[STF] reduce access mode #2830

caugonnet · 2024-11-15T13:23:06Z

Description

closes

This PR intends to introduce a reduction access mode to make it much easier to write parallel_for kernels which also perform some reductions to a logical data.

Checklist

[ x] New or existing tests cover these changes.
The documentation is up to date with these changes.

…ls (eg. parallel_for).

copy-pr-bot · 2024-11-15T13:23:10Z

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

caugonnet · 2024-11-15T13:24:02Z

/ok to test

…deps

caugonnet · 2024-11-15T16:08:38Z

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh

@@ -423,7 +452,7 @@ public:
      Fun&& f                  = mv(::std::get<2>(*p));
      const sub_shape_t& shape = ::std::get<3>(*p);

-      auto explode_coords = [&](size_t i, deps_t... data) {
+      auto explode_coords = [&](size_t i, typename deps_ops_t::first_type... data) {


Need a comment to say what that type is (we can't use an alias because it's a pack, not a tuple)

…n access mode, and start to implement all the mechanisms for reductions in parallel_for

… to cuda::std::tuple

cudax/include/cuda/experimental/__stf/internal/task_dep.cuh

caugonnet · 2024-11-21T08:44:37Z

/ok to test

caugonnet · 2024-11-21T11:25:13Z

/ok to test

caugonnet · 2024-11-21T14:04:46Z

/ok to test

…_host method

caugonnet · 2024-11-21T16:01:03Z

/ok to test

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh

caugonnet · 2024-11-21T16:04:16Z

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh

+  // Write the block's result to the output array
+  if (tid == 0)
+  {
+    tuple_set_op<tuple_ops>(redux_buffer[blockIdx.x], per_block_redux_buffer[0].get());


specialize if only one block...

…ntext

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh

github-actions · 2024-11-26T12:47:20Z

🟩 CI finished in 36m 00s: Pass: 100%/54 | Total: 11h 41m | Avg: 12m 59s | Max: 20m 16s | Hits: 90%/246

🟩 cudax: Pass: 100%/54 | Total: 11h 41m | Avg: 12m 59s | Max: 20m 16s | Hits: 90%/246

