Specialize `cuda::std::numeric_limits` for FP8 types #3478

davebayer · 2025-01-22T14:36:56Z

Partially implements #525.

copy-pr-bot · 2025-01-22T14:37:02Z

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.

libcudacxx/include/cuda/std/limits

miscco · 2025-01-22T15:18:26Z

/ok to test

libcudacxx/include/cuda/std/limits

…er (NVIDIA#3472)

Also ensure that we actually can enable FP8 due to FP16 and BF16 requirements Co-authored-by: Michael Schellenberger Costa <[email protected]>

* moves emptykernel to detail ns * second batch * third batch * fourth batch * fixes cuda parallel * concatenates nested namespaces

bernhardmgruber · 2025-01-22T16:30:23Z

I would like to backport this PR to CCCL 2.8, so I can also backport the deprecation of cub::Traits, so some limited functionality should work in C++11. We don't need constexpr support in C++11 though. And after the backport, you can cleanup the implementation to use C++17. So please don't invest too much to get everything working under C++11.

bernhardmgruber · 2025-01-22T16:36:12Z

/ok to test

bernhardmgruber · 2025-01-22T16:37:44Z

/ok to test

github-actions · 2025-01-22T19:03:34Z

🟩 CI finished in 2h 23m: Pass: 100%/135 | Total: 3d 05h | Avg: 34m 30s | Max: 1h 15m | Hits: 277%/23291

🟩 cub: Pass: 100%/38 | Total: 1d 09h | Avg: 52m 53s | Max: 1h 15m | Hits: 39%/3528

