Backport to 2.8: Specialize `cuda::std::numeric_limits` for FP8 types (#3478) #3492

bernhardmgruber · 2025-01-22T20:03:19Z

Merge first: Backport to 2.8: Implement cuda::std::numeric_limits for __half and __nv_bfloat16 (#3361) #3490

* implement builtins for huge val, nan and nans * change `INFINITY` and `NAN` implementation for NVRTC

NVIDIA#3361) * implement `cuda::std::numeric_limits` for `__half` and `__nv_bfloat16`

github-actions · 2025-01-23T00:15:53Z

🟨 CI finished in 4h 08m: Pass: 95%/170 | Total: 3d 15h | Avg: 31m 02s | Max: 1h 17m | Hits: 262%/22584

🟨 libcudacxx: Pass: 93%/48 | Total: 19h 01m | Avg: 23m 47s | Max: 1h 10m | Hits: 452%/9880

🔍 cpu: amd64 🔍
  🔍 amd64              Pass:  93%/46  | Total: 18h 21m | Avg: 23m 57s | Max:  1h 10m | Hits: 452%/9880  
  🟩 arm64              Pass: 100%/2   | Total: 40m 04s | Avg: 20m 02s | Max: 20m 40s
🔍 cudacxx_family: nvcc 🔍
  🟩 ClangCUDA          Pass: 100%/4   | Total:  1h 10m | Avg: 17m 38s | Max: 23m 08s
  🔍 nvcc               Pass:  93%/44  | Total: 17h 51m | Avg: 24m 20s | Max:  1h 10m | Hits: 452%/9880  
🟨 ctk
  🟨 11.1               Pass:  71%/7   | Total:  2h 00m | Avg: 17m 12s | Max: 32m 46s | Hits: 462%/2287  
  🟩 12.5               Pass: 100%/2   | Total:  1h 06m | Avg: 33m 06s | Max: 36m 13s
  🟨 12.6               Pass:  97%/39  | Total: 15h 55m | Avg: 24m 29s | Max:  1h 10m | Hits: 449%/7593  
🟨 cudacxx
  🟩 ClangCUDA18        Pass: 100%/4   | Total:  1h 10m | Avg: 17m 38s | Max: 23m 08s
  🟨 nvcc11.1           Pass:  71%/7   | Total:  2h 00m | Avg: 17m 12s | Max: 32m 46s | Hits: 462%/2287  
  🟩 nvcc12.5           Pass: 100%/2   | Total:  1h 06m | Avg: 33m 06s | Max: 36m 13s
  🟨 nvcc12.6           Pass:  97%/35  | Total: 14h 44m | Avg: 25m 16s | Max:  1h 10m | Hits: 449%/7593  
🟨 cxx
  🟨 Clang9             Pass:  75%/4   | Total:  1h 12m | Avg: 18m 10s | Max: 21m 57s
  🟩 Clang10            Pass: 100%/1   | Total: 23m 46s | Avg: 23m 46s | Max: 23m 46s
  🟩 Clang11            Pass: 100%/1   | Total: 18m 55s | Avg: 18m 55s | Max: 18m 55s
  🟩 Clang12            Pass: 100%/1   | Total: 18m 18s | Avg: 18m 18s | Max: 18m 18s
  🟩 Clang13            Pass: 100%/1   | Total: 22m 06s | Avg: 22m 06s | Max: 22m 06s
  🟩 Clang14            Pass: 100%/1   | Total: 23m 36s | Avg: 23m 36s | Max: 23m 36s
  🟩 Clang15            Pass: 100%/1   | Total: 25m 14s | Avg: 25m 14s | Max: 25m 14s
  🟩 Clang16            Pass: 100%/1   | Total: 21m 11s | Avg: 21m 11s | Max: 21m 11s
  🟩 Clang17            Pass: 100%/1   | Total: 22m 09s | Avg: 22m 09s | Max: 22m 09s
  🟩 Clang18            Pass: 100%/8   | Total:  3h 23m | Avg: 25m 24s | Max:  1h 10m
