Please help an OC newbie... AMD 2800 (333fsb, Barton) proc..

[Guest]

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

My motherboard is an nForce2 based mobo, and a quick google search
affirmed the rumors that nForce2 based motherboards auto-unlock the
cpu--at least for my specific motherboard.

Without overclocking, it runs at 12.5x166

I first changed the multiplier to 13, but it had no effect. I would
restart, no problems or anything, but when I booted into linux, and
check the cpu speed... it was the same old 2087mhz. I even tried
setting the multiplier to 13.5, 14 etc... still had no affect. The
bios did save the new settings correctly...im sure. I doubble checked
each time to make sure it saved the new multiplier...and it did--it
just didnt seem to actually change it like i would expect. Why isn't
changing the multiplier working??

So then I tried changeing the FSB speed slowly... because I heard that
that's where the real overclocking power comes from on the AMD
Barton's. I changed the fsb to 168... then 169. no problems, but very
marginal speed increases (of course). When I try to up it to 170 or
higher, When my box restarts... The bios just makes a long beep,
reboots, and then goes back to being un-overclocked. I'm assuming it's
trying to tell me it didn't work or want stable at all, and as a safty
measure, reset the speed to the old stuff.

So...I'm sure I can eek out more than a few mhz of a difference from
overclocking. Why couldn't I? ... I assume I need to up some of the
voltages??

My BIOS is currently set to have the voltages on "AUTO". All 3 voltage
settings, cpu, ram, and agp are set to "AUTO."

The bios doesnt tell me what the normal voltage should be though
Can anyone tell me what they should be on a NOT overclocked box...and
then what they should be on an overclocked box. Again, I assume the
voltage increases should be proportional to the amount I overclock...
but by how much? like many volts should i try adding for every x mhz
fsb increasing? If you guys could provide me with even ballpark
figures I would appriciate it. Or at least warnings on what NOT to set
them too. I would really prefer not to fry my headware

FYI, My RAM is DDR400... and has nice heatsinks (headspeaders?) on
them...appearntly aluminum.

Thanks,

--Farrell F.

 

[Guest]

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

upgrdman wrote:
> My motherboard is an nForce2 based mobo, and a quick google search
> affirmed the rumors that nForce2 based motherboards auto-unlock the
> cpu--at least for my specific motherboard.

The MB does not unlock a locked CPU. The MB will allow all multiplers,
but only with an unlocked CPU.

> Without overclocking, it runs at 12.5x166

Yep, that's default for the 2800+.

> I first changed the multiplier to 13, but it had no effect. I would
> restart, no problems or anything, but when I booted into linux, and
> check the cpu speed... it was the same old 2087mhz. I even tried
> setting the multiplier to 13.5, 14 etc... still had no affect. The
> bios did save the new settings correctly...im sure. I doubble checked
> each time to make sure it saved the new multiplier...and it did--it
> just didnt seem to actually change it like i would expect. Why isn't
> changing the multiplier working??

Because you have a locked multiplier CPU. I haven't heard anything
recently about anyone finding a way to unlock it yet.

> So then I tried changeing the FSB speed slowly... because I heard that
> that's where the real overclocking power comes from on the AMD
> Barton's.

Correct.

> I changed the fsb to 168... then 169. no problems, but very
> marginal speed increases (of course). When I try to up it to 170 or
> higher, When my box restarts... The bios just makes a long beep,
> reboots, and then goes back to being un-overclocked. I'm assuming it's
> trying to tell me it didn't work or want stable at all, and as a safty
> measure, reset the speed to the old stuff.
>
> So...I'm sure I can eek out more than a few mhz of a difference from
> overclocking. Why couldn't I? ... I assume I need to up some of the
> voltages??

You assume correctly.

> My BIOS is currently set to have the voltages on "AUTO". All 3 voltage
> settings, cpu, ram, and agp are set to "AUTO."
>
> The bios doesnt tell me what the normal voltage should be though
> Can anyone tell me what they should be on a NOT overclocked box...and
> then what they should be on an overclocked box. Again, I assume the
> voltage increases should be proportional to the amount I overclock...
> but by how much? like many volts should i try adding for every x mhz
> fsb increasing? If you guys could provide me with even ballpark
> figures I would appriciate it. Or at least warnings on what NOT to set
> them too. I would really prefer not to fry my headware

The default core voltage for the 2800+ is 1.65V; memory, 2.5V; AGP,
1.5V, which should be left alone in most cases.

