Muhammad Awad

@hao123mh sorry for responding late. You are correct (and nice catch!). At one point the scenario you mentioned should happen. I did a test to see if it actually happens...

Hi @HakubunLuo , thanks for reporting this issue. Could you please share the complete log you get from both the CMake configure and build commands? It looks like the `CUDA_ARCHS`...

You should not pass the hash tables by reference or as pointers to kernels. You should pass them by value to kernels. It looks like you are dealing with a...

Also if every thread is trying to insert a key, you will need to serialize them within a tile. See how we do it here: https://github.com/owensgroup/BGHT/blob/140b80f859efe08c1f7bcf95b459bb56500703bd/include/detail/kernels.cuh#L26-L64

If N is not multiple of the bucket size you will into issues. The insert function expects `bucket_size` threads calling it. This if statement is problematic: ```cuda if (idx <...

Great! Yes, you will need to follow the same steps for finds. Here is an example: https://github.com/owensgroup/BGHT/blob/140b80f859efe08c1f7bcf95b459bb56500703bd/include/detail/kernels.cuh#L67-L109

You need to follow the insertion/find code. Again, this if statement is problematic: ```cuda if (idx < N) { } ``` and if values of `lookup_idx` are different per thread...

I understand the keys you are using are different. You need to follow the same strategy you followed for insertion which is similar to find as well. The two things...

@DaemonSnake I finished it sometime ago but haven't pushed it yet. I'll add it this weekend.

Hi, I've been refactoring the code so I haven't added these yet, but it should be simple to add for the current supported operations (insert, query, and delete). Which operations...