🟩 cpu
  🟩 amd64              Pass: 100%/36  | Total:  1d 07h | Avg: 52m 31s | Max:  1h 15m | Hits:  39%/3528  
  🟩 arm64              Pass: 100%/2   | Total:  1h 59m | Avg: 59m 40s | Max: 59m 50s
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  5h 00m | Avg:  1h 00m | Max:  1h 07m | Hits:  39%/882   
  🟩 12.5               Pass: 100%/2   | Total:  2h 24m | Avg:  1h 12m | Max:  1h 12m
  🟩 12.6               Pass: 100%/31  | Total:  1d 02h | Avg: 50m 29s | Max:  1h 15m | Hits:  39%/2646  
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/2   | Total:  2h 07m | Avg:  1h 03m | Max:  1h 05m
  🟩 nvcc12.0           Pass: 100%/5   | Total:  5h 00m | Avg:  1h 00m | Max:  1h 07m | Hits:  39%/882   
  🟩 nvcc12.5           Pass: 100%/2   | Total:  2h 24m | Avg:  1h 12m | Max:  1h 12m
  🟩 nvcc12.6           Pass: 100%/29  | Total: 23h 57m | Avg: 49m 34s | Max:  1h 15m | Hits:  39%/2646  
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/2   | Total:  2h 07m | Avg:  1h 03m | Max:  1h 05m
  🟩 nvcc               Pass: 100%/36  | Total:  1d 07h | Avg: 52m 17s | Max:  1h 15m | Hits:  39%/3528  
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  3h 57m | Avg: 59m 19s | Max:  1h 01m
  🟩 Clang15            Pass: 100%/1   | Total: 55m 30s | Avg: 55m 30s | Max: 55m 30s
  🟩 Clang16            Pass: 100%/1   | Total: 56m 19s | Avg: 56m 19s | Max: 56m 19s
  🟩 Clang17            Pass: 100%/1   | Total: 55m 46s | Avg: 55m 46s | Max: 55m 46s
  🟩 Clang18            Pass: 100%/7   | Total:  5h 49m | Avg: 49m 57s | Max:  1h 05m
  🟩 GCC7               Pass: 100%/2   | Total:  1h 53m | Avg: 56m 44s | Max: 57m 35s
  🟩 GCC8               Pass: 100%/1   | Total: 58m 15s | Avg: 58m 15s | Max: 58m 15s
  🟩 GCC9               Pass: 100%/2   | Total:  1h 55m | Avg: 57m 57s | Max:  1h 00m
  🟩 GCC10              Pass: 100%/1   | Total: 57m 21s | Avg: 57m 21s | Max: 57m 21s
  🟩 GCC11              Pass: 100%/1   | Total:  1h 02m | Avg:  1h 02m | Max:  1h 02m
  🟩 GCC12              Pass: 100%/3   | Total:  1h 45m | Avg: 35m 15s | Max: 59m 12s
  🟩 GCC13              Pass: 100%/8   | Total:  5h 18m | Avg: 39m 46s | Max:  1h 05m
  🟩 MSVC14.29          Pass: 100%/2   | Total:  2h 13m | Avg:  1h 06m | Max:  1h 07m | Hits:  39%/1764  
  🟩 MSVC14.39          Pass: 100%/2   | Total:  2h 26m | Avg:  1h 13m | Max:  1h 15m | Hits:  39%/1764  
  🟩 NVHPC24.7          Pass: 100%/2   | Total:  2h 24m | Avg:  1h 12m | Max:  1h 12m
🟩 cxx_family
  🟩 Clang              Pass: 100%/14  | Total: 12h 34m | Avg: 53m 53s | Max:  1h 05m
  🟩 GCC                Pass: 100%/18  | Total: 13h 51m | Avg: 46m 11s | Max:  1h 05m
  🟩 MSVC               Pass: 100%/4   | Total:  4h 40m | Avg:  1h 10m | Max:  1h 15m | Hits:  39%/3528  
  🟩 NVHPC              Pass: 100%/2   | Total:  2h 24m | Avg:  1h 12m | Max:  1h 12m
🟩 gpu
  🟩 h100               Pass: 100%/2   | Total: 46m 34s | Avg: 23m 17s | Max: 27m 14s
  🟩 v100               Pass: 100%/36  | Total:  1d 08h | Avg: 54m 32s | Max:  1h 15m | Hits:  39%/3528  
🟩 jobs
  🟩 Build              Pass: 100%/31  | Total:  1d 06h | Avg: 59m 28s | Max:  1h 15m | Hits:  39%/3528  
  🟩 DeviceLaunch       Pass: 100%/1   | Total: 19m 53s | Avg: 19m 53s | Max: 19m 53s
  🟩 GraphCapture       Pass: 100%/1   | Total: 25m 03s | Avg: 25m 03s | Max: 25m 03s
  🟩 HostLaunch         Pass: 100%/3   | Total:  1h 13m | Avg: 24m 31s | Max: 32m 58s
  🟩 TestGPU            Pass: 100%/2   | Total: 48m 00s | Avg: 24m 00s | Max: 26m 05s
🟩 sm
  🟩 90                 Pass: 100%/2   | Total: 46m 34s | Avg: 23m 17s | Max: 27m 14s
  🟩 90a                Pass: 100%/1   | Total: 28m 16s | Avg: 28m 16s | Max: 28m 16s
🟩 std
  🟩 17                 Pass: 100%/14  | Total: 14h 30m | Avg:  1h 02m | Max:  1h 11m | Hits:  39%/2646  
  🟩 20                 Pass: 100%/24  | Total: 18h 59m | Avg: 47m 28s | Max:  1h 15m | Hits:  39%/882

🟩 libcudacxx: Pass: 100%/37 | Total: 15h 39m | Avg: 25m 23s | Max: 1h 01m | Hits: 486%/10061

