HOWTO: Overclock C2Q (Quads) and C2D (Duals) - A Guide v1.7

[graysky]

As far as I can tell, C1E doesn't do anything to my voltage. I have manually specified a value... I wonder if C1E only functions in "auto" voltage mode. I'm too lazy to do the experiment and see

 

[arthur2007]

Thanks Wgoldfarb, Woodbutcher, and Graysky for your help/comment. I tried to juggling many task at the same time (WOW, IE, Firefox, Burning CD ) and the CPU-z keep jumping indicator between 3200s and 2100s. I guess must be the C1E that caused this. I found a forum link that mention this as well. http://www.rage3d.com/board/ar...ex.php?t-33890505.html
Again thanks for the answer, since my comp been running stable for days now, I'd prefer not alter it more , awesome feel just like the first time ...

 

[WoodButcher]

Originally posted by: graysky
As far as I can tell, C1E doesn't do anything to my voltage. I have manually specified a value... I wonder if C1E only functions in "auto" voltage mode. I'm too lazy to do the experiment and see

My vcore was set around 1.5. The p5k I have does not have NB voltage in the bios. something was throttling this chip, not C1E, not speedstep, not temp, it was under 50c. Coretemp would read 3.6 but it was running at the 6x multi and cpuz read true, prime was still running, slow, my temps went down to 40-42. I did this CPU-z the same day. I'm currently running @ 3.2 w/ 4,4,4,11 timings, with the vcore set to 1.4 in bios and cpuz reads 1.24v as does coretemp.

 

[mrblotto]

Figured I'd add my 2 cents.....lol.
I had posted waaaaaaay back on page 4 about the 590/570i boards not being OC-friendly. Well, I thought I was SOL cause that's the board I have (590i SLi), but I figured I give it a try. To my pleasant surprise, I was able to take a E6600/2.4 up to 3.23 on water and run stable with zero issues! Took a little while, but that's part of the fun, right?
Then, of course, I just HAD to toss in a G0 Q6600....rofl. I have yet to try OC'ing that but am looking forward to monkeying with it as well.
Thanx for the great guide graysky, and to all for the helpful information in this thread.....

 

[Boyd117]

Is it possible to OC the PCIE 1.1 to a PCIE 2.0 on an ASUS P5K? or any board?

 

[WoodButcher]

Originally posted by: Boyd117
Is it possible to OC the PCIE 1.1 to a PCIE 2.0 on an ASUS P5K? or any board?

Nope, ya lose the drives, From a P5P800 w/ a Pentium4 551 I learned i couldn't increase my fsb w/o kicking up my PCI proportionally from 100. It was discussed here, I still increase my PCI to a max 120 although I find w/ the newer boards I have it is not needed to get the OC up.

 

[Boyd117]

cool lots of info, thanks for the link

 

[alexei]

Hi guys. Thanks for all the info. I have an ABIT IP35E w/ an E2140 (1.6 ghz). It's now up to 3.0 Ghz @ stock voltage, and running prime 95 it goes up to 54 degrees. I am very impressed. I'm still going to go higher, since I haven't seen any problems yet.

My question is this: I don't have a great power supply. It's a Sparkle 300w from a few years ago, rated for only 13A @ 12V. In addition to the mobo and cpu, I have a geforce 9700 gs and three hard disks (1 sata). I'll probably plug in a dvdrw and another hard disk. All I've read seems to indicate that most people think this is woefully underpowered, but, like I said, it seems rock-solid. I've played half life 2 -- no problem. Do I need a new PS? I suppose I'll get one when I see a good deal, but is it urgent? Am I destroying my parts?

Details: FSB 375x8. RAM @ 1:1.2 (OCZ 800 4-4-4-12 running at 5-5-5-15)
I had problems even without doing any OC at first, until I boosted the RAM voltage from the default 1.8 to 1.95 (the specs say 1.9-2.1). It's now at 2.05.

What worries me is that speedfan says Vcore is 1.16V, 1.22V under load (it's set at 1.275). More alarming, the +12V is at 6.65V (no joke). I guess that's what I get for using an underpowered PS, and should cut it out. But everything seems to be working all fine and dandy... is it possible speedfan is wrong?

Update: Using a multimeter, I found out that the 12v source appears to be 12v, give or take .5. So I guess speedfan has it all wrong. The BIOS reported it at 12V too, though I was thinking not all the parts were drawing power at that point.

 

[jliles01]

Hi all. First time overclocking and found this guide very helpful. I just wanted to see if I was on the right track.

Corsair lists the SPD for my memory @ 5-5-5-12. My BIOS defaulted it to 5-5-5-18 and thats what CPU-Z lists as the SPD also. I left it there for now and am running @ 1:1 and default 1.8v . I have my ext clock @ 400, X 8 multiplier, and 1.35 vcore and BIOS won't let me set it lower (options between 1.35 and 1.9x).

