I agree there should be a finalizer to reclaim GPU memory - you can see an example [here](https://github.com/unicredit/neo/blob/master/neo/cudadense.nim#L53)
- Success - calling custom CUDA kernels from Nim mratsim
- Re: Success - calling custom CUDA kernels from Nim mratsim
- Re: Success - calling custom CUDA kernels from Nim Udiknedormin
- Re: Success - calling custom CUDA kernels from N... mratsim
- Re: Success - calling custom CUDA kernels fr... andrea
- Re: Success - calling custom CUDA kerne... Udiknedormin
- Re: Success - calling custom CUDA k... andrea
- Re: Success - calling custom CU... mratsim
- Re: Success - calling custom CU... Udiknedormin
- Re: Success - calling custom CU... mratsim
- Re: Success - calling custom CU... Udiknedormin
