Archived from groups: alt.comp.hardware.overclocking (More info?)

Hello to all,

About 3 weeks ago I raised the FSB-speed from my AMD 1800+ on a MSI K7T
raid-turbo mobo from 133 to 135 up to 140 and it looked like I had no
problems at all......
Well the system worked fine and my Battlefield1942 game started faster and
no instabillity of the Xp.

So far so good.....

But the last weeks I have some strange connection problem's with the local
WiFi network. Some users have, now and then, problems to connect to the
ADSL modem that can be accesed by a switch. I am directly plugged into
that switch and 8 others by a accesspoint(wifi). When the connection
problems occur I have to power-off/on the switch to get it all working. I
can also choose to power-off/on the modem the result is the same. Some
users can work others cannot.

For more than halve a year it was working fine until... yes I raised my
FSB???(not shure about same week or so).

Questions:
Is it possible that even the smallest FSB raise can generate this kind of
problems with a Network-Interface-Card ?????
in his turn, can create problems in the switch??
If so, is there a manufacturer that is not so "sensitive" in raising FSB??

Any sugestions??


Bonobo
--
DutchDareDevil

Archived from groups: alt.comp.hardware.overclocking (More info?)

> Is it possible that even the smallest FSB raise can generate this kind
> of problems with a Network-Interface-Card ?????

Yes. If the PCI bus is not locked at 100/3 MHz, then the frequency of
the signals sent to the bus will also increase. This will screw up most,
if not all PCI cards (as they do not benefit from overclocking).

This may sound a bit patronising (best to be thorough...), but have you
tried removing your overclock and seeing if the problems still remain?

> in his turn, can create problems in the switch??

Yes. Your network card might rely on the PCI frequency for timing, in
which case it might send out packets at the wrong time. This could cause
collisions and so halt the network (at least on your end).

That said, it could be that your switch is on the blink. Does anyone
else that is connected to your switch have these problems? Do they still
have them if you disconnect your computer. If not, then your cable might
be faulty (letting in interference) - try swapping it for a new (or
known good) one.

> If so, is there a manufacturer that is not so "sensitive" in raising FSB??

That, I don't know about. Your card could have chips that are barely in
tolerances and the slight frequency jump was enough to push them over
the edge (just simple bad luck). Or your manufacturer could be using
parts at the bottom end of the stability spectrum in order to make
cheaper cards.

I might test this on an old Pentium that I have - give it a 3Com
3C905-TX card (very nice - one of the best ethernet cards around) and
raise the PCI bus frequency (if I can).

Archived from groups: alt.comp.hardware.overclocking (More info?)

Hi BananaOfTheNight, it's you again

On Mon, 11 Oct 2004 14:31:56 +0100, BananaOfTheNight
<bananaofthenight@hotmail.com.nospaam.please> wrote:

>> Is it possible that even the smallest FSB raise can generate this kind
>> of problems with a Network-Interface-Card ?????
>
> Yes. If the PCI bus is not locked at 100/3 MHz, then the frequency of
> the signals sent to the bus will also increase. This will screw up most,
> if not all PCI cards (as they do not benefit from overclocking).

No I don't think its locked at 133 and I thought "what a heck it's only a
few %..."


>
> This may sound a bit patronising (best to be thorough...),

Not at all !!

> but have you tried removing your overclock and seeing if the problems
> still remain?

Tried now for 3 days without OC and it seems like the "problems" are gone
(hmmm mixed feelings..)


>> in his turn, can create problems in the switch??
>
> Yes. Your network card might rely on the PCI frequency for timing, in
> which case it might send out packets at the wrong time. This could cause
> collisions and so halt the network (at least on your end).

Good point....


>
> That said, it could be that your switch is on the blink. Does anyone
> else that is connected to your switch have these problems?


randomly YES

> Do they still have them if you disconnect your computer. If not, then
> your cable might be faulty (letting in interference) - try swapping it
> for a new (or known good) one.
>
>> If so, is there a manufacturer that is not so "sensitive" in raising
>> FSB??
>
> That, I don't know about. Your card could have chips that are barely in
> tolerances and the slight frequency jump was enough to push them over
> the edge (just simple bad luck). Or your manufacturer could be using
> parts at the bottom end of the stability spectrum in order to make
> cheaper cards.
>
> I might test this on an old Pentium that I have - give it a 3Com
> 3C905-TX card (very nice - one of the best ethernet cards around) and
> raise the PCI bus frequency (if I can).

YES YES !! torture that 3Com and get all the information out of it!!
Ask it about the RTL8139D /2C052Q1 made in Taiwan family clan and why
that clan won't work even a few more Mhz faster.

