Muhammad Awad

Okay, I'll add a test for that (this weekend hopefully).

Hi @mfbalin, The current `insert` and `find` implementations and tests of the three bucketed hash tables (cuckoo, power-of-two, and iceberg) assume that the keys are unique. If you insert duplicate...

If we have already inserted the keys with duplicates and then tried to perform the reduction, we will be using something like `equal_range` to uniquify the keys. It won't be...

Yeah, I think if you are not wasting that much space on duplicates, then storing them should be okay. Any optimization will depend on the workload :) The performance of...

Hi @xiaoleitongxue, The `nvcc` compiler crashing would be an `nvcc` bug. I ran into a similar bug before, but they are tough to track down, so I'd recommend submitting a...

That's great! Thanks for the update. Do you mind sharing what CUDA/NVCC version worked for you? I tried CUDA 11.8 on Linux, and the code compiled successfully, so it's probably...

Hi @xiaoniandada123 , which CUDA version and operating system you are using?