  🟩 GCC6               Pass: 100%/2   | Total: 19m 38s | Avg:  9m 49s | Max: 16m 40s
  🟩 GCC7               Pass: 100%/2   | Total: 32m 59s | Avg: 16m 29s | Max: 17m 51s
  🟩 GCC8               Pass: 100%/1   | Total: 18m 44s | Avg: 18m 44s | Max: 18m 44s
  🟨 GCC9               Pass:  66%/3   | Total: 55m 12s | Avg: 18m 24s | Max: 22m 22s
  🟩 GCC10              Pass: 100%/1   | Total: 22m 47s | Avg: 22m 47s | Max: 22m 47s
  🟩 GCC11              Pass: 100%/1   | Total: 21m 30s | Avg: 21m 30s | Max: 21m 30s
  🟩 GCC12              Pass: 100%/1   | Total: 24m 16s | Avg: 24m 16s | Max: 24m 16s
  🟨 GCC13              Pass:  90%/10  | Total:  4h 16m | Avg: 25m 37s | Max:  1h 10m
  🟩 Intel2023.2.0      Pass: 100%/1   | Total: 22m 35s | Avg: 22m 35s | Max: 22m 35s
  🟩 MSVC14.16          Pass: 100%/1   | Total: 32m 46s | Avg: 32m 46s | Max: 32m 46s | Hits: 462%/2287  
  🟩 MSVC14.29          Pass: 100%/1   | Total: 36m 53s | Avg: 36m 53s | Max: 36m 53s | Hits: 450%/2482  
  🟩 MSVC14.39          Pass: 100%/2   | Total:  1h 20m | Avg: 40m 25s | Max: 45m 16s | Hits: 449%/5111  
  🟩 NVHPC24.7          Pass: 100%/2   | Total:  1h 06m | Avg: 33m 06s | Max: 36m 13s
🟨 cxx_family
  🟨 Clang              Pass:  95%/20  | Total:  7h 31m | Avg: 22m 33s | Max:  1h 10m
  🟨 GCC                Pass:  90%/21  | Total:  7h 31m | Avg: 21m 29s | Max:  1h 10m
  🟩 Intel              Pass: 100%/1   | Total: 22m 35s | Avg: 22m 35s | Max: 22m 35s
  🟩 MSVC               Pass: 100%/4   | Total:  2h 30m | Avg: 37m 37s | Max: 45m 16s | Hits: 452%/9880  
  🟩 NVHPC              Pass: 100%/2   | Total:  1h 06m | Avg: 33m 06s | Max: 36m 13s
🟨 jobs
  🟨 Build              Pass:  95%/41  | Total: 14h 53m | Avg: 21m 48s | Max: 45m 16s | Hits: 452%/9880  
  🟩 NVRTC              Pass: 100%/4   | Total:  1h 45m | Avg: 26m 16s | Max: 29m 13s
  🟨 Test               Pass:  50%/2   | Total:  2h 20m | Avg:  1h 10m | Max:  1h 10m
  🟩 VerifyCodegen      Pass: 100%/1   | Total:  1m 58s | Avg:  1m 58s | Max:  1m 58s
🟨 std
  🟩 11                 Pass: 100%/6   | Total:  2h 01m | Avg: 20m 16s | Max: 27m 45s
  🟩 14                 Pass: 100%/5   | Total:  1h 37m | Avg: 19m 35s | Max: 32m 46s | Hits: 462%/2287  
  🟨 17                 Pass:  84%/13  | Total:  4h 57m | Avg: 22m 54s | Max: 36m 53s | Hits: 450%/4964  
  🟨 20                 Pass:  95%/23  | Total: 10h 22m | Avg: 27m 03s | Max:  1h 10m | Hits: 448%/2629  
🟨 gpu
  🟨 v100               Pass:  93%/48  | Total: 19h 01m | Avg: 23m 47s | Max:  1h 10m | Hits: 452%/9880  
🟩 sm
  🟩 90                 Pass: 100%/1   | Total: 13m 21s | Avg: 13m 21s | Max: 13m 21s
  🟩 90a                Pass: 100%/2   | Total: 29m 13s | Avg: 14m 36s | Max: 15m 42s

