I had learned polars before but didn't go too far in using it because I was already running RAPIDS cuDF which used the GPU. now Polars uses the GPU so things are converging. The lazy method should put polars+GPU over RAPIDS cudf now. I think it really matters when you're running massive datasets
That's what I'm thinking. I'm honestly not super hip on all the different dataframe solutions out there, and how they compare with one another. I may test a wide array of them in a standardized way to see what happens.
I had learned polars before but didn't go too far in using it because I was already running RAPIDS cuDF which used the GPU. now Polars uses the GPU so things are converging. The lazy method should put polars+GPU over RAPIDS cudf now. I think it really matters when you're running massive datasets
That's what I'm thinking. I'm honestly not super hip on all the different dataframe solutions out there, and how they compare with one another. I may test a wide array of them in a standardized way to see what happens.