🟩 cpu
  🟩 amd64              Pass: 100%/50  | Total: 10h 52m | Avg: 13m 03s | Max: 20m 16s | Hits:  90%/246   
  🟩 arm64              Pass: 100%/4   | Total: 48m 31s | Avg: 12m 07s | Max: 13m 07s
🟩 ctk
  🟩 12.0               Pass: 100%/19  | Total:  3h 57m | Avg: 12m 31s | Max: 20m 16s | Hits:  90%/123   
  🟩 12.5               Pass: 100%/2   | Total: 14m 41s | Avg:  7m 20s | Max:  7m 57s
  🟩 12.6               Pass: 100%/33  | Total:  7h 28m | Avg: 13m 36s | Max: 18m 08s | Hits:  90%/123   
🟩 cudacxx
  🟩 nvcc12.0           Pass: 100%/19  | Total:  3h 57m | Avg: 12m 31s | Max: 20m 16s | Hits:  90%/123   
  🟩 nvcc12.5           Pass: 100%/2   | Total: 14m 41s | Avg:  7m 20s | Max:  7m 57s
  🟩 nvcc12.6           Pass: 100%/33  | Total:  7h 28m | Avg: 13m 36s | Max: 18m 08s | Hits:  90%/123   
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/54  | Total: 11h 41m | Avg: 12m 59s | Max: 20m 16s | Hits:  90%/246   
🟩 cxx
  🟩 Clang9             Pass: 100%/2   | Total: 24m 34s | Avg: 12m 17s | Max: 12m 24s
  🟩 Clang10            Pass: 100%/2   | Total: 24m 30s | Avg: 12m 15s | Max: 12m 34s
  🟩 Clang11            Pass: 100%/4   | Total: 47m 46s | Avg: 11m 56s | Max: 13m 16s
  🟩 Clang12            Pass: 100%/4   | Total: 50m 54s | Avg: 12m 43s | Max: 15m 16s
  🟩 Clang13            Pass: 100%/4   | Total: 49m 17s | Avg: 12m 19s | Max: 13m 20s
  🟩 Clang14            Pass: 100%/4   | Total: 55m 42s | Avg: 13m 55s | Max: 15m 51s
  🟩 Clang15            Pass: 100%/2   | Total: 30m 58s | Avg: 15m 29s | Max: 18m 08s
  🟩 Clang16            Pass: 100%/4   | Total: 52m 30s | Avg: 13m 07s | Max: 14m 45s
  🟩 Clang17            Pass: 100%/2   | Total: 29m 42s | Avg: 14m 51s | Max: 16m 58s
  🟩 Clang18            Pass: 100%/2   | Total: 30m 01s | Avg: 15m 00s | Max: 15m 35s
  🟩 GCC9               Pass: 100%/2   | Total: 25m 47s | Avg: 12m 53s | Max: 13m 24s
  🟩 GCC10              Pass: 100%/4   | Total: 53m 10s | Avg: 13m 17s | Max: 14m 08s
  🟩 GCC11              Pass: 100%/4   | Total: 52m 54s | Avg: 13m 13s | Max: 14m 18s
  🟩 GCC12              Pass: 100%/7   | Total:  1h 47m | Avg: 15m 20s | Max: 20m 16s
  🟩 GCC13              Pass: 100%/3   | Total: 34m 21s | Avg: 11m 27s | Max: 13m 07s
  🟩 MSVC14.36          Pass: 100%/1   | Total:  8m 05s | Avg:  8m 05s | Max:  8m 05s | Hits:  90%/123   
  🟩 MSVC14.39          Pass: 100%/1   | Total:  9m 04s | Avg:  9m 04s | Max:  9m 04s | Hits:  90%/123   
  🟩 NVHPC24.7          Pass: 100%/2   | Total: 14m 41s | Avg:  7m 20s | Max:  7m 57s
🟩 cxx_family
  🟩 Clang              Pass: 100%/30  | Total:  6h 35m | Avg: 13m 11s | Max: 18m 08s
  🟩 GCC                Pass: 100%/20  | Total:  4h 33m | Avg: 13m 40s | Max: 20m 16s
  🟩 MSVC               Pass: 100%/2   | Total: 17m 09s | Avg:  8m 34s | Max:  9m 04s | Hits:  90%/246   
  🟩 NVHPC              Pass: 100%/2   | Total: 14m 41s | Avg:  7m 20s | Max:  7m 57s
🟩 gpu
  🟩 v100               Pass: 100%/54  | Total: 11h 41m | Avg: 12m 59s | Max: 20m 16s | Hits:  90%/246   
🟩 jobs
  🟩 Build              Pass: 100%/49  | Total: 10h 17m | Avg: 12m 36s | Max: 18m 08s | Hits:  90%/246   
  🟩 Test               Pass: 100%/5   | Total:  1h 23m | Avg: 16m 42s | Max: 20m 16s
🟩 sm
  🟩 90                 Pass: 100%/1   | Total:  9m 09s | Avg:  9m 09s | Max:  9m 09s
  🟩 90a                Pass: 100%/1   | Total:  9m 43s | Avg:  9m 43s | Max:  9m 43s
🟩 std
  🟩 17                 Pass: 100%/29  | Total:  6h 11m | Avg: 12m 48s | Max: 20m 16s
  🟩 20                 Pass: 100%/25  | Total:  5h 30m | Avg: 13m 12s | Max: 18m 08s | Hits:  90%/246

👃 Inspect Changes

Modifications in project?

	Project
	CCCL Infrastructure
	libcu++
	CUB
	Thrust
+/-	CUDA Experimental
	python
	CCCL C Parallel Library
	Catch2Helper

Modifications in project or dependencies?

	Project
	CCCL Infrastructure
	libcu++
	CUB
	Thrust
+/-	CUDA Experimental
	python
	CCCL C Parallel Library
	Catch2Helper

🏃‍ Runner counts (total jobs: 54)

#	Runner
43	`linux-amd64-cpu16`
5	`linux-amd64-gpu-v100-latest-1`
4	`linux-arm64-cpu16`
2	`windows-amd64-cpu16`

andralex · 2024-11-26T16:45:09Z

cudax/examples/stf/01-axpy-reduce.cu

+
+double X0(int i)
+{
+  return sin((double) i);


Suggested change

return sin((double) i);

return sin(double(i));

In fact I think you don't even need the explicit conversion.

