[Backport 3.0] Properly specialize cub functions for `__nv_bfloat16` (#6931) #6946

miscco · 2025-12-11T08:42:17Z

(cherry picked from commit e1f5bcc)

Description

closes

Checklist

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

(cherry picked from commit e1f5bcc)

github-actions · 2025-12-11T10:38:07Z

🟨 CI finished in 1h 53m: Pass: 99%/103 | Total: 3d 03h | Avg: 44m 05s | Max: 1h 24m | Hits: 62%/143100

🟥 python: Pass: 0%/1 | Total: 1h 01m | Avg: 1h 01m | Max: 1h 01m

🟥 cpu
  🟥 amd64              Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 ctk
  🟥 12.8               Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 cudacxx
  🟥 nvcc12.8           Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 cudacxx_family
  🟥 nvcc               Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 cxx
  🟥 GCC13              Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 cxx_family
  🟥 GCC                Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 gpu
  🟥 rtx2080            Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m
🟥 jobs
  🟥 Test               Pass:   0%/1   | Total:  1h 01m | Avg:  1h 01m | Max:  1h 01m

🟩 cub: Pass: 100%/48 | Total: 2d 01h | Avg: 1h 01m | Max: 1h 24m | Hits: 37%/57168

🟩 cpu
  🟩 amd64              Pass: 100%/46  | Total:  1d 23h | Avg:  1h 01m | Max:  1h 24m | Hits:  37%/54720 
  🟩 arm64              Pass: 100%/2   | Total:  2h 15m | Avg:  1h 07m | Max:  1h 09m | Hits:  21%/2448  
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  5h 48m | Avg:  1h 09m | Max:  1h 18m | Hits:  21%/5949  
  🟩 12.6               Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
  🟩 12.8               Pass: 100%/41  | Total:  1d 16h | Avg: 59m 50s | Max:  1h 24m | Hits:  39%/48965 
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/2   | Total:  1h 52m | Avg: 56m 23s | Max: 57m 40s | Hits:  91%/2110  
  🟩 nvcc12.0           Pass: 100%/5   | Total:  5h 48m | Avg:  1h 09m | Max:  1h 18m | Hits:  21%/5949  
  🟩 nvcc12.6           Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
  🟩 nvcc12.8           Pass: 100%/39  | Total:  1d 15h | Avg:  1h 00m | Max:  1h 24m | Hits:  37%/46855 
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/2   | Total:  1h 52m | Avg: 56m 23s | Max: 57m 40s | Hits:  91%/2110  
  🟩 nvcc               Pass: 100%/46  | Total:  1d 23h | Avg:  1h 01m | Max:  1h 24m | Hits:  35%/55058 
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  4h 22m | Avg:  1h 05m | Max:  1h 06m | Hits:  21%/4904  
  🟩 Clang15            Pass: 100%/2   | Total:  2h 16m | Avg:  1h 08m | Max:  1h 09m | Hits:  21%/2448  
  🟩 Clang16            Pass: 100%/2   | Total:  2h 16m | Avg:  1h 08m | Max:  1h 12m | Hits:  21%/2448  
  🟩 Clang17            Pass: 100%/2   | Total:  2h 18m | Avg:  1h 09m | Max:  1h 09m | Hits:  21%/2448  
