Issue with Trie_Cuda example in Release configuration

[greenbrettmichael]

The CXX compiler identification is MSVC 19.32.31329.0 The CUDA compiler identification is NVIDIA 11.6.124 GPU Geforce RTX 2080

Debug output is

books/2600-0.txt
books/2701-0.txt
books/35-0.txt
books/84-0.txt
books/8800.txt
books/pg1727.txt
books/pg55.txt
books/pg6130.txt
books/pg996.txt
books/1342-0.txt
Assembled 98154 nodes on 1x1 cpu threads in 967ms.
Assembled 98154 nodes on 1x1 cpu threads in 716ms.
Assembled 98154 nodes on 1x12 cpu threads in 140ms.
Assembled 98154 nodes on 1x12 cpu threads in 141ms.
Assembled 98154 nodes on 40x1024 cuda threads in 4ms.
Assembled 98154 nodes on 40x1024 cuda threads in 5ms.

Release output is

books/2600-0.txt
books/2701-0.txt
books/35-0.txt
books/84-0.txt
books/8800.txt
books/pg1727.txt
books/pg55.txt
books/pg6130.txt
books/pg996.txt
books/1342-0.txt
Assembled 98154 nodes on 1x1 cpu threads in 180ms.
Assembled 98154 nodes on 1x1 cpu threads in 146ms.
Assembled 98154 nodes on 1x12 cpu threads in 26ms.
Assembled 98154 nodes on 1x12 cpu threads in 27ms.
Assembled 1 nodes on 0x1024 cuda threads in 0ms.
Assembled 1 nodes on 0x1024 cuda threads in 0ms.

I do not know what other information is relevant.

[wmaxey]

Thanks for the report, I'll try and repro this.