't get preprocessed, so I don't think it matters what that is
> set to. For normal module.modulemap files `Preprocessed` is `false`. I think
> it's fine to remove the true and see if anything breaks, the preprocessor
> just skips over that section anyway.
Cu
https://github.com/jmmartinez approved this pull request.
Thanks @spavloff ! I'm sorry I didn't address the remarks earlier, I was
out-of-office and didn't see the notifications.
https://github.com/llvm/llvm-project/pull/139492
___
cfe-commits mailing
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/137623
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/136853
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
jmmartinez wrote:
Thanks !
https://github.com/llvm/llvm-project/pull/137306
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/137306
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
jmmartinez wrote:
I've added a verbose version of the assertion that uses `__builtin_printf` (I
could not figure out a way to correctly emit something to `stderr` from device
code in a generic way without including `cstdio`.
```cpp
#ifdef _GLIBCXX_VERBOSE_ASSERT
__attribute__((device, noreturn
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/136133
From a602d4f56993c2fc2825997a367b4551fcd54817 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 17 Apr 2025 13:41:55 +0200
Subject: [PATCH] [CUDA][HIP] Add a __devi
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/136133
From 2136a05f0b8a0f2db0f9506a78ebf8f41c0f4d2f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 17 Apr 2025 13:41:55 +0200
Subject: [PATCH] [CUDA][HIP] Add a __devi
@@ -155,3 +155,11 @@
#endif
#pragma OPENCL EXTENSION cl_amd_media_ops2: enable
+#if (__OPENCL_C_VERSION__ >= 300)
jmmartinez wrote:
Thanks ! You're right. It's available by default on OpenCL C 2 but not 3.
https://github.com/llvm/llvm-project/pull/137636
___
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/137665
This is a draft while I'm trying to figure out what's left to do.
This has issues with the test `Modules/initializers.cpp`.
The function `CompilerInstance::createModuleFromSource` creates a
`FrontendInput`
@@ -155,3 +155,11 @@
#endif
#pragma OPENCL EXTENSION cl_amd_media_ops2: enable
+#if (__OPENCL_C_VERSION__ >= 300)
jmmartinez wrote:
I don't understand, isn't `__opencl_c_generic_address_space` available by
default with `OPENCL_C_VERSION >= 200` (from reading
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/137623
`-rewrite-objc` passes `-x objective-c++-cpp-output` as input type to the
preprocessor job. This is not correct since we would be preprocessing a
preprocessed file. The correct input type is `objective-c++`.
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/137306
When using `-dD` to generate a preprocessed output, the `#define` directives
are preserved. This includes built-in and command-line definitions.
Before, clang would warn for reserved identifiers for serialize
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/136853
From 1acf00c794bddd5aed30670915a83db5b243b7ff Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Wed, 16 Apr 2025 14:20:53 +0200
Subject: [PATCH 1/2] Pre-commit test: [Cl
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/136133
From b433add1b7c2027b89cdd9be11680c0a9a59e807 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 17 Apr 2025 13:41:55 +0200
Subject: [PATCH] [CUDA][HIP] Add a __devi
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/136853
When using -no-integrated-cpp, before, the driver won't collapse actions when
the input was not llvm-ir
or it would collapse them too aggressively with -save-temps
The original code was checking the action t
@@ -0,0 +1,35 @@
+// libstdc++ uses the non-constexpr function std::__glibcxx_assert_fail()
+// to trigger compilation errors when the __glibcxx_assert(cond) macro
+// is used in a constexpr context.
+// Compilation fails when using code from the libstdc++ (such as std::array) on
@@ -0,0 +1,35 @@
+// libstdc++ uses the non-constexpr function std::__glibcxx_assert_fail()
+// to trigger compilation errors when the __glibcxx_assert(cond) macro
+// is used in a constexpr context.
+// Compilation fails when using code from the libstdc++ (such as std::array) on
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/136133
From b98abc3e1b1ca6c81509e310b6df1f7f7cfd10b9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 17 Apr 2025 13:41:55 +0200
Subject: [PATCH] [CUDA][HIP] Add a __devi
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/136133
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -0,0 +1,35 @@
+// libstdc++ uses the non-constexpr function std::__glibcxx_assert_fail()
+// to trigger compilation errors when the __glibcxx_assert(cond) macro
+// is used in a constexpr context.
