Skip to content

Drop small deprecated entites - #3748

Merged
bernhardmgruber merged 10 commits into
NVIDIA:mainfrom
bernhardmgruber:drop
Feb 15, 2025
Merged

bernhardmgruber merged 10 commits into
NVIDIA:mainfrom
bernhardmgruber:drop

Conversation

@bernhardmgruber

@bernhardmgruber bernhardmgruber commented Feb 7, 2025 •

Copy link
Copy Markdown
Contributor
  • Drop access to cub::PolicyWrapper
  • Drop access to cub::KernelConfig
  • Drop access to triple_chevron
  • Drop thrust logical meta functions
  • Drop thrust macros from type_deduction.h
  • Drop deprecated CUB macros
  • Drop thrust::cuda_cub::terminate
  • Drop thrust::numeric_limits
  • Drop cub::IterateThreadStore
  • Drop cub::Swap
  • No SASS changes for cub.bench.copy.memcpy.base for SM86
  • No SASS changes SASS for cub.bench.radix_sort.pairs.base for SM86

@bernhardmgruber
bernhardmgruber requested review from a team as code owners February 7, 2025 18:04
@bernhardmgruber bernhardmgruber mentioned this pull request Feb 7, 2025
86 tasks done
@bernhardmgruber bernhardmgruber added thrust For all items related to Thrust. cub For all items related to CUB breaking Breaking change labels Feb 7, 2025
@copy-pr-bot

copy-pr-bot Bot commented Feb 11, 2025

Copy link
Copy Markdown
Contributor

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.

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

pre-commit.ci autofix

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

1 similar comment
@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

pre-commit.ci autofix

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

pre-commit.ci autofix

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

1 similar comment
@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

/ok to test

@elstehle

Copy link
Copy Markdown
Contributor

Given that this touches a few Agent files, would we want to check sass and, if sass changed, briefly double check performance? I think at leastMemcpy would be a candidate, potentially radix sort, if you have the chance to benchmark the legacy radix sort code path.

@miscco

miscco commented Feb 14, 2025

Copy link
Copy Markdown
Contributor

/ok to test

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

I see a fair amount of changes in the SASS of cub.bench.copy.memcpy.base for SM86 :/ Needs to be investigated.

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

I reorganized the commits so they compile individually. The bad commit is when we replaced CUB_MIN by cuda::std::min etc.

@bernhardmgruber

Copy link
Copy Markdown
Contributor Author

I removed the commit removing CUB_MIN and CUB_MAX. Now there are no SASS changes.

@github-actions

Copy link
Copy Markdown
Contributor
🟩 CI finished in 1h 45m: Pass: 100%/93 | Total: 2d 16h | Avg: 41m 24s | Max: 1h 24m | Hits: 63%/134193
  • 🟩 cub: Pass: 100%/45 | Total: 1d 17h | Avg: 55m 12s | Max: 1h 24m | Hits: 47%/53761