It’s probably necessary due to nasty compilers (one day we will support MSVC…)

andralex · 2024-11-26T16:45:28Z

cudax/examples/stf/01-axpy-reduce.cu

+
+double Y0(int i)
+{
+  return cos((double) i);


Suggested change

return cos((double) i);

return cos(double(i));

andralex · 2024-11-26T16:45:51Z

cudax/examples/stf/01-axpy-reduce.cu

+
+int main()
+{
+  context ctx;


move down just before its first use

andralex · 2024-11-26T17:22:15Z

cudax/examples/stf/CMakeLists.txt

  01-axpy-parallel_for.cu
+  01-axpy-reduce.cu
  01-axpy-launch.cu


Suggested change

01-axpy-parallel_for.cu

01-axpy-reduce.cu

01-axpy-launch.cu

01-axpy-launch.cu

01-axpy-parallel_for.cu

01-axpy-reduce.cu

i.e. keep the list sorted

andralex · 2024-11-26T19:06:01Z

cudax/examples/stf/word_count_reduce.cu

+// determines whether the character is alphabetical
+__host__ __device__ bool is_alpha(const char c)
+{
+  return (c >= 'A' && c <= 'z');


Actually that's incorrect because there are a few nonalpha characters between Z and a.

Suggested change

return (c >= 'A' && c <= 'z');

return c >= 'A' && c <= 'Z' || c >= 'a' && c <= 'z';

I will check what the Thrust example does …

Whoa chatgpt just suggested this clever thing:

Suggested change

return (c >= 'A' && c <= 'z');

return ((c | 0x20) - 'a') < 26;

andralex · 2024-11-26T19:08:24Z

cudax/examples/stf/word_count_reduce.cu

+  ctx.parallel_for(ltext.shape(), ltext.read(), lcnt.reduce(reducer::sum<int>{}))
+      ->*[] _CCCL_DEVICE(size_t i, auto text, int& s) {
+            /* When we have the beginning of a new word, increment the counter */
+            if (!is_alpha(text(i)) && is_alpha(text(i + 1)))


Careful here with out of bounds accesses.

Suggested change

if (!is_alpha(text(i)) && is_alpha(text(i + 1)))

if (is_alpha(text(i)) && (i == 0 || !is_alpha(text(i - 1))))

andralex · 2024-11-26T19:09:16Z

cudax/examples/stf/word_count_reduce.cu

+  int ref_cnt = 0;
+  for (size_t i = 0; i < sizeof(raw_input) - 1; i++)
+  {
+    if (!is_alpha(raw_input[i]) && is_alpha(raw_input[i + 1]))


Suggested change

if (!is_alpha(raw_input[i]) && is_alpha(raw_input[i + 1]))

if (is_alpha(raw_input[i]) && (i == 0 || is_alpha(raw_input[i - 1])))

This calls for a better choice of the shape instead of the test

andralex · 2024-11-26T19:33:44Z

cudax/include/cuda/experimental/__stf/graph/graph_ctx.cuh

@@ -517,6 +517,23 @@ public:
    cudaGraphDebugDotPrint(get_graph(), filename.c_str(), flags);
  }

+  template <typename T>
+  auto transfer_host(cuda::experimental::stf::logical_data<T>& ldata)


As discussed, this would be best renamed as wait or host_wait. There should be a clear messaging that this is a blocking function (unless most of cudastf).

I disagree wait suggests the main goal is to synchronize with the data, while the goal is to get its content, synchronization is a necessary byproduct

we could also have a wait method on logical data we have missed that for ages

andralex · 2024-11-26T19:35:16Z

cudax/include/cuda/experimental/__stf/graph/graph_ctx.cuh

+    using valT = typename owning_container_of<T>::type;
+    valT out;


No need for the using - you're using the name just once.

Suggested change

using valT = typename owning_container_of<T>::type;

valT out;

typename owning_container_of<T>::type out;

andralex · 2024-11-26T20:04:48Z

cudax/include/cuda/experimental/__stf/internal/constants.cuh

+  reduce_no_init = 8,
+  reduce_do_init = 16,


Suggested change

reduce_no_init = 8,

reduce_do_init = 16,

reduce_init = 8,

reduce = 16,

I mean the implicit operation is "reduce" and there's a special thing "reduce initialize".

I guess the default should be the reduce, and the less common case would be reduce_no_init, so i'll do that instead

andralex · 2024-11-26T20:06:44Z

cudax/include/cuda/experimental/__stf/internal/constants.cuh

+    case access_mode::reduce_no_init:
+      return "reduce (no init)"; // op ?
+    case access_mode::reduce_do_init:
+      return "reduce (do init)"; // op ?