  🟩 Clang18            Pass: 100%/7   | Total:  6h 23m | Avg: 54m 42s | Max:  1h 14m | Hits:  62%/8230  
  🟩 GCC7               Pass: 100%/2   | Total:  2h 14m | Avg:  1h 07m | Max:  1h 08m | Hits:  21%/2452  
  🟩 GCC8               Pass: 100%/1   | Total:  1h 04m | Avg:  1h 04m | Max:  1h 04m | Hits:  21%/1226  
  🟩 GCC9               Pass: 100%/2   | Total:  2h 24m | Avg:  1h 12m | Max:  1h 14m | Hits:  21%/2452  
  🟩 GCC10              Pass: 100%/2   | Total:  2h 11m | Avg:  1h 05m | Max:  1h 06m | Hits:  21%/2452  
  🟩 GCC11              Pass: 100%/2   | Total:  2h 11m | Avg:  1h 05m | Max:  1h 06m | Hits:  21%/2448  
  🟩 GCC12              Pass: 100%/2   | Total:  2h 12m | Avg:  1h 06m | Max:  1h 07m | Hits:  21%/2448  
  🟩 GCC13              Pass: 100%/12  | Total:  8h 58m | Avg: 44m 51s | Max:  1h 09m | Hits:  60%/14688 
  🟩 MSVC14.29          Pass: 100%/2   | Total:  2h 38m | Avg:  1h 19m | Max:  1h 19m | Hits:  19%/2090  
  🟩 MSVC14.42          Pass: 100%/4   | Total:  5h 09m | Avg:  1h 17m | Max:  1h 24m | Hits:  18%/4180  
  🟩 NVHPC25.1          Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
🟩 cxx_family
  🟩 Clang              Pass: 100%/17  | Total: 17h 36m | Avg:  1h 02m | Max:  1h 14m | Hits:  38%/20478 
  🟩 GCC                Pass: 100%/23  | Total: 21h 17m | Avg: 55m 33s | Max:  1h 14m | Hits:  41%/28166 
  🟩 MSVC               Pass: 100%/6   | Total:  7h 47m | Avg:  1h 17m | Max:  1h 24m | Hits:  18%/6270  
  🟩 NVHPC              Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
🟩 gpu
  🟩 h100               Pass: 100%/3   | Total:  1h 22m | Avg: 27m 39s | Max: 31m 52s | Hits:  73%/3672  
  🟩 rtx2080            Pass: 100%/37  | Total:  1d 18h | Avg:  1h 08m | Max:  1h 24m | Hits:  24%/43704 
  🟩 rtxa6000           Pass: 100%/8   | Total:  5h 35m | Avg: 41m 59s | Max:  1h 14m | Hits:  80%/9792  
🟩 jobs
  🟩 Build              Pass: 100%/40  | Total:  1d 21h | Avg:  1h 07m | Max:  1h 24m | Hits:  24%/47376 
  🟩 DeviceLaunch       Pass: 100%/1   | Total: 33m 11s | Avg: 33m 11s | Max: 33m 11s | Hits:  99%/1224  
  🟩 GraphCapture       Pass: 100%/1   | Total: 26m 44s | Avg: 26m 44s | Max: 26m 44s | Hits:  99%/1224  
  🟩 HostLaunch         Pass: 100%/3   | Total:  1h 34m | Avg: 31m 23s | Max: 37m 25s | Hits:  99%/3672  
  🟩 TestGPU            Pass: 100%/3   | Total:  1h 30m | Avg: 30m 19s | Max: 33m 24s | Hits:  99%/3672  
🟩 sm
  🟩 90                 Pass: 100%/3   | Total:  1h 22m | Avg: 27m 39s | Max: 31m 52s | Hits:  73%/3672  
  🟩 90;90a             Pass: 100%/2   | Total:  1h 59m | Avg: 59m 38s | Max:  1h 09m | Hits:  19%/2269  
  🟩 100;120            Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 13m | Hits:  20%/2269  
🟩 std
  🟩 17                 Pass: 100%/20  | Total: 23h 12m | Avg:  1h 09m | Max:  1h 24m | Hits:  24%/23693 
  🟩 20                 Pass: 100%/28  | Total:  1d 02h | Avg: 56m 07s | Max:  1h 24m | Hits:  46%/33475