    🟩 cpu
      🟩 amd64              Pass: 100%/43  | Total:  1d 15h | Avg: 54m 58s | Max:  1h 24m | Hits:  47%/51319 
      🟩 arm64              Pass: 100%/2   | Total:  2h 00m | Avg:  1h 00m | Max:  1h 00m | Hits:  38%/2442  
    🟩 ctk
      🟩 12.0               Pass: 100%/5   | Total:  5h 11m | Avg:  1h 02m | Max:  1h 06m | Hits:  34%/5939  
      🟩 12.5               Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 10m | Hits:  28%/2260  
      🟩 12.8               Pass: 100%/38  | Total:  1d 09h | Avg: 53m 26s | Max:  1h 24m | Hits:  50%/45562 
    🟩 cudacxx
      🟩 ClangCUDA18        Pass: 100%/2   | Total:  2h 05m | Avg:  1h 02m | Max:  1h 05m | Hits:  40%/2114  
      🟩 nvcc12.0           Pass: 100%/5   | Total:  5h 11m | Avg:  1h 02m | Max:  1h 06m | Hits:  34%/5939  
      🟩 nvcc12.5           Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 10m | Hits:  28%/2260  
      🟩 nvcc12.8           Pass: 100%/36  | Total:  1d 07h | Avg: 52m 55s | Max:  1h 24m | Hits:  50%/43448 
    🟩 cudacxx_family
      🟩 ClangCUDA          Pass: 100%/2   | Total:  2h 05m | Avg:  1h 02m | Max:  1h 05m | Hits:  40%/2114  
      🟩 nvcc               Pass: 100%/43  | Total:  1d 15h | Avg: 54m 51s | Max:  1h 24m | Hits:  47%/51647 
    🟩 cxx
      🟩 Clang14            Pass: 100%/4   | Total:  4h 01m | Avg:  1h 00m | Max:  1h 03m | Hits:  39%/4892  
      🟩 Clang15            Pass: 100%/2   | Total:  2h 01m | Avg:  1h 00m | Max:  1h 03m | Hits:  39%/2442  
      🟩 Clang16            Pass: 100%/2   | Total:  2h 04m | Avg:  1h 02m | Max:  1h 04m | Hits:  39%/2442  
      🟩 Clang17            Pass: 100%/2   | Total:  1h 57m | Avg: 58m 47s | Max:  1h 02m | Hits:  39%/2442  
      🟩 Clang18            Pass: 100%/7   | Total:  5h 47m | Avg: 49m 37s | Max:  1h 05m | Hits:  57%/8219  
      🟩 GCC7               Pass: 100%/2   | Total:  2h 02m | Avg:  1h 01m | Max:  1h 03m | Hits:  38%/2446  
      🟩 GCC8               Pass: 100%/1   | Total: 55m 23s | Avg: 55m 23s | Max: 55m 23s | Hits:  38%/1223  
      🟩 GCC9               Pass: 100%/2   | Total:  2h 04m | Avg:  1h 02m | Max:  1h 06m | Hits:  38%/2446  
      🟩 GCC10              Pass: 100%/2   | Total:  2h 04m | Avg:  1h 02m | Max:  1h 03m | Hits:  38%/2446  
      🟩 GCC11              Pass: 100%/2   | Total:  2h 08m | Avg:  1h 04m | Max:  1h 04m | Hits:  38%/2442  
      🟩 GCC12              Pass: 100%/2   | Total:  2h 05m | Avg:  1h 02m | Max:  1h 05m | Hits:  38%/2442  
      🟩 GCC13              Pass: 100%/11  | Total:  6h 54m | Avg: 37m 38s | Max:  1h 12m | Hits:  71%/13431 
      🟩 MSVC14.29          Pass: 100%/2   | Total:  2h 18m | Avg:  1h 09m | Max:  1h 13m | Hits:  12%/2094  
      🟩 MSVC14.42          Pass: 100%/2   | Total:  2h 36m | Avg:  1h 18m | Max:  1h 24m | Hits:  12%/2094  
      🟩 NVHPC24.7          Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 10m | Hits:  28%/2260  
    🟩 cxx_family
      🟩 Clang              Pass: 100%/17  | Total: 15h 52m | Avg: 56m 03s | Max:  1h 05m | Hits:  46%/20437 
      🟩 GCC                Pass: 100%/22  | Total: 18h 14m | Avg: 49m 45s | Max:  1h 12m | Hits:  55%/26876 
      🟩 MSVC               Pass: 100%/4   | Total:  4h 55m | Avg:  1h 13m | Max:  1h 24m | Hits:  12%/4188  
      🟩 NVHPC              Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 10m | Hits:  28%/2260  
    🟩 gpu
      🟩 h100               Pass: 100%/3   | Total:  1h 15m | Avg: 25m 18s | Max: 29m 14s | Hits:  79%/3663  
      🟩 rtx2080            Pass: 100%/34  | Total:  1d 12h | Avg:  1h 03m | Max:  1h 24m | Hits:  35%/40330 
      🟩 rtxa6000           Pass: 100%/8   | Total:  4h 07m | Avg: 30m 56s | Max:  1h 04m | Hits:  84%/9768  
    🟩 jobs
      🟩 Build              Pass: 100%/37  | Total:  1d 14h | Avg:  1h 02m | Max:  1h 24m | Hits:  35%/43993 
      🟩 DeviceLaunch       Pass: 100%/1   | Total: 20m 58s | Avg: 20m 58s | Max: 20m 58s | Hits:  99%/1221  
      🟩 GraphCapture       Pass: 100%/1   | Total: 16m 31s | Avg: 16m 31s | Max: 16m 31s | Hits:  99%/1221  
      🟩 HostLaunch         Pass: 100%/3   | Total:  1h 13m | Avg: 24m 20s | Max: 25m 39s | Hits:  99%/3663  
      🟩 TestGPU            Pass: 100%/3   | Total:  1h 01m | Avg: 20m 31s | Max: 21m 13s | Hits:  99%/3663  
    🟩 sm
      🟩 90                 Pass: 100%/3   | Total:  1h 15m | Avg: 25m 18s | Max: 29m 14s | Hits:  79%/3663  
      🟩 90;90a;100         Pass: 100%/1   | Total:  1h 12m | Avg:  1h 12m | Max:  1h 12m | Hits:  38%/1221  
    🟩 std
      🟩 17                 Pass: 100%/20  | Total: 20h 42m | Avg:  1h 02m | Max:  1h 13m | Hits:  34%/23659 
      🟩 20                 Pass: 100%/25  | Total: 20h 41m | Avg: 49m 39s | Max:  1h 24m | Hits:  57%/30102 
    
