On Saturday 17 February 2001 07:00, you wrote:
> civileme wrote:
> > How about doing this? Since you have 2.4 on a machine with this
> > particular chip (we have NO examples) check the speed with hdparm and the
> > settings. How much tuning has happened? What can you produce if you
> > leave autotune out and try hdparm settings yourself? Which method works
> > better? And a dmesg with your report would be helpful.
>
> Cooker with 2.4 kernel and ATA100 proved totally disappointing -
> there must be a lot of work still to be done. Max cached throughput
> is about 20 MB/sec. Multisector = 0 and dma on. Multisector 16
> takes it back to 12 MB/sec. Very strange!
>
> 7.2 with the 2.2.27-28mdk kernel (this kernel supports IDE 2 & 3)
> immediately shows the advantage of ATA100. hda is an ATA 66 IBM
> drive on an ATA66 interface. hdg is an ATA100 IBM disk on a Promise
> ATA100 interface (on a Gigabyte GA-7ZXR motherboard with an AMD
> Athlon 900MHz CPU). This is the same Promise chip as is used on
> their add-on PCI mass storage card. DMA and 32 bit are not on as
> installed. I set them on.
>
> [root@small ron]# hdparm -v /dev/hda
>
> /dev/hda:
> multcount = 16 (on)
> I/O support = 1 (32-bit)
> unmaskirq = 0 (off)
> using_dma = 1 (on)
> keepsettings = 0 (off)
> nowerr = 0 (off)
> readonly = 0 (off)
> readahead = 8 (on)
> geometry = 1869/255/63, sectors = 30033360, start = 0
> [root@small ron]# hdparm -v /dev/hdg
>
> /dev/hdg:
> multcount = 16 (on)
> I/O support = 1 (32-bit)
> unmaskirq = 0 (off)
> using_dma = 1 (on)
> keepsettings = 0 (off)
> nowerr = 0 (off)
> readonly = 0 (off)
> readahead = 8 (on)
> geometry = 3737/255/63, sectors = 60036480, start = 0
>
> [root@small ron]# hdparm -tT /dev/hda
>
> /dev/hda:
> Timing buffer-cache reads: 128 MB in 1.02 seconds =125.49 MB/sec
> Timing buffered disk reads: 64 MB in 4.33 seconds = 14.78 MB/sec
>
> [root@small ron]# hdparm -tT /dev/hdg
>
> /dev/hdg:
> Timing buffer-cache reads: 128 MB in 0.96 seconds =133.33 MB/sec
> Timing buffered disk reads: 64 MB in 1.81 seconds = 35.36 MB/sec
>
> Conclusion:
>
> It looks like ATA100 more than doubles the buffered disk throughput
> of ATA66.
Thank you, Ron.
The 2.4 kernel doesn't have much turned on--we're stretched between locking
up old systems and detecting new ones, and we are working on moving most of
the tuning to post-install and avoiding autodetection of udma4 and 5.
Unfortunately, the data rates that seem so blinding in those tests with
hdparm don't hold in most real life situations. Copy a whole partition of a
Gig or more from one drive to another (on separate ide channels) Whether you
set the udma to mode 2 4 or 5 you get the same rate to the limits of accuracy
of the test. I have doe this more than once using UDMA5 capable drives and a
Promise controller.
A lot of these tests are highly drive and controller dependent. as well as
dependent on what else is running at time of test. Nevertheless your
information is very valuable in helping us refine this procedure.
from test 1 to test 2
Killed Netscape
from test 2 to test 3
Killed Konqueror
I _never_ kill emacs<g>
[root@civileme /root]# hdparm -i /dev/hda
/dev/hda:
Model=Maxtor 51536H2, FwRev=JAC61HU0, SerialNo=F208FBPC
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=57
BuffType=DualPortCache, BuffSize=2048kB, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=29888820
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 *udma4 udma5
[root@civileme /root]# hdparm -t /dev/hda
/dev/hda:
Timing buffered disk reads: 64 MB in 2.57 seconds = 24.90 MB/sec
[root@civileme /root]#
[root@civileme /root]# hdparm -t /dev/hda
/dev/hda:
Timing buffered disk reads: 64 MB in 2.14 seconds = 29.91 MB/sec
[root@civileme /root]#
[root@civileme /root]# hdparm -t /dev/hda
/dev/hda:
Timing buffered disk reads: 64 MB in 2.02 seconds = 31.68 MB/sec
[root@civileme /root]#
Please note that this is a 66 disk on a 66-capable chipset.
Civileme
