[reilly3000]

Check out https://www.omnisci.com/learn/resources/gpu-database

In summary, you get snappy, interactive query speeds on large data sets. I've ran that locally and the results are pretty amazing compared to Postgres or even Tableau in-memory.

I'm personally more excited about GPUs in stream processing; its just quite a natural fit: https://github.com/rapidsai/cudf

[kichik]

If you're interested in stream processing, check out FASTDATA.io PlasmaENGINE. We do both stream and batch processing with Apache Spark on the GPU.

https://fastdata.io/plasma-engine/

* It's not open-source and I work there.

[arnon]

Hi Kichik :)

[felipe_aramburu]

Blazingsql is built on top of CUDF. We are contributors to rapidsai

[throwaway082729]

Isn't the speed bottlenecked by the storage speed. Is the data fully loaded into memory first?

[tmostak]

OmniSci transparently caches data across the memory of the CPUs and GPUs on a server, so after the initial read, it is likely that the data for subsequent queries will be in memory.

We've also optimized our storage formats and multithreaded our disk reads, such that we can easily hit many gigabytes per second on flash storage. Plus, new persistent memory technologies like Intel Optane will enable even more instant reads from "cold" storage.