Advice on overclocking a QX6700

Hi all,
I have the following setup:

M/b: Asus P5B deluxe, 1066MHz FSB
CPU: Intel QX6700
RAM: 2 x Corsair XMS2 DDR2-800 (2Gb)
PSU: Tagan TG580-U15 580W
Video: 2 x Asus GeForce 7300GS (running three LCD's)
Cooling: Scythe Ninja Plus (B), actively cooled with a 1200rpm 120mm fan.
1200rpm 120mm case fan (sucking from front of case).

Temperatures range from 35deg (at idling) to 65deg (pushing all 4 cores
quite hard). On average it's sitting between 40 and 50deg.

It's running at the stock 2.66GHz (266 x 10).

I'd like to overclock this to around 3GHz with the current cooling system.
Caveat: I've never overclocked a computer before.

The P5B geive a lot of control over overcocking settings. I can select
between 6 and 12 for the multiplier. I bumped this up to 11 to get 2.9GHz
and things seemed to work find, although temps ran about 4deg hotter. It
didn't see any faster though (slower on some tests).

Does anyone have any experience with this MB/CPU combination and have
overclocking results they'd like to share?

Should I start by dropping the multiplier and slowly increasing the FSB
until things get unstable? All sites I've seen suggest this starting with
method but it seems quite tedious and time consuming.

Thanks,
Allistar.

Allistar wrote:
Should I start by dropping the multiplier and slowly
increasing the FSB until things get unstable? All sites I've
seen suggest this starting with method but it seems quite
tedious and time consuming.

Well, since your multiplier seems to be unlocked, you could
just bump it up to eleven and be done. If you are going to be
fully utilizing all four cores, you should install a faster fan.

Allistar wrote:
Hi all,
I have the following setup:

M/b: Asus P5B deluxe, 1066MHz FSB
CPU: Intel QX6700
RAM: 2 x Corsair XMS2 DDR2-800 (2Gb)
PSU: Tagan TG580-U15 580W
Video: 2 x Asus GeForce 7300GS (running three LCD's)
Cooling: Scythe Ninja Plus (B), actively cooled with a 1200rpm 120mm fan.
1200rpm 120mm case fan (sucking from front of case).

Temperatures range from 35deg (at idling) to 65deg (pushing all 4 cores
quite hard). On average it's sitting between 40 and 50deg.

It's running at the stock 2.66GHz (266 x 10).

I'd like to overclock this to around 3GHz with the current cooling system.
Caveat: I've never overclocked a computer before.

The P5B geive a lot of control over overcocking settings. I can select
between 6 and 12 for the multiplier. I bumped this up to 11 to get 2.9GHz
and things seemed to work find, although temps ran about 4deg hotter. It
didn't see any faster though (slower on some tests).

Does anyone have any experience with this MB/CPU combination and have
overclocking results they'd like to share?

Should I start by dropping the multiplier and slowly increasing the FSB
until things get unstable? All sites I've seen suggest this starting with
method but it seems quite tedious and time consuming.

Thanks,
Allistar.


There are some private forums that have threads, either on your
QX6700, or on your motherboard. Some of the forums allow search
of all the forums, then when you are inside a thread, you can
search for a particular word within the thread. This site has
pretty good facilities for guests, at least for searching for
things:

"Asus P5B-Deluxe; Problems & Fixes"
http://www.xtremesystems.org/forums/...d.php?t=112648

(Main search page)
http://www.xtremesystems.org/forums/search.php

I would at least get a copy of CPUZ from cpuid.com and verify
that the core clock is what you think it is. Apparently,
Asus BIOS for 965 boards, seem to vary from release to release,
in their treatment of the multiplier. On some Asus boards, only
a few versions of BIOS have unlocked multiplier for QX6700.

Paul

Fishface wrote:

Allistar wrote:
Should I start by dropping the multiplier and slowly
increasing the FSB until things get unstable? All sites I've
seen suggest this starting with method but it seems quite
tedious and time consuming.

Well, since your multiplier seems to be unlocked, you could
just bump it up to eleven and be done. If you are going to be
fully utilizing all four cores, you should install a faster fan.

I have read that increasing the FSB speed will give better performance that
upping the multiplier as it removes a bottle neck - is there any truth in
that?

I'll find some good benchmarking software (perhaps sprime) and come up with
quantitative before and after results).

Allistar.

Paul wrote:

Allistar wrote:
Hi all,
I have the following setup:

M/b: Asus P5B deluxe, 1066MHz FSB
CPU: Intel QX6700
RAM: 2 x Corsair XMS2 DDR2-800 (2Gb)
PSU: Tagan TG580-U15 580W
Video: 2 x Asus GeForce 7300GS (running three LCD's)
Cooling: Scythe Ninja Plus (B), actively cooled with a 1200rpm 120mm fan.
1200rpm 120mm case fan (sucking from front of case).

Temperatures range from 35deg (at idling) to 65deg (pushing all 4 cores
quite hard). On average it's sitting between 40 and 50deg.

