<br><br><div><span class="gmail_quote">On 3/31/06, <b class="gmail_sendername">Wander Winkelhorst</b> <<a href="mailto:w.winkelhorst@gmail.com">w.winkelhorst@gmail.com</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On 3/31/06, Brian Wood <<a href="mailto:beww@beww.org">beww@beww.org</a>> wrote:<br>><br>> On Mar 30, 2006, at 10:33 PM, Martin Bene wrote:<br>><br>> >> Since the OP's hdparm -tT numbers are way low it's obvious he has a
<br>> >> problem not related to filesystem, although he might have those<br>> >> troubles as well.<br>> ><br>> >> /dev/hdc:<br>> >> Timing cached reads: 1392 MB in 2.01 seconds =
692.30 MB/sec<br>> >> Timing buffered disk reads: 188 MB in 3.00 seconds = 62.63 MB/sec<br>> >> [root@mythtv mythtv]# hdparm -tT /dev/hda<br>> >><br>> >> /dev/hda:<br>> >> Timing cached reads: 1368 MB in
2.00 seconds = 683.76 MB/sec<br>> >> Timing buffered disk reads: 124 MB in 3.02 seconds = 41.12 MB/sec<br>> ><br>> > I don't see any indication of slow harddisks in these numbers -<br>> > "buffered disk reads" are OK.
<br>> ><br>> > "cached reads" don't touch the disk hardware or interface at all, this<br>> > is cpu <-> memory bandwidth. Agreed, for a new system the<br>> > throughput is<br>> > rather pathetic.
<br>> ><br>><br>> I had assumed that "cached reads" were reading from the drive's<br>> internal cache, and would thus use the drives interface, but that was<br>> only an sssumption.<br><br>Well, ATA133 (The fastest parralel ATA to date) has only a maximum
<br>throughput of 133 megabyte/sec, so there is no way you can do 683.76<br>MB/sec<br><br>><br>> In any case, hdparm clearly showed that his hda drive was running in<br>> udma5 mode while the hdc drive was not, most likely due to the latter
<br>> being on the same bus with an optical drive, and this clearly could<br>> and should be corrected.<br><br>Optical drives and hard discs can share a ATA bus without any slowdown<br>at all. The optical won't limit the hard disc to ATA33.
<br><br>What will cause problems is using a 40 wire ATA cable instead of a 80<br>wire one. That WILL limit you to 33MB/sec speeds.<br><br>Allso, ATA is a 1 device-at-a-time protocol, so keep the devices that<br>you use the most at the same time on seperate busses.
<br>So don't put the CDROM and the harddisc you are ripping to on the same<br>bus, and dont put the two hard discs on the same bus either<br><br>For best performance, buy each device (HD, CDROM, DVD, whatever) it's own bus.
</blockquote><div><br><br>That's why I had put them on seperate busses in the first place. I swapped the drives around, moving hdc to hdb, and got almost identical results this morning. I'll have to see later whether it helps. The main difference is now hdb shows udma5 as being enabled.
<br><br><br><br>/dev/hdb:<br><br> Model=ST3250823A, FwRev=3.02, SerialNo=5ND064BR<br> Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs RotSpdTol>.5% }<br> RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
<br> BuffType=unknown, BuffSize=8192kB, MaxMultSect=16, MultSect=16<br> CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=268435455<br> IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}<br> PIO modes: pio0 pio1 pio2 pio3 pio4
<br> DMA modes: mdma0 mdma1 mdma2<br> UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5<br> AdvancedPM=no WriteCache=enabled<br> Drive conforms to: device does not report version:<br><br> * signifies the current active mode
<br><br>[root@mythtv mythtv]# hdparm -tT /dev/hdb<br><br>/dev/hdb:<br> Timing cached reads: 1448 MB in 2.01 seconds = 721.94 MB/sec<br> Timing buffered disk reads: 184 MB in 3.01 seconds = 61.22 MB/sec<br>[root@mythtv
mythtv]# hdparm -tT /dev/hda<br><br>/dev/hda:<br> Timing cached reads: 1460 MB in 2.00 seconds = 729.38 MB/sec<br> Timing buffered disk reads: 148 MB in 3.01 seconds = 49.24 MB/sec<br><br><br><br>Ryan<br></div><br>
</div><br>