Thanks again from Bonobo !!



--
DutchDareDevil

Archived from groups: alt.comp.hardware.overclocking (More info?)

"Bonobo" <news.wanadoo.nl> wrote in message newspsfpxg8dzabsjsi@h-l.lan...
> Hi BananaOfTheNight, it's you again
>
> On Mon, 11 Oct 2004 14:31:56 +0100, BananaOfTheNight
> <bananaofthenight@hotmail.com.nospaam.please> wrote:
>
>>> Is it possible that even the smallest FSB raise can generate this kind
>>> of problems with a Network-Interface-Card ?????
>>
>> Yes. If the PCI bus is not locked at 100/3 MHz, then the frequency of
>> the signals sent to the bus will also increase. This will screw up most,
>> if not all PCI cards (as they do not benefit from overclocking).
>
> No I don't think its locked at 133 and I thought "what a heck it's only a
> few %..."
>

thats 100 / 3 (33MHz) not 133MHz. If you got that high things whould be
getting pretty screwed.

hamman

Archived from groups: alt.comp.hardware.overclocking (More info?)

> Hi BananaOfTheNight, it's you again

Yep. I'm like a bad smell - I tend to linger.

> Tried now for 3 days without OC and it seems like the "problems" are
> gone (hmmm mixed feelings..)

>> That said, it could be that your switch is on the blink. Does anyone
>> else that is connected to your switch have these problems?
>
> randomly YES

It seems like the problem might have been caused by the anomalous
frequency of your network card - it could have been pissing off the
switch. I think that you ought to investigate the switch - if you can
(try replacing it with a new one, but only if you can borrow another -
as it might give no results and there is no point in wasting money)

> YES YES !! torture that 3Com and get all the information out of it!!
> Ask it about the RTL8139D /2C052Q1 made in Taiwan family clan and why
> that clan won't work even a few more Mhz faster.

I..er....can't. The Pentium that I was going to stress it in doesn't let
me adjust the FSB speed in the BIOS, and I don't have the manual for the
motherboard to tell me what the jumpers mean. The PC that I am writing
this on won't give me the fine tuning I want - the only settings for the
PCI bus go up in increments of 12 MHz (33-->45-->???).

A quick google on that part number gave me this interesting results from
Realtek's web page:

--Supports PCI clock 16.75MHz-40MHz

(note by me - this seems like quite an overclocker's network card)

--Supports 25MHz crystal or 25MHz OSC as the internal clock source. The
frequency deviation of either crystal or OSC must be within 50 PPM

These two seem to indicate that your card doesn't care about the PCI bus
frequency and so will not send its packets out at the wrong time due to
inaccurate estimation

--Supports a 32-bit general-purpose timer with the external PCI clock as
clock source, to generate timer-interrupt

This is one part where it may slip up - but this would happen only in
your PC as it interrupts too fast?


My diagnosis - your switch is FUBAR (well, not quite BAR, but definetly
on the blink)

Archived from groups: alt.comp.hardware.overclocking (More info?)

On Mon, 11 Oct 2004 20:02:08 +0100, Hamman <none@example.com> wrote:

>
> "Bonobo" <news.wanadoo.nl> wrote in message
> newspsfpxg8dzabsjsi@h-l.lan...
>> Hi BananaOfTheNight, it's you again
>>
>> On Mon, 11 Oct 2004 14:31:56 +0100, BananaOfTheNight
>> <bananaofthenight@hotmail.com.nospaam.please> wrote:
>>
>>>> Is it possible that even the smallest FSB raise can generate this kind
>>>> of problems with a Network-Interface-Card ?????
>>>
>>> Yes. If the PCI bus is not locked at 100/3 MHz, then the frequency of
>>> the signals sent to the bus will also increase. This will screw up
>>> most,
>>> if not all PCI cards (as they do not benefit from overclocking).
>>
>> No I don't think its locked at 133 and I thought "what a heck it's only
>> a
>> few %..."
>>
>
> thats 100 / 3 (33MHz) not 133MHz. If you got that high things whould be
> getting pretty screwed.
>

sorry sorry sorry.....

sorry


Stupid monkey thought that PCI was running on FSB bus speed, that I raised
from 133 to 140.
The only thing I can find in my (MSI K7T Turbo MS-6330 mobo) manual is:

CPU Host Bus Frequency Clock (100Mhz) ???
and a jumper to switch from 100Mhz to 133Mhz FSB
In the Bios, I can raise the "clock by slight adjust - by 1 Mhz"

Can you give me some options to, for OC the system??