🟩 thrust: Pass: 100%/48 | Total: 1d 00h | Avg: 30m 46s | Max: 1h 00m | Hits: 79%/85612

🟩 cmake_options
  🟩 -DTHRUST_DISPATCH_TYPE=Force32bit Pass: 100%/2   | Total: 42m 44s | Avg: 21m 22s | Max: 28m 26s | Hits:  88%/3570  
🟩 cpu
  🟩 amd64              Pass: 100%/46  | Total: 23h 42m | Avg: 30m 55s | Max:  1h 00m | Hits:  79%/82043 
  🟩 arm64              Pass: 100%/2   | Total: 54m 34s | Avg: 27m 17s | Max: 28m 55s | Hits:  76%/3569  
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  2h 57m | Avg: 35m 29s | Max: 54m 28s | Hits:  74%/8916  
  🟩 12.6               Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
  🟩 12.8               Pass: 100%/41  | Total: 19h 49m | Avg: 29m 01s | Max:  1h 00m | Hits:  80%/73128 
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/2   | Total:  9m 56s | Avg:  4m 58s | Max:  5m 12s | Hits: 100%/3568  
  🟩 nvcc12.0           Pass: 100%/5   | Total:  2h 57m | Avg: 35m 29s | Max: 54m 28s | Hits:  74%/8916  
  🟩 nvcc12.6           Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
  🟩 nvcc12.8           Pass: 100%/39  | Total: 19h 39m | Avg: 30m 15s | Max:  1h 00m | Hits:  79%/69560 
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/2   | Total:  9m 56s | Avg:  4m 58s | Max:  5m 12s | Hits: 100%/3568  
  🟩 nvcc               Pass: 100%/46  | Total:  1d 00h | Avg: 31m 54s | Max:  1h 00m | Hits:  78%/82044 
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  1h 58m | Avg: 29m 35s | Max: 31m 52s | Hits:  76%/7136  
  🟩 Clang15            Pass: 100%/2   | Total: 56m 59s | Avg: 28m 29s | Max: 28m 54s | Hits:  76%/3568  
  🟩 Clang16            Pass: 100%/2   | Total:  1h 02m | Avg: 31m 03s | Max: 31m 27s | Hits:  76%/3568  
  🟩 Clang17            Pass: 100%/2   | Total:  1h 00m | Avg: 30m 16s | Max: 31m 45s | Hits:  76%/3568  
  🟩 Clang18            Pass: 100%/7   | Total:  1h 52m | Avg: 16m 03s | Max: 28m 03s | Hits:  90%/12488 
  🟩 GCC7               Pass: 100%/2   | Total:  1h 03m | Avg: 31m 30s | Max: 31m 32s | Hits:  76%/3570  
  🟩 GCC8               Pass: 100%/1   | Total: 31m 17s | Avg: 31m 17s | Max: 31m 17s | Hits:  76%/1785  
  🟩 GCC9               Pass: 100%/2   | Total:  1h 05m | Avg: 32m 56s | Max: 33m 51s | Hits:  76%/3570  
  🟩 GCC10              Pass: 100%/2   | Total:  1h 05m | Avg: 32m 41s | Max: 32m 44s | Hits:  76%/3570  
  🟩 GCC11              Pass: 100%/2   | Total:  1h 06m | Avg: 33m 21s | Max: 33m 54s | Hits:  76%/3570  
  🟩 GCC12              Pass: 100%/2   | Total:  1h 08m | Avg: 34m 05s | Max: 34m 33s | Hits:  76%/3570  
  🟩 GCC13              Pass: 100%/11  | Total:  3h 57m | Avg: 21m 36s | Max: 34m 57s | Hits:  85%/19635 
  🟩 MSVC14.29          Pass: 100%/2   | Total:  1h 52m | Avg: 56m 00s | Max: 57m 32s | Hits:  64%/3556  
  🟩 MSVC14.42          Pass: 100%/5   | Total:  4h 06m | Avg: 49m 19s | Max:  1h 00m | Hits:  71%/8890  
  🟩 NVHPC25.1          Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
🟩 cxx_family
  🟩 Clang              Pass: 100%/17  | Total:  6h 50m | Avg: 24m 08s | Max: 31m 52s | Hits:  82%/30328 
  🟩 GCC                Pass: 100%/22  | Total:  9h 58m | Avg: 27m 11s | Max: 34m 57s | Hits:  80%/39270 
  🟩 MSVC               Pass: 100%/7   | Total:  5h 58m | Avg: 51m 13s | Max:  1h 00m | Hits:  69%/12446 
  🟩 NVHPC              Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
🟩 gpu
  🟩 h100               Pass: 100%/2   | Total: 29m 26s | Avg: 14m 43s | Max: 19m 20s | Hits:  88%/3570  
  🟩 rtx2080            Pass: 100%/36  | Total: 20h 16m | Avg: 33m 47s | Max:  1h 00m | Hits:  75%/64209 
  🟩 rtx4090            Pass: 100%/10  | Total:  3h 51m | Avg: 23m 08s | Max: 57m 37s | Hits:  89%/17833 
🟩 jobs
  🟩 Build              Pass: 100%/41  | Total: 23h 01m | Avg: 33m 41s | Max:  1h 00m | Hits:  75%/73126 
  🟩 TestCPU            Pass: 100%/3   | Total: 45m 10s | Avg: 15m 03s | Max: 28m 59s | Hits:  99%/5347  
  🟩 TestGPU            Pass: 100%/4   | Total: 50m 39s | Avg: 12m 39s | Max: 14m 18s | Hits:  99%/7139  
🟩 sm
  🟩 90                 Pass: 100%/2   | Total: 29m 26s | Avg: 14m 43s | Max: 19m 20s | Hits:  88%/3570  
  🟩 90;90a             Pass: 100%/2   | Total:  1h 11m | Avg: 35m 41s | Max: 47m 28s | Hits:  70%/3563  
  🟩 100;120            Pass: 100%/2   | Total:  1h 16m | Avg: 38m 07s | Max: 51m 51s | Hits:  70%/3563  
🟩 std
  🟩 17                 Pass: 100%/20  | Total: 11h 50m | Avg: 35m 30s | Max:  1h 00m | Hits:  75%/35671 
  🟩 20                 Pass: 100%/26  | Total: 12h 04m | Avg: 27m 51s | Max: 57m 37s | Hits:  81%/46371

