Hi,
I have run some "files bigger than memory" (streaming) tests.
Some significant differences with earlier kernel:
* write - lowered throughput (26 MB/s => 22 MB/s)
* copy - throughput better by 2/3 (16 MB/s => 25 MB/s) !
* dbench 32 - back down to lower than 2.4.12 (due to increased fairness?)
2.4.11 18.9 MB/s
2.4.12 23.3 MB/s
2.4.16 34.9 MB/s
2.4.17-rc1 20.3 MB/s
2.4.17-rc1 (file-readahead:1000) 24.5 MB/s
[lets forget about them now...]
* diff - usage of "file-readahead" more than doubles efficency
(most-kernels: 11 MB/s => 2.4.17-rc1 with readahead: 25 MB/s)
The nicest thing is that it is the first kernel where readahead tuning is not
necessary for the copy operation.
Now it is only multiple-big-concurrent-reads that _needs_ "file-readahead"
/RogerL
-- Roger Larsson Skellefteċ Sweden - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/