Bonobo

> hamman
>
>



--
DutchDareDevil

Archived from groups: alt.comp.hardware.overclocking (More info?)

On Mon, 11 Oct 2004 20:45:42 +0100, BananaOfTheNight
<bananaofthenight@hotmail.com.nospaam.please> wrote:

>> Hi BananaOfTheNight, it's you again
>
> Yep. I'm like a bad smell - I tend to linger.
>
>> Tried now for 3 days without OC and it seems like the "problems" are
>> gone (hmmm mixed feelings..)
>
>>> That said, it could be that your switch is on the blink. Does anyone
>>> else that is connected to your switch have these problems?
>> randomly YES
>
> It seems like the problem might have been caused by the anomalous
> frequency of your network card - it could have been pissing off the
> switch. I think that you ought to investigate the switch - if you can
> (try replacing it with a new one, but only if you can borrow another -
> as it might give no results and there is no point in wasting money)

I will try to "borrow" another but it will be a same type/model


>
>> YES YES !! torture that 3Com and get all the information out of it!!
>> Ask it about the RTL8139D /2C052Q1 made in Taiwan family clan and
>> why that clan won't work even a few more Mhz faster.
>
> I..er....can't. The Pentium that I was going to stress it in doesn't let
> me adjust the FSB speed in the BIOS, and I don't have the manual for the
> motherboard to tell me what the jumpers mean. The PC that I am writing
> this on won't give me the fine tuning I want - the only settings for the
> PCI bus go up in increments of 12 MHz (33-->45-->???).

From another replyer "Hamman" I understood that I mixed FSB with PCI speed.
I thougt that PCI is running on FSB speed but now I understand there bust
be ANOTHER devider on the mobo that creates the PCI bus speed
seperatly.....
In my MSI K7T-Turbo-raid MS-6330 I can't find where or how it's done....
(if it can be done manually)
Where or how can I check the PCI bus speed??
CPU-Z program gives: core speed 1527.2 multiplier x11.5 FSB
132.8 Busspeed 265.6

>
> A quick google on that part number gave me this interesting results from
> Realtek's web page:
>
> --Supports PCI clock 16.75MHz-40MHz
>
> (note by me - this seems like quite an overclocker's network card)
>
> --Supports 25MHz crystal or 25MHz OSC as the internal clock source. The
> frequency deviation of either crystal or OSC must be within 50 PPM
>
> These two seem to indicate that your card doesn't care about the PCI bus
> frequency and so will not send its packets out at the wrong time due to
> inaccurate estimation
>
> --Supports a 32-bit general-purpose timer with the external PCI clock as
> clock source, to generate timer-interrupt
>
> This is one part where it may slip up - but this would happen only in
> your PC as it interrupts too fast?
>
>

A lot of technical information that I just can understand, thanks for the
effort to help me!!


> My diagnosis - your switch is FUBAR (well, not quite BAR, but definetly
> on the blink)

I will try to swap this switch with a same type in a few days (end of the
week I think) and raise FSB again to 140 and see what happens.
BTW the switch I use is a :

E-Tech 5-NWay ports (1 uplink port??) 10/100Mbps 256kb buffer
p/n: 18-QH-H500-FH

Can you tell me the differance between a "uplink"port and a normal?? the
ADSL modem is connected to the uplink....

Regards Bonobo

--
DutchDareDevil

Archived from groups: alt.comp.hardware.overclocking (More info?)

> I will try to "borrow" another but it will be a same type/model

Excellent. Keeping your switch brand and model constant will help limit
the variables.

> From another replyer "Hamman" I understood that I mixed FSB with PCI
> speed.
> I thougt that PCI is running on FSB speed but now I understand there
> bust be ANOTHER devider on the mobo that creates the PCI bus speed
> seperatly.....
> In my MSI K7T-Turbo-raid MS-6330 I can't find where or how it's
> done.... (if it can be done manually)
> Where or how can I check the PCI bus speed??
> CPU-Z program gives: core speed 1527.2 multiplier x11.5 FSB
> 132.8 Busspeed 265.6

The PCI speed is usually set as a multiplier of the FSB - in the case of
a 133 MHz FSB, the PCI speed will be 1/4 of the FSB. I found the manual
for your motherboard online and I can't see anything in the BIOS
settings or in the jumper combinations that would let you change the PCI
multiplier or speed. I believe that it is fixed with respect to the
jumper that tells about the CPU's FSB - see page 2-3 of your manual (I
think).

I don't know of any program that would tell you the PCI bus speed.

