Re: [Ext2-devel] Re: [RFC][PATCH 0/2]Extend ext3 filesystem limitfrom 8TB to 16TB

[Laurent Vivier]

Le mar 18/04/2006 Ã 09:30, Arjan van de Ven a Ãcrit :
> On Tue, 2006-04-18 at 09:14 +0200, Laurent Vivier wrote:
> > Le lun 17/04/2006 Ã 23:32, Ravikiran G Thirumalai a Ãcrit :
> > > On Mon, Apr 17, 2006 at 11:09:36PM +0200, Arjan van de Ven wrote:
> > > > On Mon, 2006-04-17 at 14:07 -0700, Ravikiran G Thirumalai wrote:
> > > > > 
> > > > > 
> > > > > I ran the same tests on a 16 core EM64T box very similar to the one
> > > > > you ran
> > > > > dbench on :). Dbench results on ext3 varies quite a bit.  I couldn't
> > > > > get 
> > > > > to a statistically significant conclusion  For eg,
> > > > 
> > > > 
> > > > dbench is not a good performance benchmark. At all. Don't use it for
> > > > that ;)
> > > 
> > > Agreed. (I did not mean to use it in the first place :).  I was just trying 
> > > to verify the benchmark results posted earlier)
> > > 
> > > Thanks,
> > > Kiran
> > 
> > What is the good performance benchmark to know if we should use atomic_t
> > instead of percpu_counter ?
> 
> you probably want something like postal/postmark instead or so (although
> that's not ideal either), at least that's reproducable

I made some tests with kernbench too:

***** With percpu_counter:

16 cpus found
Cleaning source tree...
Caching kernel source in ram...
No old config found, using defconfig
Making mrproper
Making defconfig...
Kernel 2.6.16
Performing 5 runs of
make -j 8
make -j 64
make -j

All data logged to kernbench.log
Warmup run...
Half load -j 8 run number 1...
Half load -j 8 run number 2...
Half load -j 8 run number 3...
Half load -j 8 run number 4...
Half load -j 8 run number 5...
Average Half load -j 8 Run (std deviation):
Elapsed Time 120.68 (0.425558)
User Time 583.488 (0.54099)
System Time 84.716 (0.345948)
Percent CPU 553 (2)
Context Switches 13146.4 (66.3272)
Sleeps 26998.2 (297.078)

Optimal load -j 64 run number 1...
Optimal load -j 64 run number 2...
Optimal load -j 64 run number 3...
Optimal load -j 64 run number 4...
Optimal load -j 64 run number 5...
Average Optimal load -j 64 Run (std deviation):
Elapsed Time 86.496 (0.335827)
User Time 809.699 (238.449)
System Time 103.137 (19.423)
Percent CPU 945.3 (413.544)
Context Switches 32549.5 (20471.2)
Sleeps 34308 (7795.17)

Maximal load -j run number 1...
Maximal load -j run number 2...
Maximal load -j run number 3...
Maximal load -j run number 4...
Maximal load -j run number 5...
Average Maximal load -j Run (std deviation):
Elapsed Time 86.47 (0.321636)
User Time 883.568 (219.647)
System Time 108.728 (17.597)
Percent CPU 1073.8 (381.226)
Context Switches 31920.4 (16443.3)
Sleeps 30472.5 (8402.01)

***** With atomic_long_t

16 cpus found
Cleaning source tree...
Caching kernel source in ram...
No old config found, using defconfig
Making mrproper
Making defconfig...
Kernel 2.6.16
Performing 5 runs of
make -j 8
make -j 64
make -j

All data logged to kernbench.log
Warmup run...
Half load -j 8 run number 1...
Half load -j 8 run number 2...
Half load -j 8 run number 3...
Half load -j 8 run number 4...
Half load -j 8 run number 5...
Average Half load -j 8 Run (std deviation):
Elapsed Time 120.468 (0.724134)
User Time 581.226 (0.497624)
System Time 84.358 (0.45417)
Percent CPU 551.8 (3.19374)
Context Switches 13085.6 (108.579)
Sleeps 26965.8 (189.384)

Optimal load -j 64 run number 1...
Optimal load -j 64 run number 2...
Optimal load -j 64 run number 3...
Optimal load -j 64 run number 4...
Optimal load -j 64 run number 5...
Average Optimal load -j 64 Run (std deviation):
Elapsed Time 86.25 (0.263439)
User Time 805.828 (236.752)
System Time 102.262 (18.8792)
Percent CPU 942.7 (412.059)
Context Switches 32339.7 (20299.6)
Sleeps 34301.9 (7741.15)

Maximal load -j run number 1...
Maximal load -j run number 2...
Maximal load -j run number 3...
Maximal load -j run number 4...
Maximal load -j run number 5...
Average Maximal load -j Run (std deviation):
Elapsed Time 85.868 (0.757905)
User Time 879.129 (218.053)
System Time 107.847 (17.2136)
Percent CPU 1072.73 (381.349)
Context Switches 31854.5 (16297.1)
Sleeps 30436.2 (8399.98)


Laurent
-- 
Laurent Vivier
Bull, Architect of an Open World (TM)
http://www.bullopensource.org/ext4

Attachment: signature.asc
Description: Ceci est une partie de message=?ISO-8859-1?Q?num=E9riquement?= =?ISO-8859-1?Q?_sign=E9e=2E?=