Sign in with
Sign up | Sign in
Your question

Stuck at 3.8Ghz on P5W DH using TEC...

Tags:
  • Motherboards
  • Overclocking
Last response: in Overclocking
Share
October 3, 2006 8:17:44 PM

Following much advice and some of my own experiments -- my TEC solution only gained me 50 Mhz from 3.75 (dual loop watercooled) to 3.8 Ghz (dual loop using TEC).

I've read most that can be read and filtered out the fact from fiction and followed "the guide" and even went beyond the guide. I've added two 40mm fans to the chipset (removed the Asus heat sink covers, thermal pasted, etc. etc.) and still can't move beyond 3.8 Ghz reliabily (prime95). I can get it to post at 4.1 Ghz but is not stable. Can't get it to post at anything beyond 4.1.

The only benefit TEC has provided is that I was able to drop my vCore considerably. But any attempt to push the FSB beyond 1520 Mhz or so will result in stability problems and/or not even post.

The voltage fluctuation is now excellent (rarely moves and never more than .01 v). CPU temp (according to Asus onboard) under Prim95 loads is about 33-48 degrees F pending which tests are being running. When idle I get error in Asus Probe as apparently it can't handle sub zero readings.

Anything above 4.1 Ghz I can't even get the system to post. Question is, I'm not sure what to look at -- is it the RAM (thought I had some pretty good Patriot RAM rated to PC8000)? is it the motherboard? is it the CPU? Tried all combination of multipliers but have stayed at 1:1 for DDR2.

Locking the PCI bus at 33.33 Mhz didn't help either. Maybe I just have a marginal motherboard - luck of the draw?

Any suggestions?

Thanks, Rob.

P.S. on this Tech setup I was expecting to hit atleast 4.1-4.2 Ghz as I know others are at 4.4+ Ghz.

More about : stuck 8ghz p5w tec

a b V Motherboard
October 3, 2006 8:31:52 PM

Quote:
Following much advice and some of my own experiments -- my TEC solution only gained me 50 Mhz from 3.75 (dual loop watercooled) to 3.8 Ghz (dual loop using TEC).

I've read most that can be read and filtered out the fact from fiction and followed "the guide" and even went beyond the guide. I've added two 40mm fans to the chipset (removed the Asus heat sink covers, thermal pasted, etc. etc.) and still can't move beyond 3.8 Ghz reliabily (prime95). I can get it to post at 4.1 Ghz but is not stable. Can't get it to post at anything beyond 4.1.

The only benefit TEC has provided is that I was able to drop my vCore considerably. But any attempt to push the FSB beyond 1520 Mhz or so will result in stability problems and/or not even post.

The voltage fluctuation is now excellent (rarely moves and never more than .01 v). CPU temp (according to Asus onboard) under Prim95 loads is about 33-48 degrees F pending which tests are being running. When idle I get error in Asus Probe as apparently it can't handle sub zero readings.

Anything above 4.1 Ghz I can't even get the system to post. Question is, I'm not sure what to look at -- is it the RAM (thought I had some pretty good Patriot RAM rated to PC8000)? is it the motherboard? is it the CPU? Tried all combination of multipliers but have stayed at 1:1 for DDR2.

Locking the PCI bus at 33.33 Mhz didn't help either. Maybe I just have a marginal motherboard - luck of the draw?

Any suggestions?

Thanks, Rob.

P.S. on this Tech setup I was expecting to hit atleast 4.1-4.2 Ghz as I know others are at 4.4+ Ghz.


You mentioned the ASUS MB you are using, but I don't see where you mentioned the processor you are using. Try increasing the PCI bus speed to 110, then to 120.
October 3, 2006 8:43:57 PM

Sorry my error: Core 2 X6800.

Haven't tried increasing PCI bus speed -- most have recommend setting it to 100, but heck, I'll try it and see.

I was thinking of using something other than 1:1 for DDR ... as it stands now at 3.8 Ghz the post reports the RAM as PC5700. The Patriot RAM is listed to run at 2.2v for PC8000, increasing to 2.3v didn't help. I have DRAM timing set to SPD as that was recommend for this MB.

Thanks, Rob.
Related resources
a b V Motherboard
October 3, 2006 8:48:04 PM

Quote:
Sorry my error: Core 2 X6800.

Haven't tried increasing PCI bus speed -- most have recommend setting it to 100, but heck, I'll try it and see.

I was thinking of using something other than 1:1 for DDR ... as it stands now at 3.8 Ghz the post reports the RAM as PC5700. The Patriot RAM is listed to run at 2.2v, increasing to 2.3v didn't help.

Thanks, Rob.


I would try increasing PCI Bus Speed to 110 then to 120. If You have it locked at 100 while the rest of your system is being clocked faster and faster. My Pentium D 805 Smithfield on a 945G ASUS board will do no better than a 10% overclock until I set the PCI BUS SPeed at 118. I get a 1 ghz. OC there. 2.6 to 3.6 ghz.
October 3, 2006 10:08:42 PM

Maybe try setting the mem timings really loose, instead of useing the spd settings?

other than that, tinker around and maybe you will have luck. But if nothing works then its most likely you didn't get the best motherboard off that production line.
October 4, 2006 3:55:16 AM

What kind of TEC water block are you using (module rating, voltage, company)? I am not sure of the differences between the 6800 and my 955 Presler but I am able to OC 4+ easily (adjusting the FSB, NB Strap and various voltages) on my Abit AW8D. I intend on going further once I've set up my second loop for the voltage regulators around the CPU.