> Can you tell me the differance between a "uplink"port and a normal??
> the ADSL modem is connected to the uplink....

I believe that an 'uplink' port on a switch is for connecting to another
router or switch. Since an ADSL modem is technically a router (it routs
ADSL traffic to the network...), then it is probably good to put it there.

Archived from groups: alt.comp.hardware.overclocking (More info?)

On Tue, 12 Oct 2004 12:14:11 +0100, BananaOfTheNight
<bananaofthenight@hotmail.com.nospaam.please> wrote:

>> I will try to "borrow" another but it will be a same type/model
>
> Excellent. Keeping your switch brand and model constant will help limit
> the variables.
>
>> From another replyer "Hamman" I understood that I mixed FSB with PCI
>> speed.
>> I thougt that PCI is running on FSB speed but now I understand there
>> bust be ANOTHER devider on the mobo that creates the PCI bus speed
>> seperatly.....
>> In my MSI K7T-Turbo-raid MS-6330 I can't find where or how it's
>> done.... (if it can be done manually)
>> Where or how can I check the PCI bus speed??
>> CPU-Z program gives: core speed 1527.2 multiplier x11.5 FSB
>> 132.8 Busspeed 265.6
>
> The PCI speed is usually set as a multiplier of the FSB - in the case of
> a 133 MHz FSB, the PCI speed will be 1/4 of the FSB. I found the manual
> for your motherboard online and I can't see anything in the BIOS
> settings or in the jumper combinations that would let you change the PCI
> multiplier or speed. I believe that it is fixed with respect to the
> jumper that tells about the CPU's FSB - see page 2-3 of your manual (I
> think).

WOW.. Where did you find that manual info?? I've been on the MSI site but
could find something like that or onfo about OC for this mobo. Yes it's
page 2-3 but doesn't tell anything about 133/4 PCI freq.

But, if it's a 133/4 =33Mhz PCI speed, than I realy cant understand why a
raise of my FSB to 140 gives such a strange effect on my network. 140/4
=only 35 Mhz i.s.o. 33Mhz....that's let's see...yes 6%
Is that realy much for a network card??

> I don't know of any program that would tell you the PCI bus speed.
>
>> Can you tell me the differance between a "uplink"port and a normal??
>> the ADSL modem is connected to the uplink....
>
> I believe that an 'uplink' port on a switch is for connecting to another
> router or switch. Since an ADSL modem is technically a router (it routs
> ADSL traffic to the network...), then it is probably good to put it
> there.

just luck...


Bonobo
--
DutchDareDevil

ps: I will let you know what the results are after switching the switch.

Archived from groups: alt.comp.hardware.overclocking (More info?)

Bonobo wrote:
> BananaOfTheNight wrote:
[...]
>>> If so, is there a manufacturer that is not so "sensitive" in raising
>>> FSB??
>>
>> That, I don't know about. Your card could have chips that are barely
>> in tolerances and the slight frequency jump was enough to push them
>> over the edge (just simple bad luck). Or your manufacturer could be
>> using parts at the bottom end of the stability spectrum in order to
>> make cheaper cards.
>>
>> I might test this on an old Pentium that I have - give it a 3Com
>> 3C905-TX card (very nice - one of the best ethernet cards around) and
>> raise the PCI bus frequency (if I can).
>
> YES YES !! torture that 3Com and get all the information out of it!!
> Ask it about the RTL8139D /2C052Q1 made in Taiwan family clan and
> why that clan won't work even a few more Mhz faster.

For what it's worth, I've got several RTL8139D-based cards (Xnet 3000RDM)
that all run happily at a 152Mhz FSB (PCI at 38MHz) and probably higher. I
can't read the top of the chips at the moment (they're inside my computer
) but a spare I have lying around has the second line 26113Q1 (with 229E
in small text on the next line).

--
Michael Brown
www.emboss.co.nz : OOS/RSI software and more
Add michael@ to emboss.co.nz - My inbox is always open

Archived from groups: alt.comp.hardware.overclocking (More info?)

