8000 [ATen][CUDA] Use new CCCL API in v2.8 (#160554) · pytorch/pytorch@d670304 · GitHub
[go: up one dir, main page]

Skip to content

Commit d670304

Browse files
Aidyn-Apytorchmergebot
authored andcommitted
[ATen][CUDA] Use new CCCL API in v2.8 (#160554)
Silences deprecation warnings like: ``` In file included from tmpxft_003a195d_00000000-6_Nonzero.cudafe1.stub.c:1: /tmp/tmpxft_003a195d_00000000-6_Nonzero.cudafe1.stub.c: At global scope: /tmp/tmpxft_003a195d_00000000-6_Nonzero.cudafe1.stub.c:243:219: warning: 'template<class ValueType, class OffsetT> class at_cuda_detail::cub::CountingInputIterator' is deprecated: Use thrust::counting_iterator instead [-Wdeprecated-declarations] 243 | static void __device_stub__ZN2at6native43_GLOBAL__N__3cee4041_10_Nonzero_cu_cba1aaa011flag_kernelILi512ELi16EhEEvPKT1_PlPKllli( const _ZN3c104impl20ScalarTypeToCPPTypeTILNS_10ScalarTypeE0EEE *__par0, int64_t *__par1, const int64_t *__par2, int64_t __par3, int64_t __par4, int __par5) { __cudaLaunchPrologue(6); __cudaSetupArgSimple(__par0, 0UL); __cudaSetupArgSimple(__par1, 8UL); __cudaSetupArgSimple(__par2, 16UL); __cudaSetupArgSimple(__par3, 24UL); __cudaSetupArgSimple(__par4, 32UL); __cudaSetupArgSimple(__par5, 40UL); __cudaLaunch(((char *)((void ( *)(const _ZN3c104impl20ScalarTypeToCPPTypeTILNS_10ScalarTypeE0EEE *, int64_t *, const int64_t *, int64_t, int64_t, int))at::native::_NV_ANON_NAMESPACE::flag_kernel<(int)512, (int)16, unsigned char> ))); }namespace at{ | ^~~~~~~~~~~~~~~~~~~~~ /usr/local/cuda-12.9/include/cub/iterator/counting_input_iterator.cuh:93:63: note: declared here 93 | class CCCL_DEPRECATED_BECAUSE("Use thrust::counting_iterator instead") CountingInputIterator | ^~~~~~~~~~~~~~~~~~~~~ ``` Pull Request resolved: #160554 Approved by: https://github.com/ngimel, https://github.com/malfet, https://github.com/atalman
1 parent c5efc5c commit d670304

File tree

1 file changed

+1
-1
lines changed

1 file changed

+1
-1
lines changed

aten/src/ATen/cuda/cub_definitions.cuh

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -54,7 +54,7 @@
5454

5555
// There were many bc-breaking changes in major version release of CCCL v3.0.0
5656
// Please see https://nvidia.github.io/cccl/cccl/3.0_migration_guide.html
57-
#if CUB_VERSION >= 300000
57+
#if CUB_VERSION >= 200800
5858
#define CUB_V3_PLUS() true
5959
#else
6060
#define CUB_V3_PLUS() false

0 commit comments

Comments
 (0)
0