Re: problems with I/O performance with 2.4.12-ac3

Re: problems with I/O performance with 2.4.12-ac3

[safemode]

On Friday 19 October 2001 10:30, Andrei Lahun wrote:
> hello.
>
> I did a bonnie++ tests with 2.4.13-pre3(+aa vm)
>  and 2.4.12-ac3(+Rick patch).
> And i found that ac kernel have a big problems here.
> With linux kernel i got 22 Mb/c for read 19 for write and
> with ac  12 for read and 11 for write (both with ext2)
> If i used ext3 with ac kernel results little bit better.
> What is the reason for such regression in ac kernel ?
> And yes of course i had udma for both test.
>
> Andrei.

You should give ide and drive chipset info.  This is not a problem seen by 
everyone using the ac3 kernel.  Mine for instance run just fine.  What 
settings did you use with bonnie++?  all of this is required info if someone 
wanted to look and see why you are getting those numbers.

problems with I/O performance with 2.4.12-ac3

[Andrei Lahun]

hello.

I did a bonnie++ tests with 2.4.13-pre3(+aa vm)
 and 2.4.12-ac3(+Rick patch).
And i found that ac kernel have a big problems here.
With linux kernel i got 22 Mb/c for read 19 for write and
with ac  12 for read and 11 for write (both with ext2)
If i used ext3 with ac kernel results little bit better.
What is the reason for such regression in ac kernel ?
And yes of course i had udma for both test.

Andrei.
 

Re: problems with I/O performance with 2.4.12-ac3

[Dmitry Volkoff]

Hi,
On my machine HDD is slow with all ac-kernels I've tried so far.
There is even more simple test -- hdparm.

2.4.12-ac3:
# hdparm -tT /dev/hda
/dev/hda:
Timing buffer-cache reads:   128 MB in  0.70 seconds =182.86 MB/sec
Timing buffered disk reads:  64 MB in  2.88 seconds = 22.22 MB/sec

2.4.13-pre3:
# hdparm -tT /dev/hda
/dev/hda
Timing buffer-cache reads:   128 MB in  0.63 seconds =203.00 MB/sec
Timing buffered disk reads:  64 MB in  1.57 seconds = 40.76 MB/sec

Same config options in both kernels. Same hardware.

# hdparm /dev/hda
/dev/hda:
multcount    = 16 (on)
I/O support  =  1 (32-bit)
unmaskirq    =  1 (on)
using_dma    =  1 (on)
keepsettings =  1 (on)
nowerr       =  0 (off)
readonly     =  0 (off)
readahead    =  8 (on)
geometry     = 4865/255/63, sectors = 78165360, start = 0
	 
This is on Athlon-1.4Ghz, chipset amd761/via8231, 
IDE Seagate Barracuda ATA-IV ST340016A 40G.
Can somebody explain such results?

-- 

    DV

Re: problems with I/O performance with 2.4.12-ac3

[Andrei Lahun]

On Fri, 19 Oct 2001 09:27:35 -0400
safemode <safemode@speakeasy.net> wrote:

>
> You should give ide and drive chipset info.  This is not a problem seen by 
> everyone using the ac3 kernel.  Mine for instance run just fine.  What 
> settings did you use with bonnie++?  all of this is required info if someone 
> wanted to look and see why you are getting those numbers.
> 
My chipset is VIA apollo 133-A ( old shirt).
I have run bonnie++ just like bonnie -s 256 , and i have 128 MB of memory,
disk is MAXTOR 52049H4.With old kernels i always had results something like i have with 2.4.13-pre now .
And i have read one post in lkml about 2 times less write output with 2.4.13 -pre comparing with ac,
mine result are opposite.

Re: problems with I/O performance with 2.4.12-ac3

[safemode]

On Friday 19 October 2001 13:14, Andrei Lahun wrote:
> On Fri, 19 Oct 2001 09:27:35 -0400
>
> safemode <safemode@speakeasy.net> wrote:
> > You should give ide and drive chipset info.  This is not a problem seen
> > by everyone using the ac3 kernel.  Mine for instance run just fine.  What
> > settings did you use with bonnie++?  all of this is required info if
> > someone wanted to look and see why you are getting those numbers.
>
> My chipset is VIA apollo 133-A ( old shirt).
> I have run bonnie++ just like bonnie -s 256 , and i have 128 MB of memory,
> disk is MAXTOR 52049H4.With old kernels i always had results something like
> i have with 2.4.13-pre now . And i have read one post in lkml about 2 times
> less write output with 2.4.13 -pre comparing with ac, mine result are
> opposite.

2.4.12-ac3-preempt-hogstop   (no vm patches after the hogstop patch of a 
couple days ago).
PDC20262: chipset revision 1
(my drive on the VIA vt82c686a (rev 22)  is too full to run the test) 
Maxtor 32049H2, ATA DISK drive
All tests were done on this drive.  

(not clean boot) 

test results (bonnie++)  no extra arguments.
Version  1.02       ------Sequential Output------ --Sequential Input- 
--Random-
                    -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- 
--Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec 
%CP
psuedomode    1504M  6668  95 27108  39  5429   8  6217  82 22005  11  76.9   
0
                    ------Sequential Create------ --------Random 
Create--------
                    -Create-- --Read--- -Delete-- -Create-- --Read--- 
-Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec 
%CP
                 16  1501  97 +++++ +++ +++++ +++  1548  99 +++++ +++  4056  
99
psuedomode,1504M,6668,95,27108,39,5429,8,6217,82,22005,11,76.9,0,16,1501,97,+++++,+++,+++++,+++,1548,99,+++++,+++,4056,99

hdparm results 
 Timing buffered disk reads:  64 MB in  2.26 seconds = 28.32 MB/sec
 Timing buffer-cache reads:   128 MB in  1.00 seconds =128.00 MB/sec
(run in that order)

dbench results (./dbench 32)
Throughput 47.3829 MB/sec (NB=59.2286 MB/sec  473.829 MBit/sec)

Re: problems with I/O performance with 2.4.12-ac3

[Andrea Arcangeli]

On Fri, Oct 19, 2001 at 07:53:28PM +0400, Dmitry Volkoff wrote:
> Hi,
> On my machine HDD is slow with all ac-kernels I've tried so far.
> There is even more simple test -- hdparm.
> 
> 2.4.12-ac3:
> # hdparm -tT /dev/hda
> /dev/hda:
> Timing buffer-cache reads:   128 MB in  0.70 seconds =182.86 MB/sec
> Timing buffered disk reads:  64 MB in  2.88 seconds = 22.22 MB/sec
> 
> 2.4.13-pre3:
> # hdparm -tT /dev/hda
> /dev/hda
> Timing buffer-cache reads:   128 MB in  0.63 seconds =203.00 MB/sec
> Timing buffered disk reads:  64 MB in  1.57 seconds = 40.76 MB/sec

this should be mostly the blkdev in pagecache and possibly also the vm.

Andrea