Using stock cooling. Core Temp reports idle @ 41C. I ran Prime 95 for 7+ hours with no errors and Core Temp showed 62-63C temps. before I went to bed and after I woke up. CPU-Z shows core voltage fluctuating between 1.300 - 1.320.

I was shooting for 3.2Ghz and I think I got there. I guess now I should play with the memory timings and voltages. And then stress w/Prime 95 for 8-12 hours. Have a 4 day weekend coming up.

Am I heading in the right direction? Anyone see any problems with the temps I'm getting? Does the voltage fluctuation showing in CPU-Z indicate a problem? Any and all advice is welcome.

Ands thanks for a great guide for us newbies to overclocking. It made it real easy to get a basic understanding of what I'm doing. I know there's a he$$ of a lot more to learn.

 

[graysky]

@jliles - Your temps seem normal for those settings. You can also consider following the suggestions in the temperature guide section of my guide (upgrade cooling, lapping, etc.) As to the mem timings... the most I would try w/ the mem would be 5-5-5-15 and you could try 4-4-4-12 if that's stable. You can also try lower voltage but 1.8 is pretty low. Dunno why you can't go lower than 1.35 in the bios... that's kinda high. My q6600 runs 9x333 @ 1.2625 for example. You might wanna ask around in the abit forums how to go lower.

The voltage fluctuation is probably vdroop... does it bounce around when idle or load or just going from idle to load?

 

[imported_LoKe]

Would running Boinc w/ SETI @ Home on all four cores of my Q6600 be sufficient in testing stability? I use Linux and I'm not familiar with benchmarking tools available, and would hate to boot up my windows HD just to test.

Q6600 @ 3.6GHz (401x9)
DDR2 80Mhz 1:1
vcore is quite high...1.4xx

Low core is 27C, high is 33*C idle, air cooled.

 

[graysky]

loke - tough to say... p95 is pretty much the standard. There is a version of p95 for LINUX, I dunno if it'll use all 4 cores though.

 

[imported_LoKe]

Well, I'm at 3.8 and so far so good. Gonna stress a bit, see what my load temps are. If they're acceptable, I'll beef the voltage another step and load up Windows and give it 4 hours of prime. 4GHz is the goal.

Unstable at 3.8GHz, stable at 3.6 Not sure what the problem is here. I was using 1.42 and 1.44 vcore and 3.8 would still fail. Memory divider is 1:1, tried dimm voltages of 2.00/2.10/2.20 and no go. What else is there to try?

 

[graysky]

Honestly dude, 3.6 for 4 cores cooled on air is damn good... add more vcore if you dare, but remember that temp increases with the square of voltage.

 

[WoodButcher]

Originally posted by: LoKe
Well, I'm at 3.8 and so far so good. Gonna stress a bit, see what my load temps are. If they're acceptable, I'll beef the voltage another step and load up Windows and give it 4 hours of prime. 4GHz is the goal.

Unstable at 3.8GHz, stable at 3.6 Not sure what the problem is here. I was using 1.42 and 1.44 vcore and 3.8 would still fail. Memory divider is 1:1, tried dimm voltages of 2.00/2.10/2.20 and no go. What else is there to try?

Air? wow, I got up there too, not stable, I can run at 3.4 stable but find the temps higher than I'm happy at so until I get to rebuild and lap my chip I've backed to 3.2.

Nice clock, what are your temps and what are you cooling that chip with?

 

[imported_LoKe]

Oddly enough, the temps aren't my issue. 55-60 on load, but this chip (G0 stepping) is said to handle 71*C. Cooling it with the ThermalRight Ultra 120 (I'm using some crappy thermalpaste, and badly applied). I need to pick up some new paste and apply it properly. Haven't tried AS5, so maybe I'll get some of that.

Also, I'm trying to run my memory at 1:1 with 800MHz (400) ram, does that technically mean I can't go above 400FSB on my chip? Or am I not really understanding how dividers factor in?

 

[WoodButcher]

AS5 is good, you've got the mem thing right, over 400 you'll be OC'ing the memory. No biggie, though it may be the cause of your stability loss. If you slow the memory w/ dividers and boost the fsb you may find your OC to be stable higher, or not,,,,,

 

[jliles01]

graysky-thanks for the advice. I read in the Abit forum, and not sure if I understand it properly, but it said the BIOS reads the VID of the processor and sets the min voltage accordingly. In Core Temp, under Processor Information, it says the VID is 1.35v. Maybe that's why I can't go below that. Intel spec sheet says .8 - 1.5v. My BIOS also only gives me multiplier options of 6-8, can't use 9 like you are. I'll keep reading, and learning. I think I'm happy now with what I have. 3.2G (400x8). Memory @1:1 & 5-5-5-12 @ 1.8v. Ran P95 for 9+ hours today while stuffing myself with turkey, and no errors. It would be nice to bring the core voltage down if I could and try for lower temps. Like I said, a lot more to learn. I'm not going to get into chasing benchmarks. Stability is my 1st priority, and a little more uuumph from my system is just icing on the cake. I want to be able to get as much out of CRYSIS as I can without any crashes. I'm also a big fan of Tweakguides.com, so I've done a lot of customizing in the OS and do the same in my games. This is my 1st Intel processor since my Pentium II @ 866Mhz was top-of-the-line. I've had 3 AMD's between then and now, but never OC'd any of them. I still have an Opteron 175 in my 2nd computer that I hear is a good OC processor.