🟩 stdpar: Pass: 100%/4 | Total: 18m 47s | Avg: 4m 41s | Max: 4m 45s

🟩 cpu
  🟩 amd64              Pass: 100%/2   | Total:  9m 28s | Avg:  4m 44s | Max:  4m 45s
  🟩 arm64              Pass: 100%/2   | Total:  9m 19s | Avg:  4m 39s | Max:  4m 43s
🟩 ctk
  🟩 12.6               Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cudacxx
  🟩 nvcc12.6           Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cxx
  🟩 NVHPC25.1          Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cxx_family
  🟩 NVHPC              Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 gpu
  🟩 rtx2080            Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 jobs
  🟩 Build              Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 std
  🟩 17                 Pass: 100%/2   | Total:  9m 26s | Avg:  4m 43s | Max:  4m 43s
  🟩 20                 Pass: 100%/2   | Total:  9m 21s | Avg:  4m 40s | Max:  4m 45s

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 19m 10s | Avg: 9m 35s | Max: 16m 54s | Hits: 96%/320

🟩 cpu
  🟩 amd64              Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 ctk
  🟩 12.8               Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cudacxx
  🟩 nvcc12.8           Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cxx
  🟩 GCC13              Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cxx_family
  🟩 GCC                Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 gpu
  🟩 rtx2080            Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 jobs
  🟩 Build              Pass: 100%/1   | Total:  2m 16s | Avg:  2m 16s | Max:  2m 16s | Hits:  94%/160   
  🟩 Test               Pass: 100%/1   | Total: 16m 54s | Avg: 16m 54s | Max: 16m 54s | Hits:  98%/160

👃 Inspect Changes

Modifications in project?

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

Modifications in project or dependencies?

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

🏃‍ Runner counts (total jobs: 103)

#	Runner
70	`linux-amd64-cpu16`
13	`windows-amd64-cpu16`
6	`linux-arm64-cpu16`
6	`linux-amd64-gpu-rtxa6000-latest-1`
3	`linux-amd64-gpu-h100-latest-1`
3	`linux-amd64-gpu-rtx4090-latest-1`
2	`linux-amd64-gpu-rtx2080-latest-1`

github-actions · 2025-12-12T09:04:46Z

🟨 CI finished in 1d 00h: Pass: 99%/103 | Total: 3d 03h | Avg: 44m 00s | Max: 1h 24m | Hits: 62%/143100

🟥 python: Pass: 0%/1 | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s

🟥 cpu
  🟥 amd64              Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 ctk
  🟥 12.8               Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 cudacxx
  🟥 nvcc12.8           Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 cudacxx_family
  🟥 nvcc               Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 cxx
  🟥 GCC13              Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 cxx_family
  🟥 GCC                Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 gpu
  🟥 rtx2080            Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s
🟥 jobs
  🟥 Test               Pass:   0%/1   | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s

🟩 cub: Pass: 100%/48 | Total: 2d 01h | Avg: 1h 01m | Max: 1h 24m | Hits: 37%/57168

