Does anyone have any metrics on how fast tsearch2 actually is?

I tried it on a synthetic dataset of a million documents of a hundred
words each and while insertions were impressively fast I gave up on
the search after 10 minutes.

Broken? Unusable slow? This was on the last 7.4 release candidate.

I just created a 1.1million row dataset by copying one of our 30000 row production tables and just taking out the txtidx column. Then I inserted it into itself until it had 1.1 million rows.

Then I created the GiST index - THAT took forever - seriously like 20 mins or half an hour or something.

Now, to find a word:

select * from tsearchtest where ftiidx ## 'curry';
Time: 9760.75 ms

The AND of two words:
Time: 103.61 ms

The AND of three words:
select * from tsearchtest where ftiidx ## 'curry&green&thai';
Time: 61.86 ms

And now a one word query now that buffers are cached:
select * from tsearchtest where ftiidx ## 'curry';
Time: 444.89 ms

So, I have no idea why you think it's slow? Perhaps you forgot the 'create index using gist' step?

Also, if you use the NOT (!) operand, you can get yourself into a really slow situation.


---------------------------(end of broadcast)--------------------------- TIP 3: if posting/reading through Usenet, please send an appropriate subscribe-nomail command to [EMAIL PROTECTED] so that your message can get through to the mailing list cleanly

Reply via email to