If I do try to tighten the mem timings, should I go straight to 4-4-4-12, or go down in increments, like 4-5-5-12, then 4-4-5-12, and so on. In other words, should I change those 1st 3 numbers together or separately?

And the voltage fluctuations showing on CPU-Z happen all the time. Idle it goes from 1.31-1.32 every 1-2 seconds. During P95 it was jumping from 1.30 -1.32 every 1-2 seconds. But everything seems stable.

I also disabled those "power saving" features in the BIOS that you mentioned in your guide. I may go back & enable them one at a time to see if anything changes.

Another thing I noticed before I started changing the settings. Default for the memory was 5-5-5-18 @ 2t. I changed it to 1T in the BIOS, noticed no problems but hadn't stressed anything yet. However, CPU-Z still showed it at 2T. I went back to the BIOS and it still said 1T. Not sure what's happening there. And it may not even matter. This is also my 1st venture into DDR2. I changhed it back to 2T before starting to OC.

Once again, thanks for the advice.

 

[graysky]

Originally posted by: jliles01
graysky-thanks for the advice. I read in the Abit forum, and not sure if I understand it properly, but it said the BIOS reads the VID of the processor and sets the min voltage accordingly. In Core Temp, under Processor Information, it says the VID is 1.35v. Maybe that's why I can't go below that. Intel spec sheet says .8 - 1.5v. My BIOS also only gives me multiplier options of 6-8, can't use 9 like you are.

Wow, if that's true, no one should ever buy an abit board as that 'feature' is retarded. For example, my VID as read by coretemp is 1.2875. My machine runs p95 stable @ 9x266 (stock settings) at 1.1000V in the BIOS. It runs 8x333 @ 1.1500V in the BIOS. Finally, it's p95 stable @ 9x333 at 1.2625V in the BIOS. If my min. vcore was my VID that would be a tremendous waste of energy and an enormous production of heat. Is there no way to 'unlock' more settings in that BIOS?

Your 6-8 is normal for the E6750; the q6600 is 6-9.

Originally posted by: jliles01If I do try to tighten the mem timings, should I go straight to 4-4-4-12, or go down in increments, like 4-5-5-12, then 4-4-5-12, and so on. In other words, should I change those 1st 3 numbers together or separately?

The rule of thumb is x-x-x-3x-(x+1) for the first 5 timings. In otherwords, x=4 then 4-4-4-12-5; if x=5 then 5-5-5-15-6. I wouldn't recommend doing it incrementally, but if you have time, try it and test each one via superpi mod 1.5 doing the 16M or 32M and not using the pc for anything else during the test. If you do end-up doing this, please report the results.

Originally posted by: jliles01And the voltage fluctuations showing on CPU-Z happen all the time. Idle it goes from 1.31-1.32 every 1-2 seconds. During P95 it was jumping from 1.30 -1.32 every 1-2 seconds. But everything seems stable.

I wouldn't worry about it... could be vdroop. My system (before I vdroop modded it) didn't do that though...

Originally posted by: jliles01I also disabled those "power saving" features in the BIOS that you mentioned in your guide. I may go back & enable them one at a time to see if anything changes.

Give it a try sooner than later. Speedstep is your friend. C1E also.

Originally posted by: jliles01Another thing I noticed before I started changing the settings. Default for the memory was 5-5-5-18 @ 2t. I changed it to 1T in the BIOS, noticed no problems but hadn't stressed anything yet. However, CPU-Z still showed it at 2T. I went back to the BIOS and it still said 1T. Not sure what's happening there. And it may not even matter. This is also my 1st venture into DDR2. I changhed it back to 2T before starting to OC.

My BIOS doesn't have a 2T/1T option so I can't help you there. I'd do two things: 1) test via superpi mod 1.5 each setting to see if there is a difference as there should be and 2) report the bug via CPU-Z's website.

 

[jliles01]

Thanks graysky

I confirmed the VID min/vcore topic. There's a few people there unhappy about it also and are writing Abit about including an undervolting option. I assume it would be a BIOS update. I increased my ext clock in increments-333>350>360>380>400 and stressed each one for about 2 hours with 9+ hrs @ 400. The temps were always between 60-63C. So I think I'm "wasting" some of that voltage. I feel like I could stay stable at lower vcore and lose some of that heat.