It's running at the stock 2.66GHz (266 x 10).

I'd like to overclock this to around 3GHz with the current cooling
system. Caveat: I've never overclocked a computer before.

The P5B geive a lot of control over overcocking settings. I can select
between 6 and 12 for the multiplier. I bumped this up to 11 to get 2.9GHz
and things seemed to work find, although temps ran about 4deg hotter. It
didn't see any faster though (slower on some tests).

Does anyone have any experience with this MB/CPU combination and have
overclocking results they'd like to share?

Should I start by dropping the multiplier and slowly increasing the FSB
until things get unstable? All sites I've seen suggest this starting with
method but it seems quite tedious and time consuming.

Thanks,
Allistar.


There are some private forums that have threads, either on your
QX6700, or on your motherboard. Some of the forums allow search
of all the forums, then when you are inside a thread, you can
search for a particular word within the thread. This site has
pretty good facilities for guests, at least for searching for
things:

"Asus P5B-Deluxe; Problems & Fixes"
http://www.xtremesystems.org/forums/...d.php?t=112648

(Main search page)
http://www.xtremesystems.org/forums/search.php

Thanks, looks like a lot of good information there.

I would at least get a copy of CPUZ from cpuid.com and verify
that the core clock is what you think it is. Apparently,
Asus BIOS for 965 boards, seem to vary from release to release,
in their treatment of the multiplier. On some Asus boards, only
a few versions of BIOS have unlocked multiplier for QX6700.

From what I can tell CPUZ if fow Windows only, I run Linux (specifically
Gentoo). Both my BIOS and software report the processor speed as the same,
I can't find a tool for Linux to report the FSB speed or multiplier. I'll
keep looking and see what I can find.

My board allows me to select a multiplier between 6 and 12, and apparently I
can enter the FSB speed in 1MHz increments (although I haven't yet tried).

Paul

Thanks,
Allistar.

I have read that increasing the FSB speed will give better
performance that upping the multiplier as it removes a bottle
neck - is there any truth in that?

I'll find some good benchmarking software (perhaps sprime) and
come up with quantitative before and after results).

I have read that with the advent of multi-channel memory controllers
and larger cache sizes, today's processors are not as memory
bandwidth starved as they once were-- but I'm sure it wouldn't hurt,
especially with four cores! Since you did pay for the ddr2-400,
380 x 8 should get you 3.04 GHz with minimal tedium, as long as
you can keep the beast cool. Intel specifies the Thermal Design
Power at 130 w on its new 2.93 GHz QX6800. This is under
"normal operating conditions," whatever that means to Intel...
Watch your voltage drop-- you can Google for "vdroop" and "p5b."

Fishface wrote:

I have read that increasing the FSB speed will give better
performance that upping the multiplier as it removes a bottle
neck - is there any truth in that?

I'll find some good benchmarking software (perhaps sprime) and
come up with quantitative before and after results).

I have read that with the advent of multi-channel memory controllers
and larger cache sizes, today's processors are not as memory
bandwidth starved as they once were-- but I'm sure it wouldn't hurt,
especially with four cores! Since you did pay for the ddr2-400,
380 x 8 should get you 3.04 GHz with minimal tedium, as long as
you can keep the beast cool. Intel specifies the Thermal Design
Power at 130 w on its new 2.93 GHz QX6800. This is under
"normal operating conditions," whatever that means to Intel...
Watch your voltage drop-- you can Google for "vdroop" and "p5b."

Thanks for the advice.

I tried 10x @ 320MHz with DRAM set to 800MHz (although lshw report it as
being 667MHz). It ran fine but temperatures were too high for my liking
(idle at 59deg, moderate load at about 66deg). This configuration saw a 20%
improvement in sprime benchmarks.

I then dropped it down to 10x @ 300MHz with DRAM set to 900MHz (lshw
reports it as being 800MHz). It's now running stably with acceptable
temperatures (idling at 48deg, moderate load at 58deg). This configuration
sees a 13% improvement in sprime benchmarks.

The only voltage I have modified is to set RAM to 2.1V (the manufacturers
rated voltage).

I'll see how this goes and will run an sprime torture test on it overnight
next week some time.

Allistar.

"Allistar" wrote in message
...
Fishface wrote:

I have read that increasing the FSB speed will give better
performance that upping the multiplier as it removes a bottle
neck - is there any truth in that?

I'll find some good benchmarking software (perhaps sprime) and
come up with quantitative before and after results).

I have read that with the advent of multi-channel memory controllers
and larger cache sizes, today's processors are not as memory
bandwidth starved as they once were-- but I'm sure it wouldn't hurt,
especially with four cores! Since you did pay for the ddr2-400,
380 x 8 should get you 3.04 GHz with minimal tedium, as long as
you can keep the beast cool. Intel specifies the Thermal Design
Power at 130 w on its new 2.93 GHz QX6800. This is under
"normal operating conditions," whatever that means to Intel...
Watch your voltage drop-- you can Google for "vdroop" and "p5b."