🟩 cpu
  🟩 amd64              Pass: 100%/46  | Total:  1d 23h | Avg:  1h 01m | Max:  1h 24m | Hits:  37%/54720 
  🟩 arm64              Pass: 100%/2   | Total:  2h 15m | Avg:  1h 07m | Max:  1h 09m | Hits:  21%/2448  
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  5h 48m | Avg:  1h 09m | Max:  1h 18m | Hits:  21%/5949  
  🟩 12.6               Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
  🟩 12.8               Pass: 100%/41  | Total:  1d 16h | Avg: 59m 50s | Max:  1h 24m | Hits:  39%/48965 
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/2   | Total:  1h 52m | Avg: 56m 23s | Max: 57m 40s | Hits:  91%/2110  
  🟩 nvcc12.0           Pass: 100%/5   | Total:  5h 48m | Avg:  1h 09m | Max:  1h 18m | Hits:  21%/5949  
  🟩 nvcc12.6           Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
  🟩 nvcc12.8           Pass: 100%/39  | Total:  1d 15h | Avg:  1h 00m | Max:  1h 24m | Hits:  37%/46855 
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/2   | Total:  1h 52m | Avg: 56m 23s | Max: 57m 40s | Hits:  91%/2110  
  🟩 nvcc               Pass: 100%/46  | Total:  1d 23h | Avg:  1h 01m | Max:  1h 24m | Hits:  35%/55058 
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  4h 22m | Avg:  1h 05m | Max:  1h 06m | Hits:  21%/4904  
  🟩 Clang15            Pass: 100%/2   | Total:  2h 16m | Avg:  1h 08m | Max:  1h 09m | Hits:  21%/2448  
  🟩 Clang16            Pass: 100%/2   | Total:  2h 16m | Avg:  1h 08m | Max:  1h 12m | Hits:  21%/2448  
  🟩 Clang17            Pass: 100%/2   | Total:  2h 18m | Avg:  1h 09m | Max:  1h 09m | Hits:  21%/2448  
  🟩 Clang18            Pass: 100%/7   | Total:  6h 23m | Avg: 54m 42s | Max:  1h 14m | Hits:  62%/8230  
  🟩 GCC7               Pass: 100%/2   | Total:  2h 14m | Avg:  1h 07m | Max:  1h 08m | Hits:  21%/2452  
  🟩 GCC8               Pass: 100%/1   | Total:  1h 04m | Avg:  1h 04m | Max:  1h 04m | Hits:  21%/1226  
  🟩 GCC9               Pass: 100%/2   | Total:  2h 24m | Avg:  1h 12m | Max:  1h 14m | Hits:  21%/2452  
  🟩 GCC10              Pass: 100%/2   | Total:  2h 11m | Avg:  1h 05m | Max:  1h 06m | Hits:  21%/2452  
  🟩 GCC11              Pass: 100%/2   | Total:  2h 11m | Avg:  1h 05m | Max:  1h 06m | Hits:  21%/2448  
  🟩 GCC12              Pass: 100%/2   | Total:  2h 12m | Avg:  1h 06m | Max:  1h 07m | Hits:  21%/2448  
  🟩 GCC13              Pass: 100%/12  | Total:  8h 58m | Avg: 44m 51s | Max:  1h 09m | Hits:  60%/14688 
  🟩 MSVC14.29          Pass: 100%/2   | Total:  2h 38m | Avg:  1h 19m | Max:  1h 19m | Hits:  19%/2090  
  🟩 MSVC14.42          Pass: 100%/4   | Total:  5h 09m | Avg:  1h 17m | Max:  1h 24m | Hits:  18%/4180  
  🟩 NVHPC25.1          Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
🟩 cxx_family
  🟩 Clang              Pass: 100%/17  | Total: 17h 36m | Avg:  1h 02m | Max:  1h 14m | Hits:  38%/20478 
  🟩 GCC                Pass: 100%/23  | Total: 21h 17m | Avg: 55m 33s | Max:  1h 14m | Hits:  41%/28166 
  🟩 MSVC               Pass: 100%/6   | Total:  7h 47m | Avg:  1h 17m | Max:  1h 24m | Hits:  18%/6270  
  🟩 NVHPC              Pass: 100%/2   | Total:  2h 41m | Avg:  1h 20m | Max:  1h 24m | Hits:  17%/2254  
🟩 gpu
  🟩 h100               Pass: 100%/3   | Total:  1h 22m | Avg: 27m 39s | Max: 31m 52s | Hits:  73%/3672  
  🟩 rtx2080            Pass: 100%/37  | Total:  1d 18h | Avg:  1h 08m | Max:  1h 24m | Hits:  24%/43704 
  🟩 rtxa6000           Pass: 100%/8   | Total:  5h 35m | Avg: 41m 59s | Max:  1h 14m | Hits:  80%/9792  
🟩 jobs
  🟩 Build              Pass: 100%/40  | Total:  1d 21h | Avg:  1h 07m | Max:  1h 24m | Hits:  24%/47376 
  🟩 DeviceLaunch       Pass: 100%/1   | Total: 33m 11s | Avg: 33m 11s | Max: 33m 11s | Hits:  99%/1224  
  🟩 GraphCapture       Pass: 100%/1   | Total: 26m 44s | Avg: 26m 44s | Max: 26m 44s | Hits:  99%/1224  
  🟩 HostLaunch         Pass: 100%/3   | Total:  1h 34m | Avg: 31m 23s | Max: 37m 25s | Hits:  99%/3672  
  🟩 TestGPU            Pass: 100%/3   | Total:  1h 30m | Avg: 30m 19s | Max: 33m 24s | Hits:  99%/3672  
🟩 sm
  🟩 90                 Pass: 100%/3   | Total:  1h 22m | Avg: 27m 39s | Max: 31m 52s | Hits:  73%/3672  
  🟩 90;90a             Pass: 100%/2   | Total:  1h 59m | Avg: 59m 38s | Max:  1h 09m | Hits:  19%/2269  
  🟩 100;120            Pass: 100%/2   | Total:  2h 21m | Avg:  1h 10m | Max:  1h 13m | Hits:  20%/2269  
🟩 std
  🟩 17                 Pass: 100%/20  | Total: 23h 12m | Avg:  1h 09m | Max:  1h 24m | Hits:  24%/23693 
  🟩 20                 Pass: 100%/28  | Total:  1d 02h | Avg: 56m 07s | Max:  1h 24m | Hits:  46%/33475

