Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-05-16 Thread Tom Lane
I wrote: > ... Where we are losing is mostly on the actual manipulation > of the bitmaps (particularly hash_seq_search which is done in > tbm_begin_iterate; and it looks like memory allocation for the bitmap > hashtables is nontrivial too). I had already had a TODO item to look > into speeding up

Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-05-05 Thread Sergey E. Koposov
On Thu, 5 May 2005, Tom Lane wrote: > "Sergey E. Koposov" <[EMAIL PROTECTED]> writes: > > And I coadded the "flat profiles" of first two (index scan) queries and > > compared it with the flat profile of bitmap scan: > > Thanks, I had been thinking of doing that same calculation but hadn't > got

Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-05-05 Thread Tom Lane
"Sergey E. Koposov" <[EMAIL PROTECTED]> writes: > And I coadded the "flat profiles" of first two (index scan) queries and > compared it with the flat profile of bitmap scan: Thanks, I had been thinking of doing that same calculation but hadn't got round to it yet. It looks like the bitmap case

Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-05-05 Thread Sergey E. Koposov
On Wed, 4 May 2005, Tom Lane wrote: > > I figured out part of the problem: I had made nodeBitmapIndexscan > re-open the index on each call, thinking that that would save amrescan > calls. But an amrescan is a whole lot cheaper than index open/close, > so that was a bad tradeoff. > > This seems

Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-05-04 Thread Tom Lane
"Sergey E. Koposov" <[EMAIL PROTECTED]> writes: > On Fri, 29 Apr 2005, Tom Lane wrote: >> The latter is (or should be) doing slightly *less* work, so why is it >> taking almost twice as much time? Can you get gprof profiles of the >> two cases? > I've got them. I attached two files with a little

Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-04-29 Thread Sergey E. Koposov
On Fri, 29 Apr 2005, Tom Lane wrote: > >-> Index Scan using ipix_idx on q3c (cost=0.01..9686.37 rows=35 > > width=48) (actual time=0.006..0.006 rows=0 loops=300) > > Index Cond: ((q3c.ipix >= ("outer".ipix - 1000)) AND (q3c.ipix <= > > ("outer".ipix - 993))) > > >

Re: [HACKERS] performance of bitmap scans in nested loop joins

2005-04-29 Thread Tom Lane
"Sergey E. Koposov" <[EMAIL PROTECTED]> writes: > I'd like to report about suprising (for me) results of performance testing of > bitmap indexes in nested loop join plans. I'm surprised too. There's something weird in the indexscans themselves: >-> Index Scan using ipix_idx on q3c (cost=

[HACKERS] performance of bitmap scans in nested loop joins

2005-04-29 Thread Sergey E. Koposov
Hello All, I'd like to report about suprising (for me) results of performance testing of bitmap indexes in nested loop join plans. I have the table q3c test=# \d q3c Table "public.q3c" Column | Type | Modifiers ++--- ipix | bigint | errbox | box| ra