Partitioning questions

Richard Lightman richard at nezumi.plus.com
Fri Oct 4 05:59:26 PDT 2002


Matthias Benkmann wrote:
>On Sat, 21 Sep 2002 11:59:58 +1000 Ben <benhoskings at hotmail.com> wrote:
>
>
>>This will work perfectly but one thing to think about is that the start 
>>of a HDD is quite a bit faster than the end,
>
>
>AFAIK, this is outdated for modern disks. If you map increasing LBA
>numbers to access speed you will get a more complex pattern than a simple
>linear increase or decrease. If you want to optimize partition placement
>you will need a program that reads (and writes) the whole disk raw (i.e.
>without any buffering by the kernel) from front to back and measures
>speed. But I think this is a wasted effort. Seek times (which are orders
>of magnitude higher than read/write time for a sector), buffering by the
>filesystem driver and the fact that these days disk access is done in the
>background with DMA are likely to annihilate any differences the placement
>would make.
>
It in not linear. There is only a small drop in performance over most
of the disc, but near the end performance drops quickly. This is
because there is a lower data transfer rate at the end of the disk,
and also the lower data density increases the chances that a head
movement is required.

For a heavily loaded machine, this often makes little difference to
throughput, but can show up more as increased latency.

Here are some results for my system (about 2 years old). I have added
the sizes to the output of 'hdparm -tT'. Volume group urusai is on
partition 1, but apart from that I have put these in order from the
outside to the inside if the disk.

/dev/discs/disc0/disc:  10240M
 Timing buffer-cache reads:   128 MB in  0.82 seconds =156.10 MB/sec
 Timing buffered disk reads:  64 MB in  2.80 seconds = 22.86 MB/sec

/dev/discs/disc0/part1: 10133M
 Timing buffer-cache reads:   128 MB in  0.93 seconds =137.63 MB/sec
 Timing buffered disk reads:  64 MB in  2.80 seconds = 22.86 MB/sec

/dev/urusai/stable:      2097M
 Timing buffer-cache reads:   128 MB in  0.93 seconds =137.63 MB/sec
 Timing buffered disk reads:  64 MB in  2.84 seconds = 22.54 MB/sec

/dev/urusai/swap1:       1178M
 Timing buffer-cache reads:   128 MB in  0.89 seconds =143.82 MB/sec
 Timing buffered disk reads:  64 MB in  2.80 seconds = 22.86 MB/sec

/dev/urusai/var_s:        104M
 Timing buffer-cache reads:   128 MB in  0.94 seconds =136.17 MB/sec
 Timing buffered disk reads:  64 MB in  2.86 seconds = 22.38 MB/sec

/dev/urusai/home:        1048M
 Timing buffer-cache reads:   128 MB in  0.93 seconds =137.63 MB/sec
 Timing buffered disk reads:  64 MB in  2.83 seconds = 22.61 MB/sec

/dev/urusai/var_t:        104M
 Timing buffer-cache reads:   128 MB in  0.91 seconds =140.66 MB/sec
 Timing buffered disk reads:  64 MB in  2.91 seconds = 21.99 MB/sec

/dev/urusai/swap2:       1178M
 Timing buffer-cache reads:   128 MB in  0.93 seconds =137.63 MB/sec
 Timing buffered disk reads:  64 MB in  2.88 seconds = 22.22 MB/sec

/dev/urusai/test:        2097M
 Timing buffer-cache reads:   128 MB in  0.90 seconds =142.22 MB/sec
 Timing buffered disk reads:  64 MB in  3.09 seconds = 20.71 MB/sec

/dev/urusai/backup:      2097M
 Timing buffer-cache reads:   128 MB in  0.92 seconds =139.13 MB/sec
 Timing buffered disk reads:  64 MB in  3.47 seconds = 18.44 MB/sec

/dev/discs/disc0/part2:   107M
 Timing buffer-cache reads:   128 MB in  0.91 seconds =140.66 MB/sec
 Timing buffered disk reads:  64 MB in  4.50 seconds = 14.22 MB/sec

I am a bit suprised by the buffer-cache reads. AFAIK, the bus speed
is 66MHz, and the bus is 16 bits, so how can I get results over
133 MB/sec? The controller is AMD-756 [Viper] IDE (rev 3).

Richard

-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-support' in the subject header of the message



More information about the lfs-support mailing list