+// Compilation fails when using code from the libstdc++ (such as std::array) on
jmmartinez wrote:
See
https://github.com/gcc-mirror/gcc/blob/c237297ee596545552f48ec9582d7d16703d8949/libstdc%2B%2B-v3/include/bits/c%2B%2Bconfig#L645-L659
for the problematic libstdc++ snippet
https://github.com/llvm/llvm-project/pull/136133
___
cfe
jmmartinez wrote:
This patch is cruelly missing some tests. Is there a place for tests of this
kind? I haven't found an obvious one for other headers.
https://github.com/llvm/llvm-project/pull/136133
___
cfe-commits mailing list
cfe-commits@lists.llvm
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/136133
libstdc++ 15 uses the non-constexpr function
std::__glibcxx_assert_fail() to trigger compilation errors when the
__glibcxx_assert(cond) macro is used in a constantly evaluated context.
Compilation fails when
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/135027
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/135376
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/135376
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/135376
The vmem-to-lds-loads-insts feature is only available on gfx9/10. While
target-parser was also enabling it for gfx6,7,8.
From 90c95af57317103f399ee4059a4e4ec01ddc98de Mon Sep 17 00:00:00 2001
From: =?UTF-8?q
jmmartinez wrote:
> IIRC there were bugs when you try to use lambda in conjunction with the
> target attribute. Can you add a test where you try to use a builtin that
> depends on a subtarget feature, in a function marked with the target
> attribute (and not available in the main compile targe
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/135027
From a22e56d063e91df667d89ed5238bd886ce10bd0a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Wed, 9 Apr 2025 15:11:19 +0200
Subject: [PATCH] [Clang][AMDGPU] Accept bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/135027
From 4e5736d1b90ae477642ea03dcf39231f5f0a8e59 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Wed, 9 Apr 2025 15:11:19 +0200
Subject: [PATCH 1/4] [Clang][AMDGPU] Accep
@@ -0,0 +1,24 @@
+// REQUIRES: amdgpu-registered-target
+// RUN: %clang_cc1 -std=c++20 -triple amdgcn -target-cpu tahiti -fsyntax-only
-fcuda-is-device -verify=expected -o - %s
+// RUN: %clang_cc1 -std=c++20 -triple amdgcn -target-cpu gfx950 -fsyntax-only
-fcuda-is-device -o - %
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/135027
`Sema::getCurFunctionDecl(AllowLambda = false)` returns a nullptr when the
lambda declaration is outside a function (for example, when assigning a lambda
to a static constexpr variable).
This triggered an a
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/134598
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
jmmartinez wrote:
Thanks for the quick review !
https://github.com/llvm/llvm-project/pull/134598
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/134598
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/134598
The pointer argument for `wait_for_event(int, event_t*)` should take the
default address space: generic if available, otherwise private.
Before this patch it would always be generic with `-fdeclare-opencl-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/132048
From 27b90ab19cc8c8694784b41b53b2623c718071ab Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/3] [AMDGPU] Add "lds-bu
jmmartinez wrote:
I've added a commit to have a common attribute that we could use for both,
global-load-lds and buffer-load-lds builtins.
https://github.com/llvm/llvm-project/pull/133055
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https:/
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133741
From 41af38793161b0f1535c98c4695c36e081ef2f67 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 27 Mar 2025 17:46:34 +0100
Subject: [PATCH 01/10] [Clang][AMDGPU] Ad
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/132048
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/132048
From 5f039ef32105132a7edf6203eb0b7825d4b8b503 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/6] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From 5f039ef32105132a7edf6203eb0b7825d4b8b503 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/5] [AMDGPU] Add "lds-bu
@@ -729,6 +729,25 @@ GCNTTIImpl::instCombineIntrinsic(InstCombiner &IC,
IntrinsicInst &II) const {
break;
}
+ case Intrinsic::amdgcn_cvt_off_f32_i4: {
+Value* Arg = II.getArgOperand(0);
+Type *Ty = II.getType();
+
+if (isa(Arg))
+ return IC.replaceInst
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133741
From 41af38793161b0f1535c98c4695c36e081ef2f67 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 27 Mar 2025 17:46:34 +0100
Subject: [PATCH 1/9] [Clang][AMDGPU] Add
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/133741
This built-in that maps to `V_CVT_OFF_F32_I4` which treats its input as a 4-bit
signed integer and returns `0.0625f * src`.
