Note on the phoronix numbers
Added 2023-11-04 15:26:48 +0000 UTCPhoronix recently posted some bcachefs benchmarks, and their results looked a bit... off.
Here's what I just got, testing 4k random writes with a similar fio configuration. SSD is a Samsung 970 Evo plus. Default mkfs options for all three filesystems.
xfs: 1 job 456k iops, 8 jobs 548k iops
btrfs: 1 job 112k iops, 8 jobs 113k iops
bcachefs: 1 job 161k, 8 jobs 538k
So comparing bcachefs to xfs - we see much higher CPU usage because we're COW (every write has to allocate, and the extent mapping is much more fragmented because we're doing 4k random writes!) - but we do still scale, and we're able to max out the device with enough jobs.
Comments
Have you reached out? Seems like something they could revise/ correct.
Colin O'Brien
2023-11-10 01:31:54 +0000 UTCAnd that nasty error with bigger SSD?
sobkas
2023-11-07 22:51:25 +0000 UTCIf that is the case, will you disable it by default in the kernel?
John-Gee
2023-11-05 00:38:19 +0000 UTCI suspect a lot of this might be from having CONFIG_BCACHEFS_DEBUG_TRANSACTIONS on by default. hmm...
Kent Overstreet
2023-11-04 17:11:16 +0000 UTCIt's not like phoronix are known for their skillful use of or interpretation of benchmark data so this isn't terribly surprising.
Scott Bruce
2023-11-04 17:09:12 +0000 UTCSomething somewhere must be different, but if it is reproducible, it must also be fixable.
Nazar Mokrynskyi
2023-11-04 16:17:50 +0000 UTC🤔 Then I wonder why the numbers are different
brunoais
2023-11-04 16:07:08 +0000 UTC