🟩 cpu
  🟩 amd64              Pass: 100%/35  | Total: 14h 56m | Avg: 25m 36s | Max:  1h 01m | Hits: 486%/10061 
  🟩 arm64              Pass: 100%/2   | Total: 43m 10s | Avg: 21m 35s | Max: 21m 36s
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  1h 48m | Avg: 21m 46s | Max: 33m 13s | Hits: 487%/2470  
  🟩 12.5               Pass: 100%/2   | Total:  1h 05m | Avg: 32m 50s | Max: 33m 55s
  🟩 12.6               Pass: 100%/30  | Total: 12h 44m | Avg: 25m 29s | Max:  1h 01m | Hits: 485%/7591  
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/4   | Total:  1h 10m | Avg: 17m 39s | Max: 22m 01s
  🟩 nvcc12.0           Pass: 100%/5   | Total:  1h 48m | Avg: 21m 46s | Max: 33m 13s | Hits: 487%/2470  
  🟩 nvcc12.5           Pass: 100%/2   | Total:  1h 05m | Avg: 32m 50s | Max: 33m 55s
  🟩 nvcc12.6           Pass: 100%/26  | Total: 11h 34m | Avg: 26m 42s | Max:  1h 01m | Hits: 485%/7591  
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/4   | Total:  1h 10m | Avg: 17m 39s | Max: 22m 01s
  🟩 nvcc               Pass: 100%/33  | Total: 14h 28m | Avg: 26m 19s | Max:  1h 01m | Hits: 486%/10061 
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  1h 20m | Avg: 20m 11s | Max: 23m 32s
  🟩 Clang15            Pass: 100%/1   | Total: 22m 15s | Avg: 22m 15s | Max: 22m 15s
  🟩 Clang16            Pass: 100%/1   | Total: 23m 18s | Avg: 23m 18s | Max: 23m 18s
  🟩 Clang17            Pass: 100%/1   | Total: 24m 02s | Avg: 24m 02s | Max: 24m 02s
  🟩 Clang18            Pass: 100%/8   | Total:  3h 17m | Avg: 24m 38s | Max:  1h 01m
  🟩 GCC7               Pass: 100%/2   | Total: 41m 45s | Avg: 20m 52s | Max: 21m 59s
  🟩 GCC8               Pass: 100%/1   | Total: 21m 48s | Avg: 21m 48s | Max: 21m 48s
  🟩 GCC9               Pass: 100%/2   | Total: 39m 44s | Avg: 19m 52s | Max: 20m 07s
  🟩 GCC10              Pass: 100%/1   | Total: 22m 16s | Avg: 22m 16s | Max: 22m 16s
  🟩 GCC11              Pass: 100%/1   | Total: 24m 26s | Avg: 24m 26s | Max: 24m 26s
  🟩 GCC12              Pass: 100%/1   | Total: 26m 10s | Avg: 26m 10s | Max: 26m 10s
  🟩 GCC13              Pass: 100%/8   | Total:  3h 25m | Avg: 25m 37s | Max:  1h 00m
  🟩 MSVC14.29          Pass: 100%/2   | Total:  1h 08m | Avg: 34m 05s | Max: 34m 58s | Hits: 487%/4950  
  🟩 MSVC14.39          Pass: 100%/2   | Total:  1h 17m | Avg: 38m 31s | Max: 39m 54s | Hits: 484%/5111  
  🟩 NVHPC24.7          Pass: 100%/2   | Total:  1h 05m | Avg: 32m 50s | Max: 33m 55s
🟩 cxx_family
  🟩 Clang              Pass: 100%/15  | Total:  5h 47m | Avg: 23m 09s | Max:  1h 01m
  🟩 GCC                Pass: 100%/16  | Total:  6h 21m | Avg: 23m 49s | Max:  1h 00m
  🟩 MSVC               Pass: 100%/4   | Total:  2h 25m | Avg: 36m 18s | Max: 39m 54s | Hits: 486%/10061 
  🟩 NVHPC              Pass: 100%/2   | Total:  1h 05m | Avg: 32m 50s | Max: 33m 55s
🟩 gpu
  🟩 v100               Pass: 100%/37  | Total: 15h 39m | Avg: 25m 23s | Max:  1h 01m | Hits: 486%/10061 
🟩 jobs
  🟩 Build              Pass: 100%/32  | Total: 12h 34m | Avg: 23m 33s | Max: 39m 54s | Hits: 486%/10061 
  🟩 NVRTC              Pass: 100%/2   | Total:  1h 01m | Avg: 30m 40s | Max: 30m 51s
  🟩 Test               Pass: 100%/2   | Total:  2h 01m | Avg:  1h 00m | Max:  1h 01m
  🟩 VerifyCodegen      Pass: 100%/1   | Total:  2m 07s | Avg:  2m 07s | Max:  2m 07s
🟩 sm
  🟩 90                 Pass: 100%/1   | Total: 13m 00s | Avg: 13m 00s | Max: 13m 00s
  🟩 90a                Pass: 100%/2   | Total: 31m 03s | Avg: 15m 31s | Max: 16m 34s
🟩 std
  🟩 17                 Pass: 100%/15  | Total:  6h 19m | Avg: 25m 17s | Max: 37m 08s | Hits: 481%/7430  
  🟩 20                 Pass: 100%/21  | Total:  9h 17m | Avg: 26m 34s | Max:  1h 01m | Hits: 499%/2631