🟨 cub: Pass: 95%/47 | Total: 1d 13h | Avg: 47m 18s | Max: 1h 13m | Hits: 33%/3132

🔍 cpu: amd64 🔍
  🔍 amd64              Pass:  95%/45  | Total:  1d 11h | Avg: 46m 42s | Max:  1h 13m | Hits:  33%/3132  
  🟩 arm64              Pass: 100%/2   | Total:  2h 01m | Avg:  1h 00m | Max:  1h 05m
🔍 ctk: 11.1 🔍
  🔍 11.1               Pass:  71%/7   | Total:  3h 21m | Avg: 28m 50s | Max:  1h 00m | Hits:  33%/783   
  🟩 12.5               Pass: 100%/2   | Total:  2h 22m | Avg:  1h 11m | Max:  1h 13m
  🟩 12.6               Pass: 100%/38  | Total:  1d 07h | Avg: 49m 27s | Max:  1h 12m | Hits:  33%/2349  
🔍 cudacxx: nvcc11.1 🔍
  🟩 ClangCUDA18        Pass: 100%/2   | Total:  2h 08m | Avg:  1h 04m | Max:  1h 12m
  🔍 nvcc11.1           Pass:  71%/7   | Total:  3h 21m | Avg: 28m 50s | Max:  1h 00m | Hits:  33%/783   
  🟩 nvcc12.5           Pass: 100%/2   | Total:  2h 22m | Avg:  1h 11m | Max:  1h 13m
  🟩 nvcc12.6           Pass: 100%/36  | Total:  1d 05h | Avg: 48m 38s | Max:  1h 11m | Hits:  33%/2349  
🔍 cudacxx_family: nvcc 🔍
  🟩 ClangCUDA          Pass: 100%/2   | Total:  2h 08m | Avg:  1h 04m | Max:  1h 12m
  🔍 nvcc               Pass:  95%/45  | Total:  1d 10h | Avg: 46m 33s | Max:  1h 13m | Hits:  33%/3132  
🔍 gpu: v100 🔍
  🟩 h100               Pass: 100%/2   | Total: 39m 05s | Avg: 19m 32s | Max: 22m 53s
  🔍 v100               Pass:  95%/45  | Total:  1d 12h | Avg: 48m 32s | Max:  1h 13m | Hits:  33%/3132  
🔍 jobs: Build 🔍
  🔍 Build              Pass:  95%/40  | Total:  1d 10h | Avg: 52m 02s | Max:  1h 13m | Hits:  33%/3132  
  🟩 DeviceLaunch       Pass: 100%/1   | Total: 19m 55s | Avg: 19m 55s | Max: 19m 55s
  🟩 GraphCapture       Pass: 100%/1   | Total: 16m 29s | Avg: 16m 29s | Max: 16m 29s
  🟩 HostLaunch         Pass: 100%/3   | Total: 54m 25s | Avg: 18m 08s | Max: 20m 29s
  🟩 TestGPU            Pass: 100%/2   | Total: 50m 40s | Avg: 25m 20s | Max: 26m 38s
🔍 std: 17 🔍
  🟩 11                 Pass: 100%/5   | Total:  3h 31m | Avg: 42m 18s | Max: 55m 15s
  🟩 14                 Pass: 100%/4   | Total:  3h 27m | Avg: 51m 57s | Max:  1h 00m | Hits:  33%/783   
  🔍 17                 Pass:  83%/12  | Total: 10h 07m | Avg: 50m 39s | Max:  1h 08m | Hits:  32%/1566  
  🟩 20                 Pass: 100%/26  | Total: 19h 56m | Avg: 46m 00s | Max:  1h 13m | Hits:  33%/783   
🟨 cxx
  🟨 Clang9             Pass:  75%/4   | Total:  2h 25m | Avg: 36m 22s | Max: 56m 41s
  🟩 Clang10            Pass: 100%/1   | Total: 55m 10s | Avg: 55m 10s | Max: 55m 10s
