| 1 |
On 10/07/2010 12:59 AM, Adam Carter wrote: |
| 2 |
> WOW! Those differences are crazy! |
| 3 |
> |
| 4 |
> |
| 5 |
> Please - I know benchmarking takes a lot of time - but could you check |
| 6 |
> something: the behavior those fs have at what time they flush data from |
| 7 |
> cache to disk is very different. Have you made sure that you measured |
| 8 |
> the time it really needs? I mean the difference between: |
| 9 |
> |
| 10 |
> $ sync; time cp source dest |
| 11 |
> and |
| 12 |
> $ sync; time (cp source dest; sync) |
| 13 |
> |
| 14 |
> Only the last measures somewhat correctly. |
| 15 |
> |
| 16 |
> |
| 17 |
> I had noticed that there was, say, 5 seconds of disk activity after the |
| 18 |
> cp command complete which I assumed was buffers getting flushed, but 5 |
| 19 |
> seconds didnt seem that significant overall. I will run the tests as you |
| 20 |
> suggest and post back. Do you think btrfs (with or without compression) |
| 21 |
> would be faster than reiser? If so I will try that as well. |
| 22 |
On my system it is twice as fast as reiser3 for _lots_ (200.000) of |
| 23 |
small files with compression on (didn't test is without compression). I |
| 24 |
didn't test if with big files. |
| 25 |
|
| 26 |
But your results may vary anyway. For example btrfs is very |
| 27 |
cpu-intensive (even more with compression). If you've got a slow cpu |
| 28 |
(like in embedded devices), jfs might perform better. |
| 29 |
|
| 30 |
BTW: _all_ my partitions are encrypted and on LVM, so your use case is |
| 31 |
probably very different :) |
| 32 |
|
| 33 |
Bye, |
| 34 |
Daniel |
| 35 |
|
| 36 |
-- |
| 37 |
PGP key @ http://pgpkeys.pca.dfn.de/pks/lookup?search=0xBB9D4887&op=get |
| 38 |
# gpg --recv-keys --keyserver hkp://subkeys.pgp.net 0xBB9D4887 |
Archives