🟩 thrust: Pass: 100%/37 | Total: 22h 48m | Avg: 36m 59s | Max: 1h 13m | Hits: 147%/9180

🟩 cmake_options
  🟩 -DTHRUST_DISPATCH_TYPE=Force32bit Pass: 100%/2   | Total: 45m 33s | Avg: 22m 46s | Max: 30m 44s
🟩 cpu
  🟩 amd64              Pass: 100%/35  | Total: 21h 44m | Avg: 37m 15s | Max:  1h 13m | Hits: 147%/9180  
  🟩 arm64              Pass: 100%/2   | Total:  1h 04m | Avg: 32m 05s | Max: 35m 10s
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  3h 28m | Avg: 41m 37s | Max:  1h 07m | Hits: 111%/1836  
  🟩 12.5               Pass: 100%/2   | Total:  2h 25m | Avg:  1h 12m | Max:  1h 13m
  🟩 12.6               Pass: 100%/30  | Total: 16h 55m | Avg: 33m 50s | Max:  1h 13m | Hits: 156%/7344  
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/2   | Total: 54m 38s | Avg: 27m 19s | Max: 27m 44s
  🟩 nvcc12.0           Pass: 100%/5   | Total:  3h 28m | Avg: 41m 37s | Max:  1h 07m | Hits: 111%/1836  
  🟩 nvcc12.5           Pass: 100%/2   | Total:  2h 25m | Avg:  1h 12m | Max:  1h 13m
  🟩 nvcc12.6           Pass: 100%/28  | Total: 16h 00m | Avg: 34m 17s | Max:  1h 13m | Hits: 156%/7344  
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/2   | Total: 54m 38s | Avg: 27m 19s | Max: 27m 44s
  🟩 nvcc               Pass: 100%/35  | Total: 21h 53m | Avg: 37m 32s | Max:  1h 13m | Hits: 147%/9180  
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  2h 11m | Avg: 32m 53s | Max: 33m 18s
  🟩 Clang15            Pass: 100%/1   | Total: 31m 35s | Avg: 31m 35s | Max: 31m 35s
  🟩 Clang16            Pass: 100%/1   | Total: 37m 41s | Avg: 37m 41s | Max: 37m 41s
  🟩 Clang17            Pass: 100%/1   | Total: 33m 16s | Avg: 33m 16s | Max: 33m 16s
  🟩 Clang18            Pass: 100%/7   | Total:  2h 54m | Avg: 24m 52s | Max: 32m 58s
  🟩 GCC7               Pass: 100%/2   | Total:  1h 11m | Avg: 35m 59s | Max: 36m 21s
  🟩 GCC8               Pass: 100%/1   | Total: 37m 40s | Avg: 37m 40s | Max: 37m 40s
  🟩 GCC9               Pass: 100%/2   | Total:  1h 14m | Avg: 37m 20s | Max: 38m 13s
  🟩 GCC10              Pass: 100%/1   | Total: 37m 30s | Avg: 37m 30s | Max: 37m 30s
  🟩 GCC11              Pass: 100%/1   | Total: 37m 53s | Avg: 37m 53s | Max: 37m 53s
  🟩 GCC12              Pass: 100%/1   | Total: 41m 08s | Avg: 41m 08s | Max: 41m 08s
  🟩 GCC13              Pass: 100%/8   | Total:  3h 21m | Avg: 25m 12s | Max: 39m 53s
  🟩 MSVC14.29          Pass: 100%/2   | Total:  2h 12m | Avg:  1h 06m | Max:  1h 07m | Hits:  96%/3672  
  🟩 MSVC14.39          Pass: 100%/3   | Total:  2h 59m | Avg: 59m 54s | Max:  1h 13m | Hits: 181%/5508  
  🟩 NVHPC24.7          Pass: 100%/2   | Total:  2h 25m | Avg:  1h 12m | Max:  1h 13m
🟩 cxx_family
  🟩 Clang              Pass: 100%/14  | Total:  6h 48m | Avg: 29m 09s | Max: 37m 41s
  🟩 GCC                Pass: 100%/16  | Total:  8h 22m | Avg: 31m 24s | Max: 41m 08s
  🟩 MSVC               Pass: 100%/5   | Total:  5h 12m | Avg:  1h 02m | Max:  1h 13m | Hits: 147%/9180  
  🟩 NVHPC              Pass: 100%/2   | Total:  2h 25m | Avg:  1h 12m | Max:  1h 13m
🟩 gpu
  🟩 v100               Pass: 100%/37  | Total: 22h 48m | Avg: 36m 59s | Max:  1h 13m | Hits: 147%/9180  
🟩 jobs
  🟩 Build              Pass: 100%/31  | Total: 21h 07m | Avg: 40m 54s | Max:  1h 13m | Hits:  92%/7344  
  🟩 TestCPU            Pass: 100%/3   | Total: 51m 19s | Avg: 17m 06s | Max: 35m 50s | Hits: 365%/1836  
  🟩 TestGPU            Pass: 100%/3   | Total: 49m 10s | Avg: 16m 23s | Max: 18m 37s
🟩 sm
  🟩 90a                Pass: 100%/1   | Total: 22m 28s | Avg: 22m 28s | Max: 22m 28s
🟩 std
  🟩 17                 Pass: 100%/14  | Total: 10h 27m | Avg: 44m 50s | Max:  1h 13m | Hits:  93%/5508  
  🟩 20                 Pass: 100%/21  | Total: 11h 35m | Avg: 33m 05s | Max:  1h 13m | Hits: 228%/3672