Suggested change

case access_mode::reduce_no_init:

return "reduce (no init)"; // op ?

case access_mode::reduce_do_init:

return "reduce (do init)"; // op ?

case access_mode::reduce:

return "reduce"; // op ?

case access_mode::reduce_init:

return "reduce (initialization stage)"; // op ?

cudax/include/cuda/experimental/__stf/internal/constants.cuh

…init to user, then use true_type and false_type internally. Also rename reduce_do_init to reduce for clarity, as this is the most common case.

caugonnet · 2024-11-27T14:40:42Z

/ok to test

caugonnet · 2024-11-27T14:47:29Z

cudax/examples/stf/01-axpy-reduce.cu

This example should be reworked as a DOT product instead, to make it look like more useful

caugonnet · 2024-11-27T14:48:09Z

cudax/include/cuda/experimental/__stf/internal/constants.cuh

    default:
      assert(false);
      abort();
  }
 }

+/**


Do this tag type belongs here ?

caugonnet · 2024-11-27T14:49:18Z

cudax/include/cuda/experimental/__stf/internal/scalar_interface.cuh

+    size_t input_cnt) override
+  {
+    cudaGraphNode_t dummy;
+    cuda_safe_call(cudaGraphAddEmptyNode(&dummy, graph, input_nodes, input_cnt));


This is not implemented ... Perhaps we need to add helpers ?

github-actions · 2024-11-27T15:14:23Z

🟩 CI finished in 32m 26s: Pass: 100%/54 | Total: 12h 23m | Avg: 13m 46s | Max: 19m 46s | Hits: 90%/256

🟩 cudax: Pass: 100%/54 | Total: 12h 23m | Avg: 13m 46s | Max: 19m 46s | Hits: 90%/256

