pci overclock

i have a hpt366 based ata-66 controller card that claims operability at
44MHz.
how much would it help my bx overclocking efforts to place both my hdds on
that controller?
are my dvd/cd devices part of that picture?

i suspect running that pci bus many of peripherals would not function
properly.

On Sun, 12 Oct 2003 20:57:44 +0200, "flekso" wrote:

:i have a hpt366 based ata-66 controller card that claims operability at
:44MHz.
:how much would it help my bx overclocking efforts to place both my hdds on
:that controller?
:are my dvd/cd devices part of that picture?
:

-----
Lee.

What motherboard? Some BX chipset motherboards allow a PCI bus speed
divider of 1/4 (the aBit BX6 rev.2, for example.) That will give an in-spec
PCI bus speed of 33 MHz with a FSB speed of 133 MHz, and a PCI bus speed of
44 MHz with a FSB speed of 176 MHz; I don't think the latter is going to
happen.

PC133 CL2 memory will likely turn flakey at ~ 156 MHz, and with a 1/4
divider, the PCI bus speed will be only 39 MHz. Most IDE controllers will
do O.K. up to 37 MHz to 39 MHz (with a 1/4 divider that would be a FSB speed
of 148 MHz to 156 MHz.) An over clocked PCI bus is problem for the
controller, one that can slow down data transfers because errors caused by
improper timing will cause data to be retransfered. Errors due to improper
timing that are NOT caught can cause data corruption. Being a controller
problem, it can affect any device hooked to the PCI bus.




-
--
Phil Weldon, pweldonatmindjumpdotcom
For communication,
replace "at" with the 'at sign'
replace "mindjump" with "mindspring."
replace "dot" with "."

"flekso" wrote in message
...
i have a hpt366 based ata-66 controller card that claims operability at
44MHz.
how much would it help my bx overclocking efforts to place both my hdds on
that controller?
are my dvd/cd devices part of that picture?

"Phil Weldon" wrote in message
hlink.net...

What motherboard? Some BX chipset motherboards allow a PCI bus speed
divider of 1/4 (the aBit BX6 rev.2, for example.) That will give an
in-spec
PCI bus speed of 33 MHz with a FSB speed of 133 MHz, and a PCI bus speed
of
44 MHz with a FSB speed of 176 MHz; I don't think the latter is going to
happen.

PC133 CL2 memory will likely turn flakey at ~ 156 MHz, and with a 1/4
divider, the PCI bus speed will be only 39 MHz. Most IDE controllers will
do O.K. up to 37 MHz to 39 MHz (with a 1/4 divider that would be a FSB
speed
of 148 MHz to 156 MHz.) An over clocked PCI bus is problem for the
controller, one that can slow down data transfers because errors caused by
improper timing will cause data to be retransfered. Errors due to
improper
timing that are NOT caught can cause data corruption. Being a controller
problem, it can affect any device hooked to the PCI bus.




-
--
Phil Weldon, pweldonatmindjumpdotcom
For communication,
replace "at" with the 'at sign'
replace "mindjump" with "mindspring."
replace "dot" with "."

"flekso" wrote in message
...
i have a hpt366 based ata-66 controller card that claims operability at
44MHz.
how much would it help my bx overclocking efforts to place both my hdds
on
that controller?
are my dvd/cd devices part of that picture?





but peripherals are connected to 44MHz capable card, doesn't that make any
difference ?

"flekso" wrote
i have a hpt366 based ata-66 controller card that claims operability at
44MHz.
how much would it help my bx overclocking efforts to place both my hdds on
that controller?
are my dvd/cd devices part of that picture?


Hola Amigo,

I think you may be taking this overclock experiment a wee bit to far. .
..know what I mean :P
--
Wayne ][

flekso wrote:
i have a hpt366 based ata-66 controller card that claims operability at
44MHz.
how much would it help my bx overclocking efforts to place both my hdds on
that controller?
are my dvd/cd devices part of that picture?



The last question is the easiest. Yes, they'd be part of the picture since
they're on an IDE controller and subject to the same interface speed problems.

The first one is harder because I don't know exactly what they 'mean' by
operable on a 44 MHz PCI bus. Do they do something special to ensure that
'operability', like regenerate a suitable clock for the IDE port so the
hard drives 'see' a 33 Mhz like signal or do they just mean they can keep
up at 44 MHz and the drive sees the same thing, so that it's got the same
'keep up' problem? I rather suspect it's the latter.

However, I suspect it would work if the drive is 100 UDMA capable. I.E. the
drive's speed capability would be faster than the overclocked UDMA66 speed.

That's not a technically rigorous analysis but it matches my general
experience. For example, if we take what was the conventional wisdom from
the days when your BX board was made, overclocking the PCI bus often caused
hard drive problems with UDMA33, 5400 RPM, drives on the typical UDAM33
controller those boards have. However, if you tried the exact same thing
with a UDMA66 drive it operates just fine even at the previously 'drastic'
overclock. In the case of a UDMA66 controller, a UDMA100 drive gives the
same 'overhead' to, hopefully, allow the controller overclocking.