🟩 cudax: Pass: 100%/20 | Total: 4h 43m | Avg: 14m 11s | Max: 18m 27s | Hits: 158%/522

🟩 cpu
  🟩 amd64              Pass: 100%/16  | Total:  3h 49m | Avg: 14m 22s | Max: 18m 27s | Hits: 158%/522   
  🟩 arm64              Pass: 100%/4   | Total: 53m 51s | Avg: 13m 27s | Max: 14m 32s
🟩 ctk
  🟩 12.0               Pass: 100%/1   | Total: 12m 03s | Avg: 12m 03s | Max: 12m 03s | Hits: 159%/261   
  🟩 12.5               Pass: 100%/2   | Total: 17m 29s | Avg:  8m 44s | Max:  8m 54s
  🟩 12.6               Pass: 100%/17  | Total:  4h 14m | Avg: 14m 57s | Max: 18m 27s | Hits: 158%/261   
🟩 cudacxx
  🟩 nvcc12.0           Pass: 100%/1   | Total: 12m 03s | Avg: 12m 03s | Max: 12m 03s | Hits: 159%/261   
  🟩 nvcc12.5           Pass: 100%/2   | Total: 17m 29s | Avg:  8m 44s | Max:  8m 54s
  🟩 nvcc12.6           Pass: 100%/17  | Total:  4h 14m | Avg: 14m 57s | Max: 18m 27s | Hits: 158%/261   
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/20  | Total:  4h 43m | Avg: 14m 11s | Max: 18m 27s | Hits: 158%/522   
🟩 cxx
  🟩 Clang14            Pass: 100%/1   | Total: 13m 44s | Avg: 13m 44s | Max: 13m 44s
  🟩 Clang15            Pass: 100%/1   | Total: 16m 01s | Avg: 16m 01s | Max: 16m 01s
  🟩 Clang16            Pass: 100%/1   | Total: 16m 07s | Avg: 16m 07s | Max: 16m 07s
  🟩 Clang17            Pass: 100%/1   | Total: 16m 58s | Avg: 16m 58s | Max: 16m 58s
  🟩 Clang18            Pass: 100%/4   | Total:  1h 00m | Avg: 15m 02s | Max: 18m 16s
  🟩 GCC10              Pass: 100%/1   | Total: 16m 27s | Avg: 16m 27s | Max: 16m 27s
  🟩 GCC11              Pass: 100%/1   | Total: 16m 22s | Avg: 16m 22s | Max: 16m 22s
  🟩 GCC12              Pass: 100%/2   | Total: 36m 31s | Avg: 18m 15s | Max: 18m 27s
  🟩 GCC13              Pass: 100%/4   | Total: 49m 10s | Avg: 12m 17s | Max: 14m 32s
  🟩 MSVC14.36          Pass: 100%/1   | Total: 12m 03s | Avg: 12m 03s | Max: 12m 03s | Hits: 159%/261   
  🟩 MSVC14.39          Pass: 100%/1   | Total: 12m 48s | Avg: 12m 48s | Max: 12m 48s | Hits: 158%/261   
  🟩 NVHPC24.7          Pass: 100%/2   | Total: 17m 29s | Avg:  8m 44s | Max:  8m 54s
🟩 cxx_family
  🟩 Clang              Pass: 100%/8   | Total:  2h 03m | Avg: 15m 22s | Max: 18m 16s
  🟩 GCC                Pass: 100%/8   | Total:  1h 58m | Avg: 14m 48s | Max: 18m 27s
