3 ms·
SQL is somewhat parallel. All row-level computations are independent and a WHERE maps nicely to a reduce(). I'm not sure if these aren't limited by I/O, though.
by matt4077 11y ago
SQL is somewhat parallel. All row-level computations are independent and a WHERE maps nicely to a reduce(). I'm not sure if these aren't limited by I/O, though.
Here's a recent article by NVIDIA on using GPUs for graph computation which is somewhat related: https://devblogs.nvidia.com/parallelforall/gpus-graph-predictive-analytics/ https://devblogs.nvidia.com/parallelforall/gpus-graph-predic...
- jandrese 11y agoI think it will be highly dependent on how you structure your queries. Some will be absolutely I/O limited while others will be barely faster than on the CPU. Piping the data out to the GPU is definitely going to be an issue too unless your database is small enough to fit in GPU memory (in which case the whole approach is probably overkill anyway).