It's not just theory. I'm running a UDMA100 WD on a UDMA66 controller with
a 38 MHz PCI bus as I type. Of course, that's not 44 MHz and I don't know
at what speed the theory would fail due to other timing constraints that
may not scale the same when overclocking vs simply operating at UDMA100,
however, I have operated at 41 MHz when it was a UDMA66 drive on a UDMA33
controller, which is getting close.

As for hitting 44 MHz it worked there too but the P-III wasn't 100% stable.
I also lost the PCI sound card at 44Mhz, but the NIC and video capture
cards operated.

Well, a lot of peripherals may be connected to the PCI bus; not just what
you connect to an IDE outboard card. If your sound card chokes a few times,
no harm except to your ears. If a hard drive chokes during a write, well,
that could be bad. And then there is the question "What does the
manufacturer mean by 44 MHz?" That any IDE hard drive connected to this
card will operate correctly at a PCI bus speed of 44 MHz? And what about
the other PCI devices that are not connected to their card? PCI sound card?
LAN card?

If the card has a buffer enough larger than the on-drive buffer, it could
insulate the drive circuitry from the overspeed PCI bus, but I seriously
doubt it does. If it doesn't, what does the drive do when data moves into
the on-drive controller (most of the controller is on-board for IDE drives)
faster than its design speed? Corrupt the data? And that is exactly what
happens when most IDE hard drives are run on a 44 MHz PCI bus.

Try it and find out if you already have the components; I just would not
risk any important data.
--
Phil Weldon, pweldonatmindjumpdotcom
For communication,
replace "at" with the 'at sign'
replace "mindjump" with "mindspring."
replace "dot" with "."

"flekso" wrote in message
...
..
..
but peripherals are connected to 44MHz capable card, doesn't that make any
difference ?

Happy Hippo wrote:
On Sun, 12 Oct 2003 19:11:09 -0500, David Maynard
wrote:


flekso wrote:

i have a hpt366 based ata-66 controller card that claims operability at
44MHz.
how much would it help my bx overclocking efforts to place both my hdds on
that controller?
are my dvd/cd devices part of that picture?



The last question is the easiest. Yes, they'd be part of the picture since
they're on an IDE controller and subject to the same interface speed problems.

The first one is harder because I don't know exactly what they 'mean' by
operable on a 44 MHz PCI bus. Do they do something special to ensure that
'operability', like regenerate a suitable clock for the IDE port so the
hard drives 'see' a 33 Mhz like signal or do they just mean they can keep
up at 44 MHz and the drive sees the same thing, so that it's got the same
'keep up' problem? I rather suspect it's the latter.

However, I suspect it would work if the drive is 100 UDMA capable. I.E. the
drive's speed capability would be faster than the overclocked UDMA66 speed.

That's not a technically rigorous analysis but it matches my general
experience. For example, if we take what was the conventional wisdom from
the days when your BX board was made, overclocking the PCI bus often caused
hard drive problems with UDMA33, 5400 RPM, drives on the typical UDAM33
controller those boards have. However, if you tried the exact same thing
with a UDMA66 drive it operates just fine even at the previously 'drastic'
overclock. In the case of a UDMA66 controller, a UDMA100 drive gives the
same 'overhead' to, hopefully, allow the controller overclocking.

It's not just theory. I'm running a UDMA100 WD on a UDMA66 controller with
a 38 MHz PCI bus as I type. Of course, that's not 44 MHz and I don't know
at what speed the theory would fail due to other timing constraints that
may not scale the same when overclocking vs simply operating at UDMA100,
however, I have operated at 41 MHz when it was a UDMA66 drive on a UDMA33
controller, which is getting close.

As for hitting 44 MHz it worked there too but the P-III wasn't 100% stable.
I also lost the PCI sound card at 44Mhz, but the NIC and video capture
cards operated.





3year ago I operated a celereon 333 at 420 (because that damn thing
did not want to do 500, since I knew nothing about overvolting back
then) with a FSB of 83.3 MHz on a Asus P2B, which is 41.65 MHz (which
is not 44 MHz though) on he PCI bus. I had a 6 GB Western Digital
drive and a Vodoo 3. (also some ISA stuff for the Soundcard and Lan)

Just note that you will be OC'ing your video card too. The vodoo 3
made no problems, it just ran faster at an agp of 83 MHz .

Today, knowing a little more I would not recommend to do such a thing
anymore, because of the problems this can cause with the harddrives
etc. and now I consider it not worth it.

HH

I wouldn't be surprised if you had hard drive problems with that PCI speed
because that WD 6 gig was a UDMA33, no doubt, and so are the P2B IDE ports.

Forcing PIO4 will usually let those drives work and I do that with old
UDMA33 drives used for miscellaneous data storage.

Btw, I presume that was a slot-1 Celeron 333 and it was rather rare to get
them to 500, even with increased Vcore, as the steppings didn't improve
enough for 500+ speeds till right at the end of Slot-1 manufacture. The
latest stepping PPGA celerons were capable of 550+.