  🟩 MSVC               Pass: 100%/2   | Total: 24m 51s | Avg: 12m 25s | Max: 12m 48s | Hits: 158%/522   
  🟩 NVHPC              Pass: 100%/2   | Total: 17m 29s | Avg:  8m 44s | Max:  8m 54s
🟩 gpu
  🟩 v100               Pass: 100%/20  | Total:  4h 43m | Avg: 14m 11s | Max: 18m 27s | Hits: 158%/522   
🟩 jobs
  🟩 Build              Pass: 100%/18  | Total:  4h 07m | Avg: 13m 45s | Max: 18m 27s | Hits: 158%/522   
  🟩 Test               Pass: 100%/2   | Total: 36m 20s | Avg: 18m 10s | Max: 18m 16s
🟩 sm
  🟩 90                 Pass: 100%/1   | Total:  9m 51s | Avg:  9m 51s | Max:  9m 51s
  🟩 90a                Pass: 100%/1   | Total: 11m 39s | Avg: 11m 39s | Max: 11m 39s
🟩 std
  🟩 17                 Pass: 100%/4   | Total: 44m 14s | Avg: 11m 03s | Max: 13m 08s
  🟩 20                 Pass: 100%/16  | Total:  3h 59m | Avg: 14m 58s | Max: 18m 27s | Hits: 158%/522

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 9m 50s | Avg: 4m 55s | Max: 7m 40s

🟩 cpu
  🟩 amd64              Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 ctk
  🟩 12.6               Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 cudacxx
  🟩 nvcc12.6           Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 cxx
  🟩 GCC13              Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 cxx_family
  🟩 GCC                Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 gpu
  🟩 v100               Pass: 100%/2   | Total:  9m 50s | Avg:  4m 55s | Max:  7m 40s
🟩 jobs
  🟩 Build              Pass: 100%/1   | Total:  2m 10s | Avg:  2m 10s | Max:  2m 10s
  🟩 Test               Pass: 100%/1   | Total:  7m 40s | Avg:  7m 40s | Max:  7m 40s

🟩 python: Pass: 100%/1 | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s

🟩 cpu
  🟩 amd64              Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 ctk
  🟩 12.6               Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 cudacxx
  🟩 nvcc12.6           Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 cxx
  🟩 GCC13              Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 cxx_family
  🟩 GCC                Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 gpu
  🟩 v100               Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s
🟩 jobs
  🟩 Test               Pass: 100%/1   | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s