Thanks for the advice.

I tried 10x @ 320MHz with DRAM set to 800MHz (although lshw report it as
being 667MHz). It ran fine but temperatures were too high for my liking
(idle at 59deg, moderate load at about 66deg). This configuration saw a
20%
improvement in sprime benchmarks.

I then dropped it down to 10x @ 300MHz with DRAM set to 900MHz (lshw
reports it as being 800MHz). It's now running stably with acceptable
temperatures (idling at 48deg, moderate load at 58deg). This configuration
sees a 13% improvement in sprime benchmarks.

The only voltage I have modified is to set RAM to 2.1V (the manufacturers
rated voltage).

I'll see how this goes and will run an sprime torture test on it overnight
next week some time.

Allistar.

You might keep an eye on your max temp under a heavy load. 58c is just fine
and well under Intel's specs. If you get to the upper 60s you might start
seeing some problems. I don't think it will start to throttle down until
70-75c IIRC, but when overclocking, stability will suffer at slightly lower
temps than when running at stock speeds. If your temps stay in the range you
have now I think you should be fine.

Ed

On Apr 12, 6:44 pm, Allistar wrote:
Fishface wrote:
I have read that increasing the FSB speed will give better
performance that upping the multiplier as it removes a bottle
neck - is there any truth in that?

I'll find some good benchmarking software (perhaps sprime) and
come up with quantitative before and after results).

I have read that with the advent of multi-channel memory controllers
and larger cache sizes, today's processors are not as memory
bandwidth starved as they once were-- but I'm sure it wouldn't hurt,
especially with four cores! Since you did pay for the ddr2-400,
380 x 8 should get you 3.04 GHz with minimal tedium, as long as
you can keep the beast cool. Intel specifies the Thermal Design
Power at 130 w on its new 2.93 GHz QX6800. This is under
"normal operating conditions," whatever that means to Intel...
Watch your voltage drop-- you can Google for "vdroop" and "p5b."

Thanks for the advice.

I tried 10x @ 320MHz with DRAM set to 800MHz (although lshw report it as
being 667MHz). It ran fine but temperatures were too high for my liking
(idle at 59deg, moderate load at about 66deg). This configuration saw a 20%
improvement in sprime benchmarks.

I then dropped it down to 10x @ 300MHz with DRAM set to 900MHz (lshw
reports it as being 800MHz). It's now running stably with acceptable
temperatures (idling at 48deg, moderate load at 58deg). This configuration
sees a 13% improvement in sprime benchmarks.

The only voltage I have modified is to set RAM to 2.1V (the manufacturers
rated voltage).

I'll see how this goes and will run an sprime torture test on it overnight
next week some time.

Allistar.- Hide quoted text -

- Show quoted text -

Hey folks,

If I went with and FSB @ 300 and a Multiplier at 11 this should get me
to 3.3Ghz which is well in the stable factor from what I read. Still,
can anyone tell me if I need to up the voltage? If so, what do I up
and how. Many thanks for the help.

wrote in message
ups.com...
On Apr 12, 6:44 pm, Allistar wrote:
Fishface wrote:
I have read that increasing the FSB speed will give better
performance that upping the multiplier as it removes a bottle
neck - is there any truth in that?

I'll find some good benchmarking software (perhaps sprime) and
come up with quantitative before and after results).

I have read that with the advent of multi-channel memory controllers
and larger cache sizes, today's processors are not as memory
bandwidth starved as they once were-- but I'm sure it wouldn't hurt,
especially with four cores! Since you did pay for the ddr2-400,
380 x 8 should get you 3.04 GHz with minimal tedium, as long as
you can keep the beast cool. Intel specifies the Thermal Design
Power at 130 w on its new 2.93 GHz QX6800. This is under
"normal operating conditions," whatever that means to Intel...
Watch your voltage drop-- you can Google for "vdroop" and "p5b."

Thanks for the advice.

I tried 10x @ 320MHz with DRAM set to 800MHz (although lshw report it as
being 667MHz). It ran fine but temperatures were too high for my liking
(idle at 59deg, moderate load at about 66deg). This configuration saw a
20%
improvement in sprime benchmarks.

I then dropped it down to 10x @ 300MHz with DRAM set to 900MHz (lshw
reports it as being 800MHz). It's now running stably with acceptable
temperatures (idling at 48deg, moderate load at 58deg). This
configuration
sees a 13% improvement in sprime benchmarks.

The only voltage I have modified is to set RAM to 2.1V (the manufacturers
rated voltage).

I'll see how this goes and will run an sprime torture test on it
overnight
next week some time.

Allistar.- Hide quoted text -

- Show quoted text -

Hey folks,

If I went with and FSB @ 300 and a Multiplier at 11 this should get me
to 3.3Ghz which is well in the stable factor from what I read. Still,
can anyone tell me if I need to up the voltage? If so, what do I up
and how. Many thanks for the help.

I wouldn't raise the core voltage until I found where the chip begins to
show instability. Then it would be in VERY small increments.

Ed