On Wed, 13 Oct 2004 01:55:02 +1300, Michael Brown <see@signature.below>
wrote:

> Bonobo wrote:
>> BananaOfTheNight wrote:
> [...]
>>>> If so, is there a manufacturer that is not so "sensitive" in raising
>>>> FSB??
>>>
>>> That, I don't know about. Your card could have chips that are barely
>>> in tolerances and the slight frequency jump was enough to push them
>>> over the edge (just simple bad luck). Or your manufacturer could be
>>> using parts at the bottom end of the stability spectrum in order to
>>> make cheaper cards.
>>>
>>> I might test this on an old Pentium that I have - give it a 3Com
>>> 3C905-TX card (very nice - one of the best ethernet cards around) and
>>> raise the PCI bus frequency (if I can).
>>
>> YES YES !! torture that 3Com and get all the information out of it!!
>> Ask it about the RTL8139D /2C052Q1 made in Taiwan family clan and
>> why that clan won't work even a few more Mhz faster.
>
> For what it's worth, I've got several RTL8139D-based cards (Xnet 3000RDM)
> that all run happily at a 152Mhz FSB (PCI at 38MHz) and probably higher.
> I
> can't read the top of the chips at the moment (they're inside my computer
> ) but a spare I have lying around has the second line 26113Q1 (with
> 229E
> in small text on the next line).
>
> --
> Michael Brown
> www.emboss.co.nz : OOS/RSI software and more
> Add michael@ to emboss.co.nz - My inbox is always open
>
>

Thanks for the extra input Michael

Bonobo

--
DutchDareDevil

Archived from groups: alt.comp.hardware.overclocking (More info?)

> WOW.. Where did you find that manual info??

http://www.msi.com.tw/index.htm

Click on 'mainboard' then pick your chipset from the drop-down menu next
to the button marked 'search' (near the top of the page). Then select
the motherboard and click on the manual link on the top left of the
resulting page.

ALternatively, go to
http://www.msi.com.tw/program/prod [...] hp?UID=100
for the easy way out...

> I've been on the MSI site
> but could find something like that or onfo about OC for this mobo.
> Yes it's page 2-3 but doesn't tell anything about 133/4 PCI freq.

True. I worked that out (the PCI is always supposed to be 33 MHz and you
would have to divide your FSB by 4 to get it)

> But, if it's a 133/4 =33Mhz PCI speed, than I realy cant understand why
> a raise of my FSB to 140 gives such a strange effect on my network.
> 140/4 =only 35 Mhz i.s.o. 33Mhz....that's let's see...yes 6%
> Is that realy much for a network card??

I don't know. I suspect not, as the manufacturer's web page for your LAN
card said quite explicitly that it worked up to 40 MHz PCI frequency. I
think that your switch is about to expire, but I don't have enough data
to back that up.

Alternatively, another PCI card that you have could be complaining about
the increased frequency and sending signals out at the wrong time,
causing bus clashes with your ethernet card (hence stopping the switch
because your network card is unable to respond).

Archived from groups: alt.comp.hardware.overclocking (More info?)

On Tue, 12 Oct 2004 14:26:38 +0100, BananaOfTheNight
<bananaofthenight@hotmail.com.nospaam.please> wrote:

>> WOW.. Where did you find that manual info??
>
> http://www.msi.com.tw/index.htm
>
> Click on 'mainboard' then pick your chipset from the drop-down menu next
> to the button marked 'search' (near the top of the page). Then select
> the motherboard and click on the manual link on the top left of the
> resulting page.
>
> ALternatively, go to
> http://www.msi.com.tw/program/prod [...] hp?UID=100
> for the easy way out...
>
> > I've been on the MSI site
> > but could find something like that or onfo about OC for this mobo.
> > Yes it's page 2-3 but doesn't tell anything about 133/4 PCI freq.
>
> True. I worked that out (the PCI is always supposed to be 33 MHz and you
> would have to divide your FSB by 4 to get it)
>
>> But, if it's a 133/4 =33Mhz PCI speed, than I realy cant understand why
>> a raise of my FSB to 140 gives such a strange effect on my network.
>> 140/4 =only 35 Mhz i.s.o. 33Mhz....that's let's see...yes 6%
>> Is that realy much for a network card??
>
> I don't know. I suspect not, as the manufacturer's web page for your LAN
> card said quite explicitly that it worked up to 40 MHz PCI frequency. I
> think that your switch is about to expire, but I don't have enough data
> to back that up.
>
> Alternatively, another PCI card that you have could be complaining about
> the increased frequency and sending signals out at the wrong time,
> causing bus clashes with your ethernet card (hence stopping the switch
> because your network card is unable to respond).

The switch is only connected to -1- (my) nic, and the uplink port to the
ADSL-modem and one port to a Senao wifi accesspoint. In the weekend I will
try to swap the switch and wait for results with 140 Mhz FSB.
If that again give's problems I can change my nic (E-tech) for a nic from
Compex
(also a RTL8139D but with 39137Q1 instead of 2C052Q1) THAT WILL TEACH MY
SWITCH!! (I hope..)

Bonobo

--
DutchDareDevil