🟩 thrust: Pass: 100%/48 | Total: 1d 00h | Avg: 30m 46s | Max: 1h 00m | Hits: 79%/85612

🟩 cmake_options
  🟩 -DTHRUST_DISPATCH_TYPE=Force32bit Pass: 100%/2   | Total: 42m 44s | Avg: 21m 22s | Max: 28m 26s | Hits:  88%/3570  
🟩 cpu
  🟩 amd64              Pass: 100%/46  | Total: 23h 42m | Avg: 30m 55s | Max:  1h 00m | Hits:  79%/82043 
  🟩 arm64              Pass: 100%/2   | Total: 54m 34s | Avg: 27m 17s | Max: 28m 55s | Hits:  76%/3569  
🟩 ctk
  🟩 12.0               Pass: 100%/5   | Total:  2h 57m | Avg: 35m 29s | Max: 54m 28s | Hits:  74%/8916  
  🟩 12.6               Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
  🟩 12.8               Pass: 100%/41  | Total: 19h 49m | Avg: 29m 01s | Max:  1h 00m | Hits:  80%/73128 
🟩 cudacxx
  🟩 ClangCUDA18        Pass: 100%/2   | Total:  9m 56s | Avg:  4m 58s | Max:  5m 12s | Hits: 100%/3568  
  🟩 nvcc12.0           Pass: 100%/5   | Total:  2h 57m | Avg: 35m 29s | Max: 54m 28s | Hits:  74%/8916  
  🟩 nvcc12.6           Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
  🟩 nvcc12.8           Pass: 100%/39  | Total: 19h 39m | Avg: 30m 15s | Max:  1h 00m | Hits:  79%/69560 
🟩 cudacxx_family
  🟩 ClangCUDA          Pass: 100%/2   | Total:  9m 56s | Avg:  4m 58s | Max:  5m 12s | Hits: 100%/3568  
  🟩 nvcc               Pass: 100%/46  | Total:  1d 00h | Avg: 31m 54s | Max:  1h 00m | Hits:  78%/82044 
🟩 cxx
  🟩 Clang14            Pass: 100%/4   | Total:  1h 58m | Avg: 29m 35s | Max: 31m 52s | Hits:  76%/7136  
  🟩 Clang15            Pass: 100%/2   | Total: 56m 59s | Avg: 28m 29s | Max: 28m 54s | Hits:  76%/3568  
  🟩 Clang16            Pass: 100%/2   | Total:  1h 02m | Avg: 31m 03s | Max: 31m 27s | Hits:  76%/3568  
  🟩 Clang17            Pass: 100%/2   | Total:  1h 00m | Avg: 30m 16s | Max: 31m 45s | Hits:  76%/3568  
  🟩 Clang18            Pass: 100%/7   | Total:  1h 52m | Avg: 16m 03s | Max: 28m 03s | Hits:  90%/12488 
  🟩 GCC7               Pass: 100%/2   | Total:  1h 03m | Avg: 31m 30s | Max: 31m 32s | Hits:  76%/3570  
  🟩 GCC8               Pass: 100%/1   | Total: 31m 17s | Avg: 31m 17s | Max: 31m 17s | Hits:  76%/1785  
  🟩 GCC9               Pass: 100%/2   | Total:  1h 05m | Avg: 32m 56s | Max: 33m 51s | Hits:  76%/3570  
  🟩 GCC10              Pass: 100%/2   | Total:  1h 05m | Avg: 32m 41s | Max: 32m 44s | Hits:  76%/3570  
  🟩 GCC11              Pass: 100%/2   | Total:  1h 06m | Avg: 33m 21s | Max: 33m 54s | Hits:  76%/3570  
  🟩 GCC12              Pass: 100%/2   | Total:  1h 08m | Avg: 34m 05s | Max: 34m 33s | Hits:  76%/3570  
  🟩 GCC13              Pass: 100%/11  | Total:  3h 57m | Avg: 21m 36s | Max: 34m 57s | Hits:  85%/19635 
  🟩 MSVC14.29          Pass: 100%/2   | Total:  1h 52m | Avg: 56m 00s | Max: 57m 32s | Hits:  64%/3556  
  🟩 MSVC14.42          Pass: 100%/5   | Total:  4h 06m | Avg: 49m 19s | Max:  1h 00m | Hits:  71%/8890  
  🟩 NVHPC25.1          Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
🟩 cxx_family
  🟩 Clang              Pass: 100%/17  | Total:  6h 50m | Avg: 24m 08s | Max: 31m 52s | Hits:  82%/30328 
  🟩 GCC                Pass: 100%/22  | Total:  9h 58m | Avg: 27m 11s | Max: 34m 57s | Hits:  80%/39270 
  🟩 MSVC               Pass: 100%/7   | Total:  5h 58m | Avg: 51m 13s | Max:  1h 00m | Hits:  69%/12446 
  🟩 NVHPC              Pass: 100%/2   | Total:  1h 50m | Avg: 55m 08s | Max: 58m 44s | Hits:  64%/3568  
🟩 gpu
  🟩 h100               Pass: 100%/2   | Total: 29m 26s | Avg: 14m 43s | Max: 19m 20s | Hits:  88%/3570  
  🟩 rtx2080            Pass: 100%/36  | Total: 20h 16m | Avg: 33m 47s | Max:  1h 00m | Hits:  75%/64209 
  🟩 rtx4090            Pass: 100%/10  | Total:  3h 51m | Avg: 23m 08s | Max: 57m 37s | Hits:  89%/17833 
🟩 jobs
  🟩 Build              Pass: 100%/41  | Total: 23h 01m | Avg: 33m 41s | Max:  1h 00m | Hits:  75%/73126 
  🟩 TestCPU            Pass: 100%/3   | Total: 45m 10s | Avg: 15m 03s | Max: 28m 59s | Hits:  99%/5347  
  🟩 TestGPU            Pass: 100%/4   | Total: 50m 39s | Avg: 12m 39s | Max: 14m 18s | Hits:  99%/7139  
🟩 sm
  🟩 90                 Pass: 100%/2   | Total: 29m 26s | Avg: 14m 43s | Max: 19m 20s | Hits:  88%/3570  
  🟩 90;90a             Pass: 100%/2   | Total:  1h 11m | Avg: 35m 41s | Max: 47m 28s | Hits:  70%/3563  
  🟩 100;120            Pass: 100%/2   | Total:  1h 16m | Avg: 38m 07s | Max: 51m 51s | Hits:  70%/3563  
🟩 std
  🟩 17                 Pass: 100%/20  | Total: 11h 50m | Avg: 35m 30s | Max:  1h 00m | Hits:  75%/35671 
  🟩 20                 Pass: 100%/26  | Total: 12h 04m | Avg: 27m 51s | Max: 57m 37s | Hits:  81%/46371