SWDEV-518861
From 41af38793161b0f1535c98c4695c36e081ef2f67 Mon Sep 17 00:00:00 200
@@ -260,7 +260,7 @@ AMDGPUTargetInfo::AMDGPUTargetInfo(const llvm::Triple
&Triple,
MaxAtomicPromoteWidth = MaxAtomicInlineWidth = 64;
CUMode = !(GPUFeatures & llvm::AMDGPU::FEATURE_WGP);
- for (auto F : {"image-insts", "gws"})
+ for (auto F : {"image-insts", "gws", "mem
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From 93232bb8255e927d62dd9a5f63e885e73ccf0eb5 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/5] [AMDGPU] Add "lds-bu
@@ -0,0 +1,45 @@
+; RUN: split-file %s %t
+;
+; RUN: not --crash llc -mtriple=amdgcn -mcpu=gfx1100 %t/struct.ll 2>&1 |
FileCheck --ignore-case %s
+; RUN: not --crash llc -global-isel -mtriple=amdgcn -mcpu=gfx1100 %t/struct.ll
2>&1 | FileCheck --ignore-case %s
+; RUN: not --crash
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From bf3f06cc99a1ac2460ffea741c0fe495f0778e70 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/4] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From bf3f06cc99a1ac2460ffea741c0fe495f0778e70 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/3] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From bf3f06cc99a1ac2460ffea741c0fe495f0778e70 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/3] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/133055
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez edited
https://github.com/llvm/llvm-project/pull/133055
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From 70fbaa4e4f2334cf6134adf02b4af2bcc0bf9847 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/3] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From eae04b28a010045bdd7fcd03f84c952abeed3ff6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/2] [AMDGPU] Add "lds-bu
jmmartinez wrote:
> At the hardware level, GFX11 removed the ability for buffer, **scratch** and
> **global** instructions to return directly to LDS. So can we use one
> attribute that covers all three of those?
I've been "grep"ping in upstream (but not yet downstream):
* Scratch load lds are
jmmartinez wrote:
> > Global load lds is available on gfx9 and gfx10.
> > However, the __builtin_amdgcn_global_load_lds builtin is restricted to
> > platforms with the attribute gfx940-insts; while it could be available on
> > gfx90a for example.
>
> This doesn't sound right, it was new in 940
@@ -3368,7 +3368,8 @@ bool AMDGPUInstructionSelector::selectG_INSERT_VECTOR_ELT(
}
bool AMDGPUInstructionSelector::selectBufferLoadLds(MachineInstr &MI) const {
- assert(!AMDGPU::isGFX12Plus(STI));
+ if (!Subtarget->hasLDSBufferLoad())
jmmartinez wrote:
Tha
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/133055
From eae04b28a010045bdd7fcd03f84c952abeed3ff6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/2] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/132048
From eae04b28a010045bdd7fcd03f84c952abeed3ff6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/2] [AMDGPU] Add "lds-bu
jmmartinez wrote:
Squashed first commit from https://github.com/llvm/llvm-project/pull/132048
https://github.com/llvm/llvm-project/pull/133055
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-c
jmmartinez wrote:
Squashed and split in 2: https://github.com/llvm/llvm-project/pull/133055
https://github.com/llvm/llvm-project/pull/132048
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-com
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/133055
This are used to restrict the availability of buffer_load_lds intrinsics to
targets that actually have this instructions.
From eae04b28a010045bdd7fcd03f84c952abeed3ff6 Mon Sep 17 00:00:00 2001
From: =?UTF-8?