Here's a link to the Abit forum about it, if you're interested:
http://forum.uabit.com/showthr...128741&highlight=vcore

Which brings up another issue. I'm running the original BIOS. There are 3 newer versions available now, so I'll probably be flashing the BIOS and starting over. I want to read about the newer versions and see if anyone is having any issues with them before I do.

I changed my mem timings to 4-4-4-15. Ran P95 blend for 2 hrs w/no errors. I may run it longer later. But I'm about ready to stop testing and start using. I'm a patient man as long as I get everything right now And I was in error earlier, I have my memory voltage at 1.9, not 1.8. Corsairs specs say it was tested at 5-5-5-12 @ 1.9.

Enabled Speedstep & C1E at the same time I changed timings. Ignoring voltage fluctuations. And I will test later the 1T/2T issue. With my old system & DDR-400, I could see the difference when going to 1T. I hope it's the same with DDR2.

You've been a great help, both in what I've learned and in easing my mind about OC'ing.

Thank you.

 

[jliles01]

I've done a little more testing and the results scared me a little. Ran superpi mod 1.5 @ 4-4-4-15-2T with no problem. I did notice that my core voltage as reported by CPU-Z stayed steady @ 1.32 while it was running. I noticed 3-4 times when the voltage dropped to 1.7 for about an eyeblink, and I think I saw my multiplier drop to 6 once or twice. It would happen almost faster than my eyes could see. Maybe that was one of the power saving features trying to kick in during a pause between iterations. Sounds logical, but what do I know. Could the steady voltage be because the cores weren't at 100% like in P95? Task Manager said CPU Usage 50%, and the graph showed about 75% on one core and 25% on the other.

When I changed to 1T, it wouldn't post. One good thing about this Abit IP35 Pro is you can save profiles in the BIOS. At least I had thought ahead and saved my settings before the change. I loaded the saved profile, instead of clearing the cmos, and everything was fine. I then tried going to 4-4-4-12, but had the same results. I'm computed out for at least a few hours. I may try raising the DDR2 voltage and trying again later.

But I'm pretty satisfied right now.

 

[graysky]

That min. voltage thing is bullsh*t if you asked me. Maybe they'll release a BIOS fix. Anyway, glad to hear your machine is up and running.

 

[jliles01]

graysky-I increased my mem voltage t 2v and am now able to post @ 4-4-4-12-1T. Ran 4+ hrs P95 blend w/no errors.

I took your advice about the 1T/2T CPU-Z problem and did some more testing. There is a difference in the results using superpi mod 1.5. However, with my inexperience, I'm not sure what the difference actually means. I used 32M calculations for both.

At 1T, the end results were:
0h 15m 28.953s Loop 24 finished
0h 16m 01.625s PI value output -> pi_data.txt

At 2T, the end results were:
0h 15m 29.156s Loop 24 finished
0h 16m 01.843s Pi value output -> pi_data.txt

With such a small difference, is it due to changing 1T to 2T, or is it just a random difference in the tests being done at different times? Even thought BIOS says 1T, am I actually running at 2T? So is BIOS reporting the command rate wrong, or CPU-Z reporting it wrong? Is this a bug for CPU-Z or Abit? I guess that's the main question.

When I first posted this issue, I was using CPU-Z version 1.41. These tests were done using 1.42. But it stills reports the command rate at 2T, no matter the setting in BIOS.

I don't have a lot of forum experience, so I hope I'm not breaking any rules in posting this here. Perhaps it should have it's own thread. If I am, I apologize.

On a side note, which may be more in line with this thread topic, could I use your guide to OC an AMD Opteron? I know there's a difference in the way they work. FSB, memory controller location, etc.

Thanks again.

 

[graysky]

@j - well, to answer the reproducibility question, you'd need to run it many times to get a standard deviation. Since you're literally talking about 200 ms, for my money, that's the same... you're literally talking about 99.978 % difference. As to the T1 vs. T2 thing, I have no experience with that; my P5B-Del doesn't offer that as an option. Finally, I don't see why you couldn't apply these principals to another brand chip. Lemme know how it goes.

 

[GeneralOreo]

Hey guys I've got a couple questions... Motherboard is P35-DS4 Rev 2.0 BTW.

1 - when I change the voltage controls to manual and change a few options, I get a "Voltage not optimized!" message above the voltage control sections... should I just ingore that if the system stays stable?

2 - on my motherboard I can only find the PCI-Express frequency, I don't see a PCI one so I can lock it down to 33.33...

I guess the BIOS just isn't allowing access to it and I shouldn't be worried about ruining my sound card?


Thanks a lot for any help.