🟩 stdpar: Pass: 100%/4 | Total: 18m 47s | Avg: 4m 41s | Max: 4m 45s

🟩 cpu
  🟩 amd64              Pass: 100%/2   | Total:  9m 28s | Avg:  4m 44s | Max:  4m 45s
  🟩 arm64              Pass: 100%/2   | Total:  9m 19s | Avg:  4m 39s | Max:  4m 43s
🟩 ctk
  🟩 12.6               Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cudacxx
  🟩 nvcc12.6           Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cxx
  🟩 NVHPC25.1          Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 cxx_family
  🟩 NVHPC              Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 gpu
  🟩 rtx2080            Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 jobs
  🟩 Build              Pass: 100%/4   | Total: 18m 47s | Avg:  4m 41s | Max:  4m 45s
🟩 std
  🟩 17                 Pass: 100%/2   | Total:  9m 26s | Avg:  4m 43s | Max:  4m 43s
  🟩 20                 Pass: 100%/2   | Total:  9m 21s | Avg:  4m 40s | Max:  4m 45s

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 19m 10s | Avg: 9m 35s | Max: 16m 54s | Hits: 96%/320

🟩 cpu
  🟩 amd64              Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 ctk
  🟩 12.8               Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cudacxx
  🟩 nvcc12.8           Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cudacxx_family
  🟩 nvcc               Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cxx
  🟩 GCC13              Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 cxx_family
  🟩 GCC                Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 gpu
  🟩 rtx2080            Pass: 100%/2   | Total: 19m 10s | Avg:  9m 35s | Max: 16m 54s | Hits:  96%/320   
🟩 jobs
  🟩 Build              Pass: 100%/1   | Total:  2m 16s | Avg:  2m 16s | Max:  2m 16s | Hits:  94%/160   
  🟩 Test               Pass: 100%/1   | Total: 16m 54s | Avg: 16m 54s | Max: 16m 54s | Hits:  98%/160

👃 Inspect Changes

Modifications in project?

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

Modifications in project or dependencies?