  🟩 Clang11            Pass: 100%/1   | Total: 51m 09s | Avg: 51m 09s | Max: 51m 09s
  🟩 Clang12            Pass: 100%/1   | Total: 59m 39s | Avg: 59m 39s | Max: 59m 39s
  🟩 Clang13            Pass: 100%/1   | Total: 53m 55s | Avg: 53m 55s | Max: 53m 55s
  🟩 Clang14            Pass: 100%/1   | Total: 57m 59s | Avg: 57m 59s | Max: 57m 59s
  🟩 Clang15            Pass: 100%/1   | Total: 52m 03s | Avg: 52m 03s | Max: 52m 03s
  🟩 Clang16            Pass: 100%/1   | Total: 56m 39s | Avg: 56m 39s | Max: 56m 39s
  🟩 Clang17            Pass: 100%/1   | Total: 54m 37s | Avg: 54m 37s | Max: 54m 37s
  🟩 Clang18            Pass: 100%/7   | Total:  5h 41m | Avg: 48m 46s | Max:  1h 12m
  🟩 GCC6               Pass: 100%/2   | Total:  1h 09m | Avg: 34m 36s | Max: 36m 00s
  🟩 GCC7               Pass: 100%/2   | Total:  1h 52m | Avg: 56m 12s | Max: 57m 10s
  🟩 GCC8               Pass: 100%/1   | Total: 55m 21s | Avg: 55m 21s | Max: 55m 21s
  🟨 GCC9               Pass:  66%/3   | Total:  1h 34m | Avg: 31m 39s | Max: 58m 43s
  🟩 GCC10              Pass: 100%/1   | Total: 57m 46s | Avg: 57m 46s | Max: 57m 46s
  🟩 GCC11              Pass: 100%/1   | Total: 58m 36s | Avg: 58m 36s | Max: 58m 36s
  🟩 GCC12              Pass: 100%/3   | Total:  1h 34m | Avg: 31m 23s | Max: 55m 06s
  🟩 GCC13              Pass: 100%/8   | Total:  4h 40m | Avg: 35m 00s | Max:  1h 05m
  🟩 Intel2023.2.0      Pass: 100%/1   | Total:  1h 05m | Avg:  1h 05m | Max:  1h 05m
  🟩 MSVC14.16          Pass: 100%/1   | Total:  1h 00m | Avg:  1h 00m | Max:  1h 00m | Hits:  33%/783   
  🟩 MSVC14.29          Pass: 100%/1   | Total:  1h 04m | Avg:  1h 04m | Max:  1h 04m | Hits:  32%/783   
  🟩 MSVC14.39          Pass: 100%/2   | Total:  2h 20m | Avg:  1h 10m | Max:  1h 11m | Hits:  33%/1566  
  🟩 NVHPC24.7          Pass: 100%/2   | Total:  2h 22m | Avg:  1h 11m | Max:  1h 13m
🟨 cxx_family
  🟨 Clang              Pass:  94%/19  | Total: 15h 28m | Avg: 48m 50s | Max:  1h 12m
  🟨 GCC                Pass:  95%/21  | Total: 13h 42m | Avg: 39m 10s | Max:  1h 05m
  🟩 Intel              Pass: 100%/1   | Total:  1h 05m | Avg:  1h 05m | Max:  1h 05m
  🟩 MSVC               Pass: 100%/4   | Total:  4h 25m | Avg:  1h 06m | Max:  1h 11m | Hits:  33%/3132  
  🟩 NVHPC              Pass: 100%/2   | Total:  2h 22m | Avg:  1h 11m | Max:  1h 13m
🟩 sm
  🟩 90                 Pass: 100%/2   | Total: 39m 05s | Avg: 19m 32s | Max: 22m 53s
  🟩 90a                Pass: 100%/1   | Total: 25m 25s | Avg: 25m 25s | Max: 25m 25s

