3 ms·
There is dask cudf which gets a lot of the way there. https://docs.rapids.ai/api/dask-cudf/stable/ https://docs.rapids.ai/api/dask-cudf/stable/
by hack_ml 2y ago
There is dask cudf which gets a lot of the way there.
https://docs.rapids.ai/api/dask-cudf/stable/ https://docs.rapids.ai/api/dask-cudf/stable/
- 3abiton 2y agoThis is huge, this was my only gripe with cudf!
- xs83 2y agoI did a conversion of 500GB of data using dask_cudf on a GTX 1060 with 6GB of VRAM and was able to do it faster than a 20 node m3.xlarge Cluster. What you can do on even consumer GPU's is mind blowing.
- ashvardanian 2y agoLast time I’ve used it, Dask was a lot worse than simple manual batching.
- iamcreasy 2y agoHow does it perform when it comes to plotting these large data points? Can I use matplotlib?