* Noah Misch (noah@leadboat.com) wrote:
> In particular, implementing a GPU-based strcoll() for bttextcmp
> sounds like quite a project in its own right.
It also wouldn't likely be helpful... To be able to use a GPU
effectively, last I looked, you need to be able to move a large
chunk of data to the GPU's memory, operate on it, and then bulk
move the results back because the cost of moving data between main
memory and GPU memory is very high.
> Those are matters we would eventually need to address as we parallelize more
> things, so I regard confronting them as an advantage. Among other benefits,
> this project is a vehicle for emplacing some infrastructure without inviting
> the full complexity entailed by loftier goals.
Agreed.
Thanks,
Stephen