🟨 thrust: Pass: 95%/46 | Total: 1d 00h | Avg: 32m 30s | Max: 1h 17m | Hits: 138%/9260

🔍 cpu: amd64 🔍
  🔍 amd64              Pass:  95%/44  | Total: 23h 46m | Avg: 32m 25s | Max:  1h 17m | Hits: 138%/9260  
  🟩 arm64              Pass: 100%/2   | Total:  1h 08m | Avg: 34m 28s | Max: 35m 43s
🔍 ctk: 11.1 🔍
  🔍 11.1               Pass:  71%/7   | Total:  1h 57m | Avg: 16m 45s | Max:  1h 05m | Hits:  89%/1852  
  🟩 12.5               Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 12m
  🟩 12.6               Pass: 100%/37  | Total: 20h 37m | Avg: 33m 26s | Max:  1h 17m | Hits: 150%/7408  
🔍 cudacxx: nvcc11.1 🔍
  🟩 ClangCUDA18        Pass: 100%/2   | Total: 57m 22s | Avg: 28m 41s | Max: 30m 02s
  🔍 nvcc11.1           Pass:  71%/7   | Total:  1h 57m | Avg: 16m 45s | Max:  1h 05m | Hits:  89%/1852  
  🟩 nvcc12.5           Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 12m
  🟩 nvcc12.6           Pass: 100%/35  | Total: 19h 39m | Avg: 33m 42s | Max:  1h 17m | Hits: 150%/7408  
🔍 cudacxx_family: nvcc 🔍
  🟩 ClangCUDA          Pass: 100%/2   | Total: 57m 22s | Avg: 28m 41s | Max: 30m 02s
  🔍 nvcc               Pass:  95%/44  | Total: 23h 58m | Avg: 32m 41s | Max:  1h 17m | Hits: 138%/9260  
🔍 jobs: Build 🔍
  🔍 Build              Pass:  95%/40  | Total: 23h 21m | Avg: 35m 02s | Max:  1h 17m | Hits:  81%/7408  
  🟩 TestCPU            Pass: 100%/3   | Total: 52m 15s | Avg: 17m 25s | Max: 35m 58s | Hits: 365%/1852  
  🟩 TestGPU            Pass: 100%/3   | Total: 41m 52s | Avg: 13m 57s | Max: 16m 21s
🔍 std: 17 🔍
  🟩 11                 Pass: 100%/5   | Total:  1h 24m | Avg: 16m 54s | Max: 28m 31s
  🟩 14                 Pass: 100%/4   | Total:  2h 30m | Avg: 37m 44s | Max:  1h 05m | Hits:  89%/1852  
  🔍 17                 Pass:  83%/12  | Total:  7h 42m | Avg: 38m 34s | Max:  1h 12m | Hits:  78%/3704  
  🟩 20                 Pass: 100%/23  | Total: 12h 33m | Avg: 32m 46s | Max:  1h 17m | Hits: 222%/3704  
🟨 cxx
  🟨 Clang9             Pass:  75%/4   | Total:  1h 12m | Avg: 18m 13s | Max: 33m 35s
  🟩 Clang10            Pass: 100%/1   | Total: 31m 24s | Avg: 31m 24s | Max: 31m 24s
  🟩 Clang11            Pass: 100%/1   | Total: 31m 42s | Avg: 31m 42s | Max: 31m 42s
  🟩 Clang12            Pass: 100%/1   | Total: 36m 06s | Avg: 36m 06s | Max: 36m 06s
  🟩 Clang13            Pass: 100%/1   | Total: 34m 22s | Avg: 34m 22s | Max: 34m 22s