What are your two cooling loops? What are you using in them and how are they arranged? Whenever I've used an ASUS board with TEC cooling the temp readings would hit 0 degrees and then go down, incrimentally, -255, -254, -253, etc.
October 4, 2006 5:57:16 AM

Swiftech everything, pumps, radiators, TEC block, etc. -- separate PS for the TEC, PC Power & Cooling 1000 Watt PSU for other components. Two 1/2" tubing loops into two dual 120mm radiators.

One cooling loop to CPU
One cooling loop to GPU

Radiators are actually housed on the outside of the case to keep things cooler in the case.

Ok, I've managed to get it running at 3.9Ghz without any Prime95 errors and 3DMark05 runs fine. What I've had to do to get this stable is:

Memory Voltage: 2.25v
CPU core Voltage: 1.5125v (Max)
FSB Termination Voltage 1.50v (Max)
MCH Chipset voltage: 1.65v (Max)
ICH Chipset voltage: 1.20v (Max)

As I increased CPU core voltage I got fewer and fewer errors in Prime95 until I reached the max vCore where Prime95 ran for an hour or so with no errors. Not sure what this tells me other than the CPU appears to be hungry for Power.

Rob.
October 4, 2006 6:20:50 AM

Is the secondary PSU (the one for the TEC) a Meanwell?
October 4, 2006 3:10:48 PM

Yes, Meanwell 320W PS

My assumption (and this could be wrong) is because I progressively got more and more stable (further along in Prime95) as I increased CPU core voltage, I would have to assume the limit is my CPU and/or voltage my Asus MB can provide.

Don't know how long my CPU can last at 1.512v - it never gets hotter than 48 degrees F and idles below freezing but I'm not sure if that will help or hurt it's life span. As I understand it, the problem (more rumors) is the on/off temp change that causes component flex which can weaken connections -- so it would actually be better for the MB/CPU if I left the PC on 24/7 (since TEC operates at extreme temp ranges). Again, some rumors and hints I picked up but without any real validation (the numbers).

Tried out the new verstion of ATITool .25 Beta 15 Pre3 on my X1950XTX and that was a complete failure -- back to RivaTuner as it seems to produce results.
October 4, 2006 3:30:28 PM

That's good to know -- I rarely keep a processor longer than 1 yr.

Maybe ASUS need to go with a 12 phase design (rather than 8) -- I must admit, putting the acrylic conformal coating on my MB and the grease left me wonder if my TEC would ever boot again (was about ready to buy a new MB).
October 4, 2006 4:23:51 PM

Quote:
Meanwhile overclockers have found 1.725V is around the point at which the processor starts to breakdown for those C2D after 2days of use. i.e. dying rapidly with lesser potential by hours passed


I haven't seen those results yet (I can't believe work is getting in the way of hobby) but I'm curious - are these CPUs run subambient at 1.725V?
October 5, 2006 4:41:19 AM

Quote:
The rule with modern processor overclocking(inlcuding extreme) is temperature has never been a factor in damage.


I've seen you post that before. The thing is, all of the mobility studies I've been involved in studied both temp and voltage effects. I've seen operating temp have an effect at higher volts. Obviously, an LN2 temp CPU will die fast at high enough V.
October 5, 2006 4:59:43 AM

Quote:
Obviously when you're using above 1.625V/1.600V on C2D, you'd need more than plain watercooling just to keep the rig running stable. i.e. again, temperature isn't a problem here
But theoratically what you said is true, but I don't give rats about that because it's not possible or practical.

Remember what I told Jack in the CPU forum, anything I said applies to practical real-life(usuable) situation only.


The practicality of my work involved devices that could only have marginal air cooling in a high ambient temp environment. So we're not talking enthusiast OC'ed rigs but developers still want to achieve the best performance within minimum lifetime requirements. So it's very practical - moreso than an extreme OC in many ways because all of the device build parameters are looked at in accelerated aging conditions. Meaning that these studies help process engineers design and manufacture better chips. What is more practical than that?
October 5, 2006 4:02:09 PM

I know a guy, who knows a guy, who works with custom BIOSs (specifically for Asus) -- will dig to see what I can get.

Wish I had the same source for someone that worked on custom ATI BIOS, I feel like my X1950XTX is being under utilized. I know the DDR4 memory is operating well below it's potential.
October 6, 2006 2:13:36 AM

Quote:
Just a statistics I've gathered for X6800 which you might be intrest you.
Watercooling w/o TEC (~38C) at 1.500V: 3.84Ghz
Phasechange (~ -50C) at 1.500V: 4.21Ghz
Cascade (~ -105C) at 1.500V: 4.64Ghz

I'm guessing for your TEC it's around -20C at the wattage(speed+voltage+unknown amp)) from processor you're putting through it now so what you're reaching with that voltage sounds about right. A little bit more tinkering(experiment...) should get it around 100Mhz higher at the same voltage you're using now.


Also somethings worth noting... not all Core 2's clock the same... that's a given. I'm able to achieve 4.45GHz (405x11) using my pelt setup. Then again my whole case was custom modded to ensure premium performance out of my TEC units.

If you go to Xtremesystems they have a thread going on regarding 4GHz+ Core 2 overclocks. I posted my results there but you'll also find MANY users breaking 4GHz with simple air cooling on E6600 and E6700 models.

I don't think your CPU is the limiting factor.. perhaps it's the motherboard. Have you tried tinkering with the Multiplier like raising it in order to lower the FSB?
!