> FYI, My RAM is DDR400... and has nice heatsinks (headspeaders?) on
> them...appearntly aluminum.

The CPU core voltage should be increased by .025V increments until the
system is stable at any particular FSB speed. Since you have PC3200
memory, you shouldn't need to raise the memory voltage, but some memory
needs 2.6V to 2.8V to run properly at 200 MHz (400 DDR) or above.
In your case, increase the core voltage to 1.7V and the memory voltage
to 2.6V (check the specs for the memory to see what the default voltage
is; you may need to increase it more). Set the FSB to 170 MHz, since
that's where you lost it last time. If it's stable there, increase the
FSB again, slowly. Go in 1 or 2 MHz steps. If the system becomes
unstable, or won't boot, increase the core voltage by .025V increments
till it's stable. Just keep doing this till the system refuses to go
any higher, with a couple caveats.
Watch the CPU temperature, since it will go up as you increase speed and
core voltage. This will be a limiting factor depending on how good a
heatsink you have. Also, the core voltage should be kept below about
1.85V. Although I have seen others run the CPU at higher voltages, I
wouldn't go higher (but that's just me). Finally, the CPU is going to
cap out anywhere between 2200 and 2500 MHz; be aware that you may get
more, or less.
Once you have it running apparently stable at a higher speed, run
Prime95 to test actual stability. If you get any errors in Prime95,
back the FSB speed down slowly till you can run it for hours with no
errors. Fine tune FSB and core voltage till you have the system running
Prime95 for at least 6 hours with no errors. Keep an eye on the CPU
temp as you go.
Prime95 is a free program available here:
http://www.mersenne.org/freesoft.htm

> Thanks,
>
> --Farrell F.

Good luck.

 

[Guest]

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

ICee wrote:
> upgrdman wrote:
>> My motherboard is an nForce2 based mobo, and a quick google search
>> affirmed the rumors that nForce2 based motherboards auto-unlock the
>> cpu--at least for my specific motherboard.
>
> The MB does not unlock a locked CPU. The MB will allow all
> multiplers, but only with an unlocked CPU.

Sorry, I was thinking you had an Abit NF7; I see nForce2 and immediately
think of the NF7. In any case, any Barton CPU manufactured after about
week 39 of 2003 has a locked multiplier, and so far no one has been able
to find a way to unlock it. The NF7 series will allow all multipliers
*if* the CPU is unlocked. Don't know about your MB since you didn't
mention what it is, but in any case, whatever it is, it can't unlock one
of these "superlocked" CPU's.