🟩 cpu
  🟩 amd64              Pass: 100%/50  | Total: 11h 28m | Avg: 13m 45s | Max: 19m 46s | Hits:  90%/256   
  🟩 arm64              Pass: 100%/4   | Total: 55m 48s | Avg: 13m 57s | Max: 15m 39s
🟩 ctk
  🟩 12.0               Pass: 100%/19  | Total:  4h 10m | Avg: 13m 11s | Max: 19m 46s | Hits:  90%/128   
  🟩 12.5               Pass: 100%/2   | Total: 17m 22s | Avg:  8m 41s | Max:  9m 07s
  🟩 12.6               Pass: 100%/33  | Total:  7h 55m | Avg: 14m 25s | Max: 16m 28s | Hits:  90%/128   
🟩 cudacxx
  🟩 nvcc12.0           Pass: 100%/19  | Total:  4h 10m | Avg: 13m 11s | Max: 19m 46s | Hits:  90%/128   
  🟩 nvcc12.5           Pass: 100%/2   | Total: 17m 22s | Avg:  8m 41s | Max:  9m 07s
  🟩 nvcc12.6           Pass: 100%/33  | Total:  7h 55m | Avg: 14m 25s | Max: 16m 28s | Hits:  90%/128   
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/54  | Total: 12h 23m | Avg: 13m 46s | Max: 19m 46s | Hits:  90%/256   
🟩 cxx
  🟩 Clang9             Pass: 100%/2   | Total: 26m 02s | Avg: 13m 01s | Max: 13m 45s
  🟩 Clang10            Pass: 100%/2   | Total: 27m 41s | Avg: 13m 50s | Max: 14m 58s
  🟩 Clang11            Pass: 100%/4   | Total: 51m 34s | Avg: 12m 53s | Max: 13m 11s
  🟩 Clang12            Pass: 100%/4   | Total: 53m 56s | Avg: 13m 29s | Max: 14m 48s
  🟩 Clang13            Pass: 100%/4   | Total: 52m 02s | Avg: 13m 00s | Max: 13m 47s
  🟩 Clang14            Pass: 100%/4   | Total: 57m 33s | Avg: 14m 23s | Max: 14m 54s
  🟩 Clang15            Pass: 100%/2   | Total: 29m 18s | Avg: 14m 39s | Max: 14m 59s
  🟩 Clang16            Pass: 100%/4   | Total: 56m 29s | Avg: 14m 07s | Max: 15m 49s
  🟩 Clang17            Pass: 100%/2   | Total: 31m 48s | Avg: 15m 54s | Max: 16m 28s
  🟩 Clang18            Pass: 100%/2   | Total: 30m 32s | Avg: 15m 16s | Max: 15m 44s
  🟩 GCC9               Pass: 100%/2   | Total: 28m 01s | Avg: 14m 00s | Max: 14m 53s
  🟩 GCC10              Pass: 100%/4   | Total: 55m 08s | Avg: 13m 47s | Max: 15m 19s
  🟩 GCC11              Pass: 100%/4   | Total:  1h 02m | Avg: 15m 38s | Max: 19m 46s
  🟩 GCC12              Pass: 100%/7   | Total:  1h 42m | Avg: 14m 36s | Max: 16m 07s
  🟩 GCC13              Pass: 100%/3   | Total: 40m 10s | Avg: 13m 23s | Max: 15m 39s
  🟩 MSVC14.36          Pass: 100%/1   | Total:  8m 53s | Avg:  8m 53s | Max:  8m 53s | Hits:  90%/128   
  🟩 MSVC14.39          Pass: 100%/1   | Total: 12m 27s | Avg: 12m 27s | Max: 12m 27s | Hits:  90%/128   
  🟩 NVHPC24.7          Pass: 100%/2   | Total: 17m 22s | Avg:  8m 41s | Max:  9m 07s
🟩 cxx_family
  🟩 Clang              Pass: 100%/30  | Total:  6h 56m | Avg: 13m 53s | Max: 16m 28s
  🟩 GCC                Pass: 100%/20  | Total:  4h 48m | Avg: 14m 24s | Max: 19m 46s
  🟩 MSVC               Pass: 100%/2   | Total: 21m 20s | Avg: 10m 40s | Max: 12m 27s | Hits:  90%/256   
  🟩 NVHPC              Pass: 100%/2   | Total: 17m 22s | Avg:  8m 41s | Max:  9m 07s
🟩 gpu
  🟩 v100               Pass: 100%/54  | Total: 12h 23m | Avg: 13m 46s | Max: 19m 46s | Hits:  90%/256   
🟩 jobs
  🟩 Build              Pass: 100%/49  | Total: 11h 06m | Avg: 13m 35s | Max: 19m 46s | Hits:  90%/256   
  🟩 Test               Pass: 100%/5   | Total:  1h 17m | Avg: 15m 33s | Max: 16m 07s
🟩 sm
  🟩 90                 Pass: 100%/1   | Total:  9m 32s | Avg:  9m 32s | Max:  9m 32s
  🟩 90a                Pass: 100%/1   | Total: 10m 19s | Avg: 10m 19s | Max: 10m 19s
🟩 std
  🟩 17                 Pass: 100%/29  | Total:  6h 29m | Avg: 13m 26s | Max: 15m 49s
  🟩 20                 Pass: 100%/25  | Total:  5h 54m | Avg: 14m 09s | Max: 19m 46s | Hits:  90%/256

👃 Inspect Changes

Modifications in project?

	Project
	CCCL Infrastructure
	libcu++
	CUB
	Thrust
+/-	CUDA Experimental
	python
	CCCL C Parallel Library
	Catch2Helper

Modifications in project or dependencies?

	Project
	CCCL Infrastructure
	libcu++
	CUB
	Thrust
+/-	CUDA Experimental
	python
	CCCL C Parallel Library
	Catch2Helper

🏃‍ Runner counts (total jobs: 54)

#	Runner
43	`linux-amd64-cpu16`
5	`linux-amd64-gpu-v100-latest-1`
4	`linux-arm64-cpu16`
2	`windows-amd64-cpu16`

caugonnet added 5 commits November 12, 2024 14:58

Rename the redux access mode into relaxed

05c696c

change redux to relaxed in the documentation

3c954c6

clang-format

9cbc0ec

Merge branch 'main' into stf_reducer_access_mode

16938fc

Experiment to start introducing a reduction access mode used in kerne…

5f71031

…ls (eg. parallel_for).

Add a trait to count the number of reductions required in a tuple of …

bdedda2

…deps

caugonnet commented Nov 15, 2024

View reviewed changes

caugonnet added 7 commits November 18, 2024 14:18

WIP: create a new scalar<T> interface which can be used in a reductio…

dec19cd

…n access mode, and start to implement all the mechanisms for reductions in parallel_for

WIP ! Introduce owning_container_of trait class

1c883fa

WIP: save progress here, lots of hardcoded things and we need to move…

09ffbd9

… to cuda::std::tuple

WIP : first prototype working...