  🟩 Clang14            Pass: 100%/1   | Total: 32m 20s | Avg: 32m 20s | Max: 32m 20s
  🟩 Clang15            Pass: 100%/1   | Total: 32m 36s | Avg: 32m 36s | Max: 32m 36s
  🟩 Clang16            Pass: 100%/1   | Total: 33m 42s | Avg: 33m 42s | Max: 33m 42s
  🟩 Clang17            Pass: 100%/1   | Total: 36m 41s | Avg: 36m 41s | Max: 36m 41s
  🟩 Clang18            Pass: 100%/7   | Total:  2h 54m | Avg: 24m 57s | Max: 33m 52s
  🟩 GCC6               Pass: 100%/2   | Total: 30m 55s | Avg: 15m 27s | Max: 18m 26s
  🟩 GCC7               Pass: 100%/2   | Total: 59m 59s | Avg: 29m 59s | Max: 33m 38s
  🟩 GCC8               Pass: 100%/1   | Total: 30m 58s | Avg: 30m 58s | Max: 30m 58s
  🟨 GCC9               Pass:  66%/3   | Total: 45m 35s | Avg: 15m 11s | Max: 35m 15s
  🟩 GCC10              Pass: 100%/1   | Total: 35m 06s | Avg: 35m 06s | Max: 35m 06s
  🟩 GCC11              Pass: 100%/1   | Total: 36m 33s | Avg: 36m 33s | Max: 36m 33s
  🟩 GCC12              Pass: 100%/1   | Total: 35m 43s | Avg: 35m 43s | Max: 35m 43s
  🟩 GCC13              Pass: 100%/8   | Total:  3h 07m | Avg: 23m 25s | Max: 35m 43s
  🟩 Intel2023.2.0      Pass: 100%/1   | Total: 53m 02s | Avg: 53m 02s | Max: 53m 02s
  🟩 MSVC14.16          Pass: 100%/1   | Total:  1h 05m | Avg:  1h 05m | Max:  1h 05m | Hits:  89%/1852  
  🟩 MSVC14.29          Pass: 100%/1   | Total:  1h 11m | Avg:  1h 11m | Max:  1h 11m | Hits:  78%/1852  
  🟩 MSVC14.39          Pass: 100%/3   | Total:  3h 06m | Avg:  1h 02m | Max:  1h 17m | Hits: 174%/5556  
  🟩 NVHPC24.7          Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 12m
🟨 cxx_family
  🟨 Clang              Pass:  94%/19  | Total:  8h 36m | Avg: 27m 11s | Max: 36m 41s
  🟨 GCC                Pass:  94%/19  | Total:  7h 42m | Avg: 24m 19s | Max: 36m 33s
  🟩 Intel              Pass: 100%/1   | Total: 53m 02s | Avg: 53m 02s | Max: 53m 02s
  🟩 MSVC               Pass: 100%/5   | Total:  5h 22m | Avg:  1h 04m | Max:  1h 17m | Hits: 138%/9260  
  🟩 NVHPC              Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 12m
🟨 gpu
  🟨 v100               Pass:  95%/46  | Total:  1d 00h | Avg: 32m 30s | Max:  1h 17m | Hits: 138%/9260  
🟩 cmake_options
  🟩 -DTHRUST_DISPATCH_TYPE=Force32bit Pass: 100%/2   | Total: 43m 39s | Avg: 21m 49s | Max: 27m 18s
🟩 sm
  🟩 90a                Pass: 100%/1   | Total: 19m 52s | Avg: 19m 52s | Max: 19m 52s