>> Without overclocking, it runs at 12.5x166
>
> Yep, that's default for the 2800+.
>
>> I first changed the multiplier to 13, but it had no effect. I would
>> restart, no problems or anything, but when I booted into linux, and
>> check the cpu speed... it was the same old 2087mhz. I even tried
>> setting the multiplier to 13.5, 14 etc... still had no affect. The
>> bios did save the new settings correctly...im sure. I doubble checked
>> each time to make sure it saved the new multiplier...and it did--it
>> just didnt seem to actually change it like i would expect. Why isn't
>> changing the multiplier working??
>
> Because you have a locked multiplier CPU. I haven't heard anything
> recently about anyone finding a way to unlock it yet.
>
>> So then I tried changeing the FSB speed slowly... because I heard
>> that that's where the real overclocking power comes from on the AMD
>> Barton's.
>
> Correct.
>
>> I changed the fsb to 168... then 169. no problems, but very
>> marginal speed increases (of course). When I try to up it to 170 or
>> higher, When my box restarts... The bios just makes a long beep,
>> reboots, and then goes back to being un-overclocked. I'm assuming
>> it's trying to tell me it didn't work or want stable at all, and as
>> a safty measure, reset the speed to the old stuff.
>>
>> So...I'm sure I can eek out more than a few mhz of a difference from
>> overclocking. Why couldn't I? ... I assume I need to up some of the
>> voltages??
>
> You assume correctly.
>
>> My BIOS is currently set to have the voltages on "AUTO". All 3
>> voltage settings, cpu, ram, and agp are set to "AUTO."
>>
>> The bios doesnt tell me what the normal voltage should be though
>> Can anyone tell me what they should be on a NOT overclocked box...and
>> then what they should be on an overclocked box. Again, I assume the
>> voltage increases should be proportional to the amount I overclock...
>> but by how much? like many volts should i try adding for every x mhz
>> fsb increasing? If you guys could provide me with even ballpark
>> figures I would appriciate it. Or at least warnings on what NOT to
>> set them too. I would really prefer not to fry my headware
>
> The default core voltage for the 2800+ is 1.65V; memory, 2.5V; AGP,
> 1.5V, which should be left alone in most cases.
>
>> FYI, My RAM is DDR400... and has nice heatsinks (headspeaders?) on
>> them...appearntly aluminum.
>
> The CPU core voltage should be increased by .025V increments until the
> system is stable at any particular FSB speed. Since you have PC3200
> memory, you shouldn't need to raise the memory voltage, but some
> memory needs 2.6V to 2.8V to run properly at 200 MHz (400 DDR) or
> above.
> In your case, increase the core voltage to 1.7V and the memory voltage
> to 2.6V (check the specs for the memory to see what the default
> voltage is; you may need to increase it more). Set the FSB to 170
> MHz, since that's where you lost it last time. If it's stable there,
> increase the FSB again, slowly. Go in 1 or 2 MHz steps. If the
> system becomes unstable, or won't boot, increase the core voltage by
> .025V increments till it's stable. Just keep doing this till the
> system refuses to go any higher, with a couple caveats.
> Watch the CPU temperature, since it will go up as you increase speed
> and core voltage. This will be a limiting factor depending on how
> good a heatsink you have. Also, the core voltage should be kept
> below about
> 1.85V. Although I have seen others run the CPU at higher voltages, I
> wouldn't go higher (but that's just me). Finally, the CPU is going to
> cap out anywhere between 2200 and 2500 MHz; be aware that you may get
> more, or less.
> Once you have it running apparently stable at a higher speed, run
> Prime95 to test actual stability. If you get any errors in Prime95,
> back the FSB speed down slowly till you can run it for hours with no
> errors. Fine tune FSB and core voltage till you have the system
> running Prime95 for at least 6 hours with no errors. Keep an eye on
> the CPU temp as you go.
> Prime95 is a free program available here:
> http://www.mersenne.org/freesoft.htm
>
>> Thanks,
>>
>> --Farrell F.
>
> Good luck.

 

[Guest]

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

On Fri, 30 Apr 2004 21:00:11 -0700, upgrdman wrote:

> My motherboard is an nForce2 based mobo, and a quick google search
> affirmed the rumors that nForce2 based motherboards auto-unlock the
> cpu--at least for my specific motherboard.
>
There's no Mb made that can unlock a multiplier locked cpu.

> Without overclocking, it runs at 12.5x166
>
> I first changed the multiplier to 13, but it had no effect. I would
> restart, no problems or anything, but when I booted into linux, and
> check the cpu speed... it was the same old 2087mhz. I even tried
> setting the multiplier to 13.5, 14 etc... still had no affect. The
> bios did save the new settings correctly...im sure. I doubble checked
> each time to make sure it saved the new multiplier...and it did--it
> just didnt seem to actually change it like i would expect. Why isn't
> changing the multiplier working??
>
1. The cpu multiplier is locked.
2. Some MB's don't allow crosing the 12.5-13 multiplier boundary because
they don't really controll the 5th multiplier bit.

> So then I tried changeing the FSB speed slowly... because I heard that
> that's where the real overclocking power comes from on the AMD Barton's.

Adds a little performance boost that way, but nothing great.

> I changed the fsb to 168... then 169. no problems, but very marginal
> speed increases (of course). When I try to up it to 170 or higher, When
> my box restarts... The bios just makes a long beep, reboots, and then
> goes back to being un-overclocked. I'm assuming it's trying to tell me
> it didn't work or want stable at all, and as a safty measure, reset the
> speed to the old stuff.
>
To be successful overclocking a FSB you need to either be lucky, or know
what effect it has on the rest of your system.

