Hello,
After some months of usage of BFQ kernel scheduler I can only talk about its effective improvement for HDD I/O.
Since this evidence, I was asking myself why BFQ it is not used as default scheduler, replacing the standard CFQ.
Cheers,
On Monday 2017-01-09 15:22, Marco Calistri wrote:
Hello,
After some months of usage of BFQ kernel scheduler I can only talk about its effective improvement for HDD I/O.
Since this evidence, I was asking myself why BFQ it is not used as default scheduler, replacing the standard CFQ.
Because it has not been merged into Linux last time I checked.
On 9 January 2017 at 15:49, Jan Engelhardt jengelh@inai.de wrote:
On Monday 2017-01-09 15:22, Marco Calistri wrote:
Hello,
After some months of usage of BFQ kernel scheduler I can only talk about its effective improvement for HDD I/O.
Since this evidence, I was asking myself why BFQ it is not used as default scheduler, replacing the standard CFQ.
Because it has not been merged into Linux last time I checked.
Yeah I cant find any evidence of that either
And looking at BFQ's own benchmarks I wouldn't say the picture is clear that BFQ is better than CFQ in all circumstances
http://algo.ing.unimo.it/people/paolo/disk_sched/results.php http://algo.ing.unimo.it/people/paolo/disk_sched/old-results/4.4.0-v7r11/
more throughput on an Hitachi & SEAGATE HDD, less on a Plextor SSD, no different on a microSDHC card
I like what they're doing and hope they get it into the kernel one day soon, but until they do I do not see a compelling case for changing openSUSE's default to an unsupported out of kernel IO scheduler
-
Jan Engelhardt -
Marco Calistri -
Richard Brown