  • 🟩 thrust: Pass: 100%/45 | Total: 22h 02m | Avg: 29m 23s | Max: 1h 08m | Hits: 74%/80136

    🟩 cmake_options
      🟩 -DTHRUST_DISPATCH_TYPE=Force32bit Pass: 100%/2   | Total: 36m 34s | Avg: 18m 17s | Max: 25m 30s | Hits:  88%/3564  
    🟩 cpu
      🟩 amd64              Pass: 100%/43  | Total: 21h 11m | Avg: 29m 34s | Max:  1h 08m | Hits:  74%/76573 
      🟩 arm64              Pass: 100%/2   | Total: 51m 05s | Avg: 25m 32s | Max: 26m 59s | Hits:  76%/3563  
    🟩 ctk
      🟩 12.0               Pass: 100%/5   | Total:  2h 44m | Avg: 32m 56s | Max: 58m 18s | Hits:  70%/8901  
      🟩 12.5               Pass: 100%/2   | Total:  1h 42m | Avg: 51m 03s | Max: 54m 01s | Hits:  55%/3562  
      🟩 12.8               Pass: 100%/38  | Total: 17h 35m | Avg: 27m 46s | Max:  1h 08m | Hits:  76%/67673 
    🟩 cudacxx
      🟩 ClangCUDA18        Pass: 100%/2   | Total: 46m 22s | Avg: 23m 11s | Max: 23m 30s | Hits:  77%/3562  
      🟩 nvcc12.0           Pass: 100%/5   | Total:  2h 44m | Avg: 32m 56s | Max: 58m 18s | Hits:  70%/8901  
      🟩 nvcc12.5           Pass: 100%/2   | Total:  1h 42m | Avg: 51m 03s | Max: 54m 01s | Hits:  55%/3562  
      🟩 nvcc12.8           Pass: 100%/36  | Total: 16h 49m | Avg: 28m 02s | Max:  1h 08m | Hits:  76%/64111 
    🟩 cudacxx_family
      🟩 ClangCUDA          Pass: 100%/2   | Total: 46m 22s | Avg: 23m 11s | Max: 23m 30s | Hits:  77%/3562  
      🟩 nvcc               Pass: 100%/43  | Total: 21h 16m | Avg: 29m 40s | Max:  1h 08m | Hits:  74%/76574 
    🟩 cxx
      🟩 Clang14            Pass: 100%/4   | Total:  1h 46m | Avg: 26m 39s | Max: 28m 04s | Hits:  77%/7124  
      🟩 Clang15            Pass: 100%/2   | Total: 53m 42s | Avg: 26m 51s | Max: 27m 05s | Hits:  77%/3562  
      🟩 Clang16            Pass: 100%/2   | Total: 58m 14s | Avg: 29m 07s | Max: 30m 31s | Hits:  77%/3562  
      🟩 Clang17            Pass: 100%/2   | Total: 56m 13s | Avg: 28m 06s | Max: 28m 20s | Hits:  77%/3562  
      🟩 Clang18            Pass: 100%/7   | Total:  2h 25m | Avg: 20m 50s | Max: 29m 58s | Hits:  83%/12467 
      🟩 GCC7               Pass: 100%/2   | Total: 54m 46s | Avg: 27m 23s | Max: 27m 47s | Hits:  76%/3564  
      🟩 GCC8               Pass: 100%/1   | Total: 29m 53s | Avg: 29m 53s | Max: 29m 53s | Hits:  76%/1782  
      🟩 GCC9               Pass: 100%/2   | Total: 57m 21s | Avg: 28m 40s | Max: 29m 15s | Hits:  76%/3564  
