Re: 1be41b78: +18% increased btrfs write throughput

[Chris Mason]

On Fri, 2014-01-03 at 23:54 +-0800, fengguang.wu+AEA-intel.com wrote:
+AD4- Hi Josef,
+AD4- 
+AD4- FYI. We are doing 0day performance tests and happen to notice that
+AD4- btrfs write throughput increased considerably during v3.10-11 time
+AD4- frame:
+AD4- 
+AD4-           v3.10                      v3.11                      v3.12                  v3.13-rc6
+AD4- ---------------  -------------------------  -------------------------  -------------------------
+AD4-      50619 +AH4- 1+ACU-     +-17.0+ACU-      59209 +AH4- 2+ACU-     +-18.8+ACU-      60159 +AH4- 2+ACU-     +-20.5+ACU-      61007 +AH4- 0+ACU-  lkp-ws02/micro/dd-write/11HDD-JBOD-cfq-btrfs-1dd
+AD4-      50619          +-17.0+ACU-      59209          +-18.8+ACU-      60159          +-20.5+ACU-      61007       TOTAL iostat.sdd.wkB/s
+AD4- 
+AD4- and it's contributed by 
+AD4- 
+AD4- commit 1be41b78bc688fc634bf30965d2be692c99fd11d
+AD4- Author:     Josef Bacik +ADw-jbacik+AEA-fusionio.com+AD4-
+AD4- AuthorDate: Wed Jun 12 13:56:06 2013 -0400
+AD4- Commit:     Josef Bacik +ADw-jbacik+AEA-fusionio.com+AD4-
+AD4- CommitDate: Mon Jul 1 08:52:28 2013 -0400
+AD4- 
+AD4-     Btrfs: fix transaction throttling for delayed refs

Bonus points for increasing the performance on purpose.  Thanks for
running these Wu.

-chris

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/