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

🏃‍ Runner counts (total jobs: 103)

#	Runner
70	`linux-amd64-cpu16`
13	`windows-amd64-cpu16`
6	`linux-arm64-cpu16`
6	`linux-amd64-gpu-rtxa6000-latest-1`
3	`linux-amd64-gpu-h100-latest-1`
3	`linux-amd64-gpu-rtx4090-latest-1`
2	`linux-amd64-gpu-rtx2080-latest-1`

Properly specialize cub functions for __nv_bfloat16 (NVIDIA#6931)

cf18326

(cherry picked from commit e1f5bcc)

miscco requested a review from a team as a code owner December 11, 2025 08:42

miscco requested a review from elstehle December 11, 2025 08:42

github-project-automation bot added this to CCCL Dec 11, 2025

github-project-automation bot moved this to Todo in CCCL Dec 11, 2025

cccl-authenticator-app bot moved this from Todo to In Review in CCCL Dec 11, 2025

elstehle approved these changes Dec 11, 2025

View reviewed changes

bernhardmgruber approved these changes Dec 11, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[Backport 3.0] Properly specialize cub functions for `__nv_bfloat16` (#6931) #6946

[Backport 3.0] Properly specialize cub functions for `__nv_bfloat16` (#6931) #6946

Uh oh!

miscco commented Dec 11, 2025

Uh oh!

github-actions bot commented Dec 11, 2025

🟥 python: Pass: 0%/1 | Total: 1h 01m | Avg: 1h 01m | Max: 1h 01m

🟩 cub: Pass: 100%/48 | Total: 2d 01h | Avg: 1h 01m | Max: 1h 24m | Hits: 37%/57168

🟩 thrust: Pass: 100%/48 | Total: 1d 00h | Avg: 30m 46s | Max: 1h 00m | Hits: 79%/85612

🟩 stdpar: Pass: 100%/4 | Total: 18m 47s | Avg: 4m 41s | Max: 4m 45s

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 19m 10s | Avg: 9m 35s | Max: 16m 54s | Hits: 96%/320

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 103)

Uh oh!

github-actions bot commented Dec 12, 2025

🟥 python: Pass: 0%/1 | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s

🟩 cub: Pass: 100%/48 | Total: 2d 01h | Avg: 1h 01m | Max: 1h 24m | Hits: 37%/57168

🟩 thrust: Pass: 100%/48 | Total: 1d 00h | Avg: 30m 46s | Max: 1h 00m | Hits: 79%/85612

🟩 stdpar: Pass: 100%/4 | Total: 18m 47s | Avg: 4m 41s | Max: 4m 45s

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 19m 10s | Avg: 9m 35s | Max: 16m 54s | Hits: 96%/320

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 103)

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

[Backport 3.0] Properly specialize cub functions for __nv_bfloat16 (#6931) #6946

Are you sure you want to change the base?

[Backport 3.0] Properly specialize cub functions for __nv_bfloat16 (#6931) #6946

Uh oh!

Conversation

miscco commented Dec 11, 2025

Description

Checklist

Uh oh!

github-actions bot commented Dec 11, 2025

🟥 python: Pass: 0%/1 | Total: 1h 01m | Avg: 1h 01m | Max: 1h 01m

🟩 cub: Pass: 100%/48 | Total: 2d 01h | Avg: 1h 01m | Max: 1h 24m | Hits: 37%/57168

🟩 thrust: Pass: 100%/48 | Total: 1d 00h | Avg: 30m 46s | Max: 1h 00m | Hits: 79%/85612

🟩 stdpar: Pass: 100%/4 | Total: 18m 47s | Avg: 4m 41s | Max: 4m 45s

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 19m 10s | Avg: 9m 35s | Max: 16m 54s | Hits: 96%/320

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 103)

Uh oh!

github-actions bot commented Dec 12, 2025

🟥 python: Pass: 0%/1 | Total: 52m 35s | Avg: 52m 35s | Max: 52m 35s

🟩 cub: Pass: 100%/48 | Total: 2d 01h | Avg: 1h 01m | Max: 1h 24m | Hits: 37%/57168

🟩 thrust: Pass: 100%/48 | Total: 1d 00h | Avg: 30m 46s | Max: 1h 00m | Hits: 79%/85612

🟩 stdpar: Pass: 100%/4 | Total: 18m 47s | Avg: 4m 41s | Max: 4m 45s

🟩 cccl_c_parallel: Pass: 100%/2 | Total: 19m 10s | Avg: 9m 35s | Max: 16m 54s | Hits: 96%/320

👃 Inspect Changes

Modifications in project?

Modifications in project or dependencies?

🏃‍ Runner counts (total jobs: 103)

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

[Backport 3.0] Properly specialize cub functions for `__nv_bfloat16` (#6931) #6946

[Backport 3.0] Properly specialize cub functions for `__nv_bfloat16` (#6931) #6946