1 |
On 10/06/2010 10:04 AM, Adam Carter wrote: |
2 |
> FYI some braindead benchmarking, reiserfs vs ext4, kernel 2.6.35-gentoo-r8 |
3 |
> |
4 |
> Copy same DVD image from internal reiserfs drive to freshly formatted |
5 |
> external drive; |
6 |
> reiserfs 1m37.530s |
7 |
> ext4 3m15.074s |
8 |
> |
9 |
> Then image copy on that external drive; |
10 |
> # time cp CentOS-5.3-x86_64-bin-DVD.iso CentOS-5.3-x86_64-bin-DVD.iso2 |
11 |
> # time cp CentOS-5.3-x86_64-bin-DVD.iso CentOS-5.3-x86_64-bin-DVD.iso3 |
12 |
> reiser 1m44.719s and 1m51.441s |
13 |
> ext4 3m24.337s and 4m30.534s |
14 |
> |
15 |
> Not that is matters, but create filesystem on 2TB drive; |
16 |
> reiserfs 1m17.373s |
17 |
> ext4 6m3.421s |
18 |
> |
19 |
> Didnt see that coming, I guess i'll stick with reiser3.... |
20 |
WOW! Those differences are crazy! |
21 |
|
22 |
Please - I know benchmarking takes a lot of time - but could you check |
23 |
something: the behavior those fs have at what time they flush data from |
24 |
cache to disk is very different. Have you made sure that you measured |
25 |
the time it really needs? I mean the difference between: |
26 |
|
27 |
$ sync; time cp source dest |
28 |
and |
29 |
$ sync; time (cp source dest; sync) |
30 |
|
31 |
Only the last measures somewhat correctly. |
32 |
|
33 |
I'm irritated, because ext4 is extends based, and should behave much |
34 |
better with big files than reiser3... not only less fragmentation, but |
35 |
should also be faster... |
36 |
|
37 |
Bye, |
38 |
Daniel |
39 |
|
40 |
-- |
41 |
PGP key @ http://pgpkeys.pca.dfn.de/pks/lookup?search=0xBB9D4887&op=get |
42 |
# gpg --recv-keys --keyserver hkp://subkeys.pgp.net 0xBB9D4887 |