      🟩 GCC10              Pass: 100%/2   | Total: 56m 04s | Avg: 28m 02s | Max: 28m 40s | Hits:  76%/3564  
      🟩 GCC11              Pass: 100%/2   | Total: 58m 42s | Avg: 29m 21s | Max: 30m 08s | Hits:  76%/3564  
      🟩 GCC12              Pass: 100%/2   | Total: 59m 05s | Avg: 29m 32s | Max: 30m 58s | Hits:  76%/3564  
      🟩 GCC13              Pass: 100%/10  | Total:  3h 28m | Avg: 20m 52s | Max: 31m 48s | Hits:  83%/17820 
      🟩 MSVC14.29          Pass: 100%/2   | Total:  1h 56m | Avg: 58m 06s | Max: 58m 18s | Hits:  39%/3550  
      🟩 MSVC14.42          Pass: 100%/3   | Total:  2h 38m | Avg: 52m 58s | Max:  1h 08m | Hits:  44%/5325  
      🟩 NVHPC24.7          Pass: 100%/2   | Total:  1h 42m | Avg: 51m 03s | Max: 54m 01s | Hits:  55%/3562  
    🟩 cxx_family
      🟩 Clang              Pass: 100%/17  | Total:  7h 00m | Avg: 24m 44s | Max: 30m 31s | Hits:  79%/30277 
      🟩 GCC                Pass: 100%/21  | Total:  8h 44m | Avg: 24m 58s | Max: 31m 48s | Hits:  80%/37422 
      🟩 MSVC               Pass: 100%/5   | Total:  4h 35m | Avg: 55m 01s | Max:  1h 08m | Hits:  42%/8875  
      🟩 NVHPC              Pass: 100%/2   | Total:  1h 42m | Avg: 51m 03s | Max: 54m 01s | Hits:  55%/3562  
    🟩 gpu
      🟩 h100               Pass: 100%/2   | Total: 33m 42s | Avg: 16m 51s | Max: 17m 17s | Hits:  76%/3564  
      🟩 rtx2080            Pass: 100%/33  | Total: 17h 33m | Avg: 31m 55s | Max: 58m 45s | Hits:  72%/58769 
      🟩 rtx4090            Pass: 100%/10  | Total:  3h 55m | Avg: 23m 31s | Max:  1h 08m | Hits:  83%/17803 
    🟩 jobs
      🟩 Build              Pass: 100%/38  | Total: 20h 26m | Avg: 32m 17s | Max:  1h 08m | Hits:  71%/67671 
      🟩 TestCPU            Pass: 100%/3   | Total: 46m 43s | Avg: 15m 34s | Max: 31m 24s | Hits:  90%/5338  
      🟩 TestGPU            Pass: 100%/4   | Total: 48m 56s | Avg: 12m 14s | Max: 16m 25s | Hits:  94%/7127  
    🟩 sm
      🟩 90                 Pass: 100%/2   | Total: 33m 42s | Avg: 16m 51s | Max: 17m 17s | Hits:  76%/3564  
      🟩 90;90a;100         Pass: 100%/1   | Total: 31m 06s | Avg: 31m 06s | Max: 31m 06s | Hits:  76%/1782  
    🟩 std
      🟩 17                 Pass: 100%/20  | Total: 11h 14m | Avg: 33m 42s | Max: 58m 45s | Hits:  69%/35611 
      🟩 20                 Pass: 100%/23  | Total: 10h 11m | Avg: 26m 35s | Max:  1h 08m | Hits:  77%/40961 
    