> So...I'm sure I can eek out more than a few mhz of a difference from
> overclocking. Why couldn't I? ... I assume I need to up some of the
> voltages??
>
Maybe, but it could be lots of things. I doubt it's vcore with just that
little overclock.

> My BIOS is currently set to have the voltages on "AUTO". All 3 voltage
> settings, cpu, ram, and agp are set to "AUTO."
>
> The bios doesnt tell me what the normal voltage should be though Can
> anyone tell me what they should be on a NOT overclocked box...and then
> what they should be on an overclocked box. Again, I assume the voltage
> increases should be proportional to the amount I overclock... but by how
> much? like many volts should i try adding for every x mhz fsb
> increasing? If you guys could provide me with even ballpark figures I
> would appriciate it. Or at least warnings on what NOT to set them too. I
> would really prefer not to fry my headware
>
> FYI, My RAM is DDR400... and has nice heatsinks (headspeaders?) on
> them...appearntly aluminum.
>
I could type for an hour and probably not cover it all. Experiment.
Or do some googling.

--
Abit KT7-Raid (KT133) Tbred B core CPU @2400MHz (24x100FSB)
http://mysite.verizon.net/res0exft/cpu.htm

 

[Guest]

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

"upgrdman" <upgrdman@mindspring.com> wrote in message
news:bf4a6c76.0404302000.146773b3@posting.google.com...
> My motherboard is an nForce2 based mobo, and a quick google search
> affirmed the rumors that nForce2 based motherboards auto-unlock the
> cpu--at least for my specific motherboard.

Eh.. It is wrong sometimes..

> Without overclocking, it runs at 12.5x166

yea, that one is correct

> I first changed the multiplier to 13, but it had no effect. I would
> restart, no problems or anything, but when I booted into linux, and
> check the cpu speed... it was the same old 2087mhz. I even tried
> setting the multiplier to 13.5, 14 etc... still had no affect. The
> bios did save the new settings correctly...im sure. I doubble checked
> each time to make sure it saved the new multiplier...and it did--it
> just didnt seem to actually change it like i would expect. Why isn't
> changing the multiplier working??

Ask google again.. hint, search for LOCKED BARTON and see what it shows ;P..
If you dont want to search then let me just say, it's a barton that's
lock'd..

> So then I tried changeing the FSB speed slowly... because I heard that
> that's where the real overclocking power comes from on the AMD
> Barton's. I changed the fsb to 168... then 169. no problems, but very
> marginal speed increases (of course). When I try to up it to 170 or
> higher, When my box restarts... The bios just makes a long beep,
> reboots, and then goes back to being un-overclocked. I'm assuming it's
> trying to tell me it didn't work or want stable at all, and as a safty
> measure, reset the speed to the old stuff.

166 to 170 isnt a big jump, how high did you go? I would expect you to
reach 176 PROBABLY but 180+ might not be possible
Granted, still have the ram to think about (read ur mobo book)

> So...I'm sure I can eek out more than a few mhz of a difference from
> overclocking. Why couldn't I? ... I assume I need to up some of the
> voltages??

You shouldnt tinker with voltages, Learn what does what and when what
happens to what and why and when it will do what it does to what it does it
to when and how and then
change voltages (or you will have a dead processor/ram (maybee more) through
user error)

> My BIOS is currently set to have the voltages on "AUTO". All 3 voltage
> settings, cpu, ram, and agp are set to "AUTO."

daz good

> The bios doesnt tell me what the normal voltage should be though
> Can anyone tell me what they should be on a NOT overclocked box...and
> then what they should be on an overclocked box. Again, I assume the
> voltage increases should be proportional to the amount I overclock...
> but by how much? like many volts should i try adding for every x mhz
> fsb increasing? If you guys could provide me with even ballpark
> figures I would appriciate it. Or at least warnings on what NOT to set
> them too. I would really prefer not to fry my headware

most bios's have a dig page and it will tell you what it's running at ..
every chip is different (well not really true, but they arn't all the same)

> FYI, My RAM is DDR400... and has nice heatsinks (headspeaders?) on
> them...appearntly aluminum.

Is it those lil ugly ones that look like the fighters out of battlestar
galatica, I wanted some of them but I didnt figgure it would help, still
they look "interesting"