👃 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: 135)

#	Runner
92	`linux-amd64-cpu16`
17	`linux-amd64-gpu-v100-latest-1`
15	`windows-amd64-cpu16`
10	`linux-arm64-cpu16`
1	`linux-amd64-gpu-h100-latest-1-testing`

bernhardmgruber · 2025-01-22T19:07:05Z

Thanks a lot for making that happen!

github-actions · 2025-01-22T19:07:11Z

Backport failed for branch/2.8.x, because it was unable to cherry-pick the commit(s).

Please cherry-pick the changes locally.

git fetch origin branch/2.8.x
git worktree add -d .worktree/backport-3478-to-branch/2.8.x origin/branch/2.8.x
cd .worktree/backport-3478-to-branch/2.8.x
git checkout -b backport-3478-to-branch/2.8.x
ancref=$(git merge-base d47c1c1681546c67d3788a80d6458e2d910f2ba1 74d86dc7862543f7881af0905c138a097b2ba506)
git cherry-pick -x $ancref..74d86dc7862543f7881af0905c138a097b2ba506

Co-authored-by: David Bayer <[email protected]>

davebayer added 2 commits January 22, 2025 15:25

implement cuda::std::numeric_limits for NVFP8 types

02a7ca8

fix missing includes

dd0daf2

davebayer requested a review from a team as a code owner January 22, 2025 14:36

davebayer requested a review from fbusato January 22, 2025 14:36

remove accidentally pushed test files

7d38320

davebayer commented Jan 22, 2025

View reviewed changes

libcudacxx/include/cuda/std/limits Outdated Show resolved Hide resolved

libcudacxx/include/cuda/std/limits Outdated Show resolved Hide resolved

libcudacxx/include/cuda/std/limits Show resolved Hide resolved

miscco approved these changes Jan 22, 2025

View reviewed changes

use bit_cast if C++14 is unavailable

ba343ea

miscco reviewed Jan 22, 2025

View reviewed changes

libcudacxx/include/cuda/std/limits Outdated Show resolved Hide resolved

libcudacxx/include/cuda/std/limits Show resolved Hide resolved

remove blank line

eee3ef6

miscco approved these changes Jan 22, 2025

View reviewed changes

bernhardmgruber reviewed Jan 22, 2025

View reviewed changes

libcudacxx/include/cuda/std/limits Outdated Show resolved Hide resolved

davebayer and others added 5 commits January 22, 2025 17:19

prefer bitcast to set values

ec1bf86

Drop universal iterator categories (NVIDIA#3474)

1d27004

Ensure that headers in <cuda/*> can be build with a C++ only compil…

1532cea

…er (NVIDIA#3472)

Specialize __is_extended_floating_point for FP8 types (NVIDIA#3470)

18441e8

Also ensure that we actually can enable FP8 due to FP16 and BF16 requirements Co-authored-by: Michael Schellenberger Costa <[email protected]>

Moves CUB kernel entry points to a detail namespace (NVIDIA#3468)

f42268b

* moves emptykernel to detail ns * second batch * third batch * fourth batch * fixes cuda parallel * concatenates nested namespaces

davebayer requested review from a team as code owners January 22, 2025 16:20

davebayer requested review from bernhardmgruber and griwes January 22, 2025 16:20

update numeric limits type selection

7875a79

bernhardmgruber added the backport branch/2.8.x label Jan 22, 2025

Merge branch 'main' into fp8_numeric_limits

74d86dc

bernhardmgruber merged commit b1422c0 into NVIDIA:main Jan 22, 2025
146 of 149 checks passed

bernhardmgruber pushed a commit to bernhardmgruber/cccl that referenced this pull request Jan 22, 2025

Specialize cuda::std::numeric_limits for FP8 types (NVIDIA#3478)

