Workarounds for CUDA 12.1 and 12.2 #1206

Job	Run time
perftests (gcc-8, release)	3m 1s
perftests (gcc-9, release)	3m 35s
perftests (gcc-10, release)	2m 24s
perftests (gcc-11, release)	3m 0s
perftests (gcc-12, debug)	4m 24s
perftests (gcc-12, release)	2m 26s
perftests (gcc-13, debug)	2m 56s
perftests (gcc-13, release)	2m 17s
perftests (clang-11, debug)	2m 54s
perftests (clang-11, release)	3m 17s
perftests (clang-12, debug)	2m 29s
perftests (clang-12, release)	2m 27s
perftests (clang-13, release)	3m 51s
perftests (clang-14, release)	3m 31s
perftests (clang-15, debug)	2m 26s
perftests (clang-15, release)	2m 54s
perftests (clang-16, debug)	2m 49s
perftests (clang-16, release)	2m 30s
perftests (clang-14-cuda-11, debug)	9m 54s
perftests (clang-14-cuda-11, release)	9m 48s
perftests (gcc-10-cuda-11.8, debug)	10m 22s
perftests (gcc-10-cuda-11.8, release)	9m 57s
perftests (gcc-11-cuda-12.0, debug)	7m 55s
perftests (gcc-11-cuda-12.0, release)	8m 32s
perftests (gcc-12-cuda-12.1, debug)	9m 13s
perftests (gcc-12-cuda-12.1, release)	11m 4s
perftests (gcc-12-cuda-12.2, debug)	13m 18s
perftests (gcc-12-cuda-12.2, release)	6m 58s
perftests (base-hip, debug)	9m 53s
perftests (base-hip, release)	8m 21s
perftests (gcc-10-hpx, release)	5m 13s
perftests (nvhpc-23.3, debug)	15m 21s
perftests (nvhpc-23.3, release)	24m 2s
perftests (nvhpc-23.7, debug)	17m 50s
perftests (nvhpc-23.7, release)	19m 51s
	4h 10m 43s

Provide feedback