Re: [patch] Revert "block: remove artifical max_hw_sectors cap"

[Christoph Hellwig <hch@infradead.org>]

Hi Jeff,

thanks for the detailed numbers!

The bigger I/O size makes a drastic impact for Linux software RAID
setups, for which this was a driver.  For the RAID5/6 over SATA disks
setups that I was benchmarking this it gives between 20 and 40% better
sequential read and write numbers.

Besides those I've tested it on various SSDs where it didn't make any
difference, e.g. on the laptop with a Samsung 840 Evo I'm currently
travelling with and your fio script.

First line max_sectors_kb = 512, second line
max_sectors_kb = max_hw_sectors_kb (32767)

Run status group 0 (all jobs):
  WRITE: io=1024.0MB, aggrb=471270KB/s, minb=471270KB/s, maxb=471270KB/s, mint=2225msec, maxt=2225msec
  WRITE: io=1024.0MB, aggrb=475976KB/s, minb=475976KB/s, maxb=475976KB/s, mint=2203msec, maxt=2203msec

Run status group 1 (all jobs):
  WRITE: io=1024.0MB, aggrb=477276KB/s, minb=477276KB/s, maxb=477276KB/s, mint=2197msec, maxt=2197msec
  WRITE: io=1024.0MB, aggrb=479677KB/s, minb=479677KB/s, maxb=479677KB/s, mint=2186msec, maxt=2186msec

Run status group 2 (all jobs):
  WRITE: io=1024.0MB, aggrb=488618KB/s, minb=488618KB/s, maxb=488618KB/s, mint=2146msec, maxt=2146msec
  WRITE: io=1024.0MB, aggrb=489302KB/s, minb=489302KB/s, maxb=489302KB/s, mint=2143msec, maxt=2143msec