@@ -162,6 +162,8 @@ BUILTIN(__builtin_amdgcn_raw_buffer_load_b64, "V2UiQbiiIi",
"n")
BUILTIN(__builtin_amdgcn_raw_buffer_load_b96, "V3UiQbiiIi", "n")
BUILTIN(__builtin_amdgcn_raw_buffer_load_b128, "V4UiQbiiIi", "n")
+TARGET_BUILTIN(__builtin_amdgcn_raw_ptr_buffer_load_lds, "v
@@ -43,7 +43,7 @@ int bar() { return no_attr() + attr_in_target() +
attr_not_in_target() + attr_in
// CHECK-LABEL: @attr_incompatible
// CHECK-SAME: () #[[ATTR_INCOMPATIBLE:[0-9]+]] {
-// CHECK: attributes #[[ATTR_BAR]] = { {{.*}} "target-cpu"="gfx90a"
"target-features"="+16
jmmartinez wrote:
> > I've also seen that gfx11 seem to have some kind of BUFFER_LOAD_LDS_(SIZE)
> > instruction (different from the BUFFER_LOAD_(SIZE)_LDS instructions
> > associated with this builtin).
> > However, I'm not sure they actually exists: we do not have any test where
> > we gener
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/132048
From f7df0f882bb2062fa63111d07aca3c91d0805904 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/6] [AMDGPU] Add "lds-bu
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/132048
From 4e2dce7968b73d54a2b9eaef964e9fe8bd1bb731 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Thu, 20 Mar 2025 13:23:25 +0100
Subject: [PATCH 1/4] [AMDGPU] Add "lds-bu
jmmartinez wrote:
I've also seen that gfx11 seem to have some kind of BUFFER_LOAD_LDS_(SIZE)
instruction (different from the BUFFER_LOAD_(SIZE)_LDS instructions associated
with this builtin).
However, I'm not sure they actually exists: we do not have any test where we
generate them and the g
@@ -162,6 +162,8 @@ BUILTIN(__builtin_amdgcn_raw_buffer_load_b64, "V2UiQbiiIi",
"n")
BUILTIN(__builtin_amdgcn_raw_buffer_load_b96, "V3UiQbiiIi", "n")
BUILTIN(__builtin_amdgcn_raw_buffer_load_b128, "V4UiQbiiIi", "n")
+BUILTIN(__builtin_amdgcn_raw_ptr_buffer_load_lds, "vQbv*3IU
https://github.com/jmmartinez created
https://github.com/llvm/llvm-project/pull/132048
CK is using either inline assembly or inline LLVM-IR builtins to generate
buffer_load_dword lds instructions.
This patch exposes this instruction as a Clang builtin.
Limitations in this patch: these instruc
jmmartinez wrote:
Thanks for the review !
https://github.com/llvm/llvm-project/pull/130284
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/130284
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/130284
From 27adc9ef08bf2ac98319885fb461975f37a98bc2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Fri, 7 Mar 2025 12:29:31 +0100
Subject: [PATCH 1/2] [OffloadBundler] Expo
@@ -1121,13 +1122,116 @@
CompressedOffloadBundle::compress(llvm::compression::Params P,
llvm::StringRef(FinalBuffer.data(), FinalBuffer.size()));
}
+// Use packed structs to avoid padding, such that the structs map the
serialized
+// format.
+LLVM_PACKED_START
+union R
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/130746
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/130746
From 43431642d278011477387ae6bfb2913103aca1ba Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Mon, 10 Mar 2025 15:35:27 +0100
Subject: [PATCH] [Clang] Add __builtin_el
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/130746
From 5c4c285df2d39e93c80b11396511b7d238b281b1 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Mon, 10 Mar 2025 15:35:27 +0100
Subject: [PATCH] [Clang] Add __builtin_el
https://github.com/jmmartinez closed
https://github.com/llvm/llvm-project/pull/130747
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/jmmartinez updated
https://github.com/llvm/llvm-project/pull/130746
From c017072c34203a25f4f179dc70e4adc27094c5dc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Juan=20Manuel=20Martinez=20Caama=C3=B1o?=
Date: Mon, 10 Mar 2025 15:40:00 +0100
Subject: [PATCH 1/2] [Clang][NFC] Rename
1 - 100 of 149 matches
Mail list logo