🟩 cudax: Pass: 100%/26 | Total: 6h 14m | Avg: 14m 23s | Max: 23m 54s | Hits: 223%/312

🟩 cpu
  🟩 amd64              Pass: 100%/22  | Total:  5h 19m | Avg: 14m 32s | Max: 23m 54s | Hits: 223%/312   
  🟩 arm64              Pass: 100%/4   | Total: 54m 19s | Avg: 13m 34s | Max: 14m 45s
🟩 ctk
  🟩 12.0               Pass: 100%/3   | Total: 41m 34s | Avg: 13m 51s | Max: 14m 45s | Hits: 223%/156   
  🟩 12.5               Pass: 100%/2   | Total: 18m 05s | Avg:  9m 02s | Max:  9m 10s
  🟩 12.6               Pass: 100%/21  | Total:  5h 14m | Avg: 14m 58s | Max: 23m 54s | Hits: 223%/156   
🟩 cudacxx
  🟩 nvcc12.0           Pass: 100%/3   | Total: 41m 34s | Avg: 13m 51s | Max: 14m 45s | Hits: 223%/156   
  🟩 nvcc12.5           Pass: 100%/2   | Total: 18m 05s | Avg:  9m 02s | Max:  9m 10s
  🟩 nvcc12.6           Pass: 100%/21  | Total:  5h 14m | Avg: 14m 58s | Max: 23m 54s | Hits: 223%/156   
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/26  | Total:  6h 14m | Avg: 14m 23s | Max: 23m 54s | Hits: 223%/312   
🟩 cxx
  🟩 Clang9             Pass: 100%/1   | Total: 14m 45s | Avg: 14m 45s | Max: 14m 45s
  🟩 Clang10            Pass: 100%/1   | Total: 15m 19s | Avg: 15m 19s | Max: 15m 19s
  🟩 Clang11            Pass: 100%/1   | Total: 15m 28s | Avg: 15m 28s | Max: 15m 28s
  🟩 Clang12            Pass: 100%/1   | Total: 13m 24s | Avg: 13m 24s | Max: 13m 24s
  🟩 Clang13            Pass: 100%/1   | Total: 15m 16s | Avg: 15m 16s | Max: 15m 16s
  🟩 Clang14            Pass: 100%/1   | Total: 14m 11s | Avg: 14m 11s | Max: 14m 11s
  🟩 Clang15            Pass: 100%/1   | Total: 16m 06s | Avg: 16m 06s | Max: 16m 06s
  🟩 Clang16            Pass: 100%/1   | Total: 15m 43s | Avg: 15m 43s | Max: 15m 43s
  🟩 Clang17            Pass: 100%/1   | Total: 16m 39s | Avg: 16m 39s | Max: 16m 39s
  🟩 Clang18            Pass: 100%/4   | Total: 59m 08s | Avg: 14m 47s | Max: 18m 23s
  🟩 GCC9               Pass: 100%/1   | Total: 14m 03s | Avg: 14m 03s | Max: 14m 03s
  🟩 GCC10              Pass: 100%/1   | Total: 14m 35s | Avg: 14m 35s | Max: 14m 35s
  🟩 GCC11              Pass: 100%/1   | Total: 16m 18s | Avg: 16m 18s | Max: 16m 18s
  🟩 GCC12              Pass: 100%/2   | Total: 39m 52s | Avg: 19m 56s | Max: 23m 54s
  🟩 GCC13              Pass: 100%/4   | Total: 49m 41s | Avg: 12m 25s | Max: 14m 45s
  🟩 MSVC14.36          Pass: 100%/1   | Total: 12m 46s | Avg: 12m 46s | Max: 12m 46s | Hits: 223%/156   
  🟩 MSVC14.39          Pass: 100%/1   | Total: 12m 55s | Avg: 12m 55s | Max: 12m 55s | Hits: 223%/156   
  🟩 NVHPC24.7          Pass: 100%/2   | Total: 18m 05s | Avg:  9m 02s | Max:  9m 10s
🟩 cxx_family
  🟩 Clang              Pass: 100%/13  | Total:  3h 15m | Avg: 15m 04s | Max: 18m 23s
  🟩 GCC                Pass: 100%/9   | Total:  2h 14m | Avg: 14m 56s | Max: 23m 54s
  🟩 MSVC               Pass: 100%/2   | Total: 25m 41s | Avg: 12m 50s | Max: 12m 55s | Hits: 223%/312   
  🟩 NVHPC              Pass: 100%/2   | Total: 18m 05s | Avg:  9m 02s | Max:  9m 10s
🟩 gpu
  🟩 v100               Pass: 100%/26  | Total:  6h 14m | Avg: 14m 23s | Max: 23m 54s | Hits: 223%/312   
🟩 jobs
  🟩 Build              Pass: 100%/24  | Total:  5h 31m | Avg: 13m 49s | Max: 16m 39s | Hits: 223%/312   
  🟩 Test               Pass: 100%/2   | Total: 42m 17s | Avg: 21m 08s | Max: 23m 54s
🟩 sm
  🟩 90                 Pass: 100%/1   | Total: 10m 15s | Avg: 10m 15s | Max: 10m 15s
  🟩 90a                Pass: 100%/1   | Total: 11m 06s | Avg: 11m 06s | Max: 11m 06s
🟩 std
  🟩 17                 Pass: 100%/6   | Total:  1h 14m | Avg: 12m 21s | Max: 14m 45s
  🟩 20                 Pass: 100%/20  | Total:  5h 00m | Avg: 15m 00s | Max: 23m 54s | Hits: 223%/312

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 11m 40s | Avg: 5m 50s | Max: 9m 25s

