On Thursday 21 June 2001 02:15 pm, Franki wrote:
> Hi all,
>
> Is there any problems switching my 7200RPM ATA100 with Mdk 7.2 and
> reiserFS???
While I've been using ReiserFS 24/7 for almost a year now with no
problems, and many benefits I'd swear by .... it's still a beta FS.
You need to thoroughly investigate the pros & cons and make your own
decision. I use it on an old 5400rpm/256k WD slave with a 250mb /swap
and the rest of the 8 gigs all in one big / partition.
> Abit KT7a (capable of ata100) with bios dated this month..
> Linux Details.
> ReiserFS.
> Kernel 2.2.19-4MDKsecure
> all updates except kde loaded.
> Can anyone make any suggestions as to what my settings should be for
> maximum speed?
>
> It currently tests at:
>
> Buffer cache: 125.49 MB/sec
> Buffered: 29.91 MB/sec
Civileme's already given you the best advice along the lines I
mentioned in a prior post. There's VIA/mobo(specially **Abit KT7a)/HDD
(specially ***WD) issues, any OS. You seem to have gathered together
the hardware that's most susceptiple.
**Research why AMD dropped Abit's and particularly KT133a chipset
models (KT7a with/without raid) from their recommended mobo list last
February. Only the Biostar M7VKD and Soyo K7VTA Pro KT133a boards
still remain.
*** Civileme seems to imply all WD HDD's are affected, I've only
heard/read that some, mostly newer models were. Still, I'd trust
anything Civileme tells you.
>
>
> I was getting 22mb buffered, with my BX chipset and an old Quantum
> 4.3gig...
>
> 22 up to 29.91 doesn't seem like that much of an impovement...
>
> Especially since I have a ATA drive and motherboard connected with
> the ATA100 cable,
An 80 wire cable is nothing more than a good'ol ata/33 40 wire cable
with the other 40 wires doin nothin more than separating the real 40
wires to reduce inductance. Count the pins on the connector ;>>
Maybe I'm infering too much, but you seem to be enamored with
ata/100. Fact is, all HDD's whether they're ata/33, /66, or /100 still
run on the motherboard's 33.3mhz IDE bus. The ata/66 and /100 are
enhancements that don't, as their nomeclature implies, double or more
than double thu-put, or as many seem to believe, bus speed. What
improvement they may provide also comes at a price of risk and
problems. IMO, more than they're worth. Also, like AGP 1x,2x,and 4x,
the enhancement comes at the price of draining off other system
resources, namely cpu/cache/ram. There's no free lunch. If you're
getting 30mb/sec, be happy.
Be aware tho that that's a benchmark rate, real world transfer rates
are much less. I doubt that with a stopwatch you could tell much
difference tranferring a huge amount of files, whether the HDD hparmed
at 20mb/sec, or 30mb/sec.
> The drive is a 7200RPM model and I believe it has a 2mb internal
> cache... (I do think the buffer cache speed is far better then it
> was.) The drive is the only ide device on the primary ide interface.
'hdparm -i /dev/hd?' will tell you the buffer (cache size). It'll be
a little less than 2mb (2048k). The HDD's firmware uses the missing k's.
IMNSHO, higher rpm and bigger buffer HDD's are a much better idea
than the current fads of ata/100 and software-firmware raid kludges
like Abit's. An IBM 7200/2mb HDD would'a been a better choice tho ;>
...and a Soyo K7VTA Pro ;~>> YMMV
--
Tom Brinkman [EMAIL PROTECTED] Galveston Bay
