https://qa.mandrakesoft.com/show_bug.cgi?id=611





------- Additional Comments From [EMAIL PROTECTED]  2002-12-15 12:31 -------
With kernel version 2.4.20.2mdk the boot problem with /dev/hde/ UDMA disk is
solved.





------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.


   
------- Reminder: -------
assigned_to: [EMAIL PROTECTED]
description: 
Since kernel 2.4.19, UDMA could been enabled successfully on i845 chipset-based
motherboards.
Up to 2.4.20.0.3mdk kernel, hard drive were performing correctly UDMA transfers.
However, since kernel 2.4.20.0.5mdk (note: 2.4.20.0.4mdk was not tested) and
2.4.20.1mdk kernels, UDMA can still be enabled on i845 motherboards, but hard
drive performance is irregular.
It seems that using buffer cache of the disk before a read/write transfer
decrease the transfer performance to non-UDMA levels.

Examples :
IDE hard drive 
Model=HITACHI_DK23DA-20, FwRev=00J2A0A1, SerialNo=11L0V8
 Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs }
 RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
 BuffType=DualPortCache, BuffSize=2048kB, MaxMultSect=16, MultSect=16
 CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=39070080
 IORDY=yes, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
 PIO modes:  pio0 pio1 pio2 pio3 pio4
 DMA modes:  mdma0 mdma1 mdma2
 UDMA modes: udma0 udma1 udma2 udma3 udma4 *udma5
 AdvancedPM=yes: mode=0x80 (128) WriteCache=enabled
 Drive conforms to: ATA/ATAPI-5 T13 1321D revision 3:  2 3 4 5

multcount    = 16 (on)
 IO_support   =  1 (32-bit)
 unmaskirq    =  1 (on)
 using_dma    =  1 (on)
 keepsettings =  0 (off)
 readonly     =  0 (off)
 readahead    =  8 (on)
 geometry     = 2432/255/63, sectors = 39070080, start = 0

1) with kernel 2.4.20.0.3mdk
hdparm -Tt /dev/hda
/dev/hda:
 Timing buffer-cache reads:   128 MB in  0.37 seconds =345.95 MB/sec
 Timing buffered disk reads:  64 MB in  2.78 seconds = 23.02 MB/sec

This result was consistent with the 2.4.20.0.3mdk kernel.

2) But with kernel 2.4.20.1mdk
hdparm -t /dev/hda
Timing buffered disk reads:  64 MB in  4.61 seconds = 13.88 MB/sec

but it gives irregular results comprised between 10 to 20

and
hdparm -Tt /dev/hda
/dev/hda:
 Timing buffer-cache reads:   128 MB in  0.37 seconds =345.95 MB/sec
 Timing buffered disk reads:  64 MB in  6.55 seconds =  9.77 MB/sec

Giving consistently about 9MB/sec, the same result as if UDMA was not enabled.
This result is always obtained if a buffer-cache read is performed before the
disk read.

This could be verified with hdparm tests, but on a normal use basis, this
globally cripples hard drive performance on that hardware.

Reply via email to