http://www.perlmonks.org?node_id=291111


in reply to sloooowwwwww code!

Any help optimizing this slow code would be great.

Good advice on profiling aside, I would start with the query and database schema. A lot of time can get chewed up by an inefficient query, or one that could be sped up by adding indexes.

Look at the query

select text_id,gene_name,protein5,protein3 from protein where org_id=$orgid order by protein5 desc"
and ask two questions:
  1. Is there an index on org_id?
  2. Is it really necessary to have the database sort the results?
Unless org_id is indexed, the database must make a linear scan over the protein table. If the table is big, the scan can be slow. If the field is indexed, the retrieval will usually be considerably faster (except in cases where you're going to match most of the records anyway.)

I can't tell from your code whether it matters whether the query results are sorted or not. If it doesn't matter, drop the order by clause.

The non-SQL thing that jumped out was the redundant trig calculations. You're doing each sine and cosine calculation twice, where once would be sufficient.

Then profile.