🟩 cpu
  🟩 amd64              Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 ctk
  🟩 12.6               Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 cudacxx
  🟩 nvcc12.6           Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 cxx
  🟩 GCC13              Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 cxx_family
  🟩 GCC                Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 gpu
  🟩 v100               Pass: 100%/2   | Total: 11m 40s | Avg:  5m 50s | Max:  9m 25s
🟩 jobs
  🟩 Build              Pass: 100%/1   | Total:  2m 15s | Avg:  2m 15s | Max:  2m 15s
  🟩 Test               Pass: 100%/1   | Total:  9m 25s | Avg:  9m 25s | Max:  9m 25s

🟩 python: Pass: 100%/1 | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s

🟩 cpu
  🟩 amd64              Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 ctk
  🟩 12.6               Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 cudacxx
  🟩 nvcc12.6           Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 cxx
  🟩 GCC13              Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 cxx_family
  🟩 GCC                Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 gpu
  🟩 v100               Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s
🟩 jobs
  🟩 Test               Pass: 100%/1   | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s

👃 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: 170)

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

davebayer added 3 commits January 22, 2025 20:59

Refactor limits and climits (NVIDIA#3221)

972076f

* implement builtins for huge val, nan and nans * change `INFINITY` and `NAN` implementation for NVRTC

Implement cuda::std::numeric_limits for __half and __nv_bfloat16 (

42ac268

NVIDIA#3361) * implement `cuda::std::numeric_limits` for `__half` and `__nv_bfloat16`

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

8b143ca

bernhardmgruber requested review from a team as code owners January 22, 2025 20:03

bernhardmgruber requested review from griwes and gonidelis January 22, 2025 20:03

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Backport to 2.8: Specialize `cuda::std::numeric_limits` for FP8 types (#3478) #3492

Backport to 2.8: Specialize `cuda::std::numeric_limits` for FP8 types (#3478) #3492

bernhardmgruber commented Jan 22, 2025

github-actions bot commented Jan 23, 2025

🟨 libcudacxx: Pass: 93%/48 | Total: 19h 01m | Avg: 23m 47s | Max: 1h 10m | Hits: 452%/9880

🟨 cub: Pass: 95%/47 | Total: 1d 13h | Avg: 47m 18s | Max: 1h 13m | Hits: 33%/3132

🟨 thrust: Pass: 95%/46 | Total: 1d 00h | Avg: 32m 30s | Max: 1h 17m | Hits: 138%/9260

🟩 cudax: Pass: 100%/26 | Total: 6h 14m | Avg: 14m 23s | Max: 23m 54s | Hits: 223%/312

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 11m 40s | Avg: 5m 50s | Max: 9m 25s

🟩 python: Pass: 100%/1 | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 170)

Backport to 2.8: Specialize cuda::std::numeric_limits for FP8 types (#3478) #3492

Are you sure you want to change the base?

Backport to 2.8: Specialize cuda::std::numeric_limits for FP8 types (#3478) #3492

Conversation

bernhardmgruber commented Jan 22, 2025

github-actions bot commented Jan 23, 2025

🟨 libcudacxx: Pass: 93%/48 | Total: 19h 01m | Avg: 23m 47s | Max: 1h 10m | Hits: 452%/9880

🟨 cub: Pass: 95%/47 | Total: 1d 13h | Avg: 47m 18s | Max: 1h 13m | Hits: 33%/3132

🟨 thrust: Pass: 95%/46 | Total: 1d 00h | Avg: 32m 30s | Max: 1h 17m | Hits: 138%/9260

🟩 cudax: Pass: 100%/26 | Total: 6h 14m | Avg: 14m 23s | Max: 23m 54s | Hits: 223%/312

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 11m 40s | Avg: 5m 50s | Max: 9m 25s

🟩 python: Pass: 100%/1 | Total: 29m 59s | Avg: 29m 59s | Max: 29m 59s

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 170)

Backport to 2.8: Specialize `cuda::std::numeric_limits` for FP8 types (#3478) #3492

Backport to 2.8: Specialize `cuda::std::numeric_limits` for FP8 types (#3478) #3492