8b143ca

fbusato mentioned this pull request Jan 27, 2025

Extend cuda::std::numeric_limits for __half, __nv_bfloat16, __nv_fp8_e4m3, __nv_fp8_e5m2 #3349

Closed

bernhardmgruber pushed a commit to bernhardmgruber/cccl that referenced this pull request Jan 28, 2025

Specialize cuda::std::numeric_limits for FP8 types (NVIDIA#3478)

dcd81e2

davebayer added a commit to davebayer/cccl that referenced this pull request Jan 29, 2025

Specialize cuda::std::numeric_limits for FP8 types (NVIDIA#3478)

d1bc32c

bernhardmgruber pushed a commit to bernhardmgruber/cccl that referenced this pull request Jan 30, 2025

Specialize cuda::std::numeric_limits for FP8 types (NVIDIA#3478)

5605766

bernhardmgruber pushed a commit to bernhardmgruber/cccl that referenced this pull request Jan 30, 2025

Specialize cuda::std::numeric_limits for FP8 types (NVIDIA#3478)

27cce48

bernhardmgruber added a commit that referenced this pull request Jan 30, 2025

Specialize cuda::std::numeric_limits for FP8 types (#3478) (#3492)

3a594e3

Co-authored-by: David Bayer <[email protected]>

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Specialize `cuda::std::numeric_limits` for FP8 types #3478

Specialize `cuda::std::numeric_limits` for FP8 types #3478

davebayer commented Jan 22, 2025

copy-pr-bot bot commented Jan 22, 2025

miscco commented Jan 22, 2025

bernhardmgruber commented Jan 22, 2025

bernhardmgruber commented Jan 22, 2025

bernhardmgruber commented Jan 22, 2025

github-actions bot commented Jan 22, 2025

🟩 cub: Pass: 100%/38 | Total: 1d 09h | Avg: 52m 53s | Max: 1h 15m | Hits: 39%/3528

🟩 libcudacxx: Pass: 100%/37 | Total: 15h 39m | Avg: 25m 23s | Max: 1h 01m | Hits: 486%/10061

🟩 thrust: Pass: 100%/37 | Total: 22h 48m | Avg: 36m 59s | Max: 1h 13m | Hits: 147%/9180

🟩 cudax: Pass: 100%/20 | Total: 4h 43m | Avg: 14m 11s | Max: 18m 27s | Hits: 158%/522

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 9m 50s | Avg: 4m 55s | Max: 7m 40s

🟩 python: Pass: 100%/1 | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 135)

bernhardmgruber commented Jan 22, 2025

github-actions bot commented Jan 22, 2025

Specialize cuda::std::numeric_limits for FP8 types #3478

Specialize cuda::std::numeric_limits for FP8 types #3478

Conversation

davebayer commented Jan 22, 2025

copy-pr-bot bot commented Jan 22, 2025

miscco commented Jan 22, 2025

bernhardmgruber commented Jan 22, 2025

bernhardmgruber commented Jan 22, 2025

bernhardmgruber commented Jan 22, 2025

github-actions bot commented Jan 22, 2025

🟩 cub: Pass: 100%/38 | Total: 1d 09h | Avg: 52m 53s | Max: 1h 15m | Hits: 39%/3528

🟩 libcudacxx: Pass: 100%/37 | Total: 15h 39m | Avg: 25m 23s | Max: 1h 01m | Hits: 486%/10061

🟩 thrust: Pass: 100%/37 | Total: 22h 48m | Avg: 36m 59s | Max: 1h 13m | Hits: 147%/9180

🟩 cudax: Pass: 100%/20 | Total: 4h 43m | Avg: 14m 11s | Max: 18m 27s | Hits: 158%/522

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 9m 50s | Avg: 4m 55s | Max: 7m 40s

🟩 python: Pass: 100%/1 | Total: 47m 07s | Avg: 47m 07s | Max: 47m 07s

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 135)

bernhardmgruber commented Jan 22, 2025

github-actions bot commented Jan 22, 2025

Specialize `cuda::std::numeric_limits` for FP8 types #3478

Specialize `cuda::std::numeric_limits` for FP8 types #3478