Hi Surya,

I found tests and benchmarks that work best, you can find from the
following links.

https://github.com/chenqin/arrow/blob/master/cpp/src/arrow/gpu/cuda_benchmark.cc
https://github.com/chenqin/arrow/blob/master/cpp/src/arrow/gpu/cuda_test.cc

One of the difficulties is to come up with the right workload that fits the
computational heavy nature of the GPU. If you are looking to run ML in GPU,
you might directly link.  https://pytorch.org/cppdocs/installing.html

Thanks,
Chen




On Fri, Apr 7, 2023 at 8:16 AM Surya Kiran Gullapalli <
[email protected]> wrote:

> Hello all,
> For Cuda support, I could not find any examples on how to use cuda for
> arrow computations. Any pointers in this direction would be greatly
> appreciated.
> Also any pointers on how to use multiple gpus with apache arrow would be
> helpful.
>
> Thanks,
> Surya
>

Reply via email to