speed up single bio_vec allocation
Quoted 10 lines
Chen, Kenneth wrote on Wednesday, December 06, 2006 10:20 AM
Quoted 6 lines
Jens Axboe wrote on Wednesday, December 06, 2006 2:09 AM This is what I had in mind, in case it wasn't completely clear. Not tested, other than it compiles. Basically it eliminates the small
bio_vecpool, and grows the bio by 16-bytes on 64-bit archs, or by 12-bytes on 32-bit archs instead and uses the room at the end for thebio_vecstructure.Yeah, I had a very similar patch queued internally for the large benchmark measurement. I will post the result as soon as I get it.
Jens, this improves 0.25% on our db transaction processing benchmark setup. The patch tested is (on top of 2.6.19):
http://marc.theaimsgroup.com/?l=linux-kernel&m=116539972229021&w=2
- Ken
Quoted 8 lines · Kenneth W
Quoted 8 lines
Chen, Kenneth wrote on Wednesday, December 06, 2006 10:20 AM
Quoted 6 lines
Jens Axboe wrote on Wednesday, December 06, 2006 2:09 AM This is what I had in mind, in case it wasn't completely clear. Not tested, other than it compiles. Basically it eliminates the small
bio_vecpool, and grows the bio by 16-bytes on 64-bit archs, or by 12-bytes on 32-bit archs instead and uses the room at the end for thebio_vecstructure.
Quoted 7 lines
Quoted 2 lines
Yeah, I had a very similar patch queued internally for the large benchmark measurement. I will post the result as soon as I get it.
Jens, this improves 0.25% on our db transaction processing benchmark setup. The patch tested is (on top of 2.6.19):
http://marc.theaimsgroup.com/?l=linux-kernel&m=116539972229021&w=2
Ok, well it's not much but if it's significant it's not too bad either :-)
Some tests I ran locally showed it being _slower_, which is a little odd. They were basically hammering requests through the block layer with a null end.
Signature · 1 line
Jens Axboe