  • 🟩 cccl_c_parallel: Pass: 100%/2 | Total: 13m 12s | Avg: 6m 36s | Max: 10m 41s | Hits: 97%/296

    🟩 cpu
      🟩 amd64              Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 ctk
      🟩 12.8               Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 cudacxx
      🟩 nvcc12.8           Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 cudacxx_family
      🟩 nvcc               Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 cxx
      🟩 GCC13              Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 cxx_family
      🟩 GCC                Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 gpu
      🟩 rtx2080            Pass: 100%/2   | Total: 13m 12s | Avg:  6m 36s | Max: 10m 41s | Hits:  97%/296   
    🟩 jobs
      🟩 Build              Pass: 100%/1   | Total:  2m 31s | Avg:  2m 31s | Max:  2m 31s | Hits:  95%/148   
      🟩 Test               Pass: 100%/1   | Total: 10m 41s | Avg: 10m 41s | Max: 10m 41s | Hits:  98%/148   
    
  • 🟩 python: Pass: 100%/1 | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s

    🟩 cpu
      🟩 amd64              Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 ctk
      🟩 12.8               Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 cudacxx
      🟩 nvcc12.8           Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 cudacxx_family
      🟩 nvcc               Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 cxx
      🟩 GCC13              Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 cxx_family
      🟩 GCC                Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 gpu
      🟩 rtx2080            Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    🟩 jobs
      🟩 Test               Pass: 100%/1   | Total: 30m 15s | Avg: 30m 15s | Max: 30m 15s
    

👃 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: 93)

# Runner
66 linux-amd64-cpu16
9 windows-amd64-cpu16
6 linux-amd64-gpu-rtxa6000-latest-1
4 linux-arm64-cpu16
3 linux-amd64-gpu-h100-latest-1
3 linux-amd64-gpu-rtx4090-latest-1
2 linux-amd64-gpu-rtx2080-latest-1

@bernhardmgruber
bernhardmgruber enabled auto-merge (squash) February 15, 2025 12:18
@elstehle

Copy link
Copy Markdown
Contributor

I removed the commit removing CUB_MIN and CUB_MAX. Now there are no SASS changes.

Thanks. I think we can integrate those in a follow-up PR after running the benchmarks.

@bernhardmgruber
bernhardmgruber merged commit d090fb6 into NVIDIA:main Feb 15, 2025
@bernhardmgruber
bernhardmgruber deleted the drop branch February 15, 2025 15:58
davebayer pushed a commit to davebayer/cccl that referenced this pull request Feb 20, 2025
* Drop cub::Swap
* Drop cub::IterateThreadStore
* Drop thrust::numeric_limits
* Drop thrust::cuda_cub::terminate
* Drop access to triple_chevron
* Drop access to cub::KernelConfig
* Drop access to cub::PolicyWrapper
* Drop access to cub::AliasTemporaries
* Drop thrust logical meta functions
* Drop thrust macros from type_deduction.h
davebayer pushed a commit to davebayer/cccl that referenced this pull request Apr 7, 2025
* Drop cub::Swap
* Drop cub::IterateThreadStore
* Drop thrust::numeric_limits
* Drop thrust::cuda_cub::terminate
* Drop access to triple_chevron
* Drop access to cub::KernelConfig
* Drop access to cub::PolicyWrapper
* Drop access to cub::AliasTemporaries
* Drop thrust logical meta functions
* Drop thrust macros from type_deduction.h
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

breaking Breaking change cub For all items related to CUB thrust For all items related to Thrust.

Projects

Archived in project

Development

Successfully merging this pull request may close these issues.

3 participants