[Tile-serving] [openstreetmap/osm2pgsql] Possible improvements for new ram middle (#1466)

[mmd]

Regarding "Reading node location back from node_locations_t": Some SIMD-based LEB128 decoders claim to provide faster speeds than the current byte-by-byte approach used in protozero. It might be fast enough to avoid the additional caching of decoded blocks. Lots of research on this topic by https://lemire.me/blog. 
Potentially interesting Rust implementation: https://github.com/as-com/varint-simd 

Besides, the current approach already turned out to be about 10% faster than the old one in my tests. My impression was that fetching node locations isn't one of the major bottleneck in the overall process anymore. This might be different for other test cases, though.

-- 
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/openstreetmap/osm2pgsql/issues/1466#issuecomment-831145321
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tile-serving/attachments/20210503/ac4e809f/attachment.htm>