97e0ba5

Proper initialization of shared memory buffers, and add another example

187be94

Some cleanups and renaming of classes for better clarity

bdfbb03

clang-format

a6a2018

caugonnet commented Nov 20, 2024

View reviewed changes

cudax/include/cuda/experimental/__stf/internal/task_dep.cuh Outdated Show resolved Hide resolved

caugonnet added the stf Sequential Task Flow programming model label Nov 21, 2024

Merge branch 'main' into stf_reducer_access_mode

86b3293

workaround some false unused captured variable warning

ce0a30f

Fix various C++ errors, and do not use the I variable

374d253

caugonnet added 2 commits November 21, 2024 16:48

Rework the CFD example to use reductions, and generalize the transfer…

7c28b1e

…_host method

clang-format

8f552e0

caugonnet commented Nov 21, 2024

View reviewed changes

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh Outdated Show resolved Hide resolved

caugonnet commented Nov 21, 2024

View reviewed changes

Implement transfer_host (name subject to change !) directly in the co…

438589b

…ntext

caugonnet commented Nov 26, 2024

View reviewed changes

cudax/include/cuda/experimental/__stf/internal/parallel_for_scope.cuh Outdated Show resolved Hide resolved

caugonnet added 3 commits November 26, 2024 13:50

Remove unused template parameter

19f26a1

Slightly simpler count_type trait

a4a75f0

clang-format

cad27fa

andralex reviewed Nov 26, 2024

View reviewed changes

cudax/examples/stf/01-axpy-reduce.cu

int main()

{

context ctx;

Copy link

Contributor

andralex Nov 26, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

move down just before its first use

andralex reviewed Nov 26, 2024

View reviewed changes

cudax/include/cuda/experimental/__stf/internal/constants.cuh Outdated Show resolved Hide resolved

caugonnet added 5 commits November 27, 2024 11:11

Add a small unit test to test count_type_v

f5240a4

Do not define both no_init and do_init types anymore, just expose no_…

34d41ae

…init to user, then use true_type and false_type internally. Also rename reduce_do_init to reduce for clarity, as this is the most common case.

sort examples in cmake

e6fa002

clang-format

697f3ba

Merge branch 'main' into stf_reducer_access_mode

80f60b7

caugonnet commented Nov 27, 2024

View reviewed changes

andralex added 3 commits November 27, 2024 15:55

Simplify redux_vars

e803587

Use ::std::monostate instead of EmptyType

14f9c12

Simplify redux_vars

8839684

	return (c >= 'A' && c <= 'z');
	return c >= 'A' && c <= 'Z' \|\| c >= 'a' && c <= 'z';

	return (c >= 'A' && c <= 'z');
	return ((c \| 0x20) - 'a') < 26;

	if (!is_alpha(text(i)) && is_alpha(text(i + 1)))
	if (is_alpha(text(i)) && (i == 0 \|\| !is_alpha(text(i - 1))))

	if (!is_alpha(raw_input[i]) && is_alpha(raw_input[i + 1]))
	if (is_alpha(raw_input[i]) && (i == 0 \|\| is_alpha(raw_input[i - 1])))

		using valT = typename owning_container_of<T>::type;
		valT out;

	using valT = typename owning_container_of<T>::type;
	valT out;
	typename owning_container_of<T>::type out;

[STF] reduce access mode #2830

Are you sure you want to change the base?

[STF] reduce access mode #2830

Conversation

caugonnet commented Nov 15, 2024

Description

Checklist

copy-pr-bot bot commented Nov 15, 2024

caugonnet commented Nov 15, 2024

Choose a reason for hiding this comment

caugonnet commented Nov 21, 2024

caugonnet commented Nov 21, 2024

caugonnet commented Nov 21, 2024

caugonnet commented Nov 21, 2024

Choose a reason for hiding this comment

github-actions bot commented Nov 26, 2024

🟩 cudax: Pass: 100%/54 | Total: 11h 41m | Avg: 12m 59s | Max: 20m 16s | Hits: 90%/246

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 54)

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

andralex Nov 26, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

caugonnet commented Nov 27, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

github-actions bot commented Nov 27, 2024

🟩 cudax: Pass: 100%/54 | Total: 12h 23m | Avg: 13m 46s | Max: 19m 46s | Hits: 90%/256

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 54)

andralex Nov 26, 2024 •

edited

Loading