Frustrated Conroe overclocker - P5B-E ocing help?

[augiem]

Dang it, I just can't win. "The great Conroe overclock" rush has just been bad luck for me.

Bought an E6600 and a DFI 975X mobo about a month ago. Max speed I could run on the DFI was 3.2GHz because the board just refuses to go higher than 360 FSB -- it's a well known issue. (But I'm stuck @ 333 because it won't cold boot at any higher) BIOS updates just make issues worse.

So, what do I do, I decide to plunk $180 down for a new mobo that's supposed to break those FSB limits... Maybe I can really see what this proc is capable of?? I buy a P5B-E -- a brand new revision C2 965 mobo. Set the thing up hoping this will be a better board than the flawed DFI. Guess what? IT'S WORSE!

On the P5B-E, I cannot even post at 333 FSB! Max I was able to get was 315FSB....
REALLLLLY disappointing. Such a waste of money.

Am I missing something here?? I disabled ALL onboard devices, tried it with only one stick of memory (Patriot 1GB 667MHz), disabled all fancy CPU options like C1E, etc, set the memory timings to the loosest possible and using the biggest divider I could, set Vcore to 1.4, Memory voltage to 2.0, locked PCI-E to 100MHz, PCI to 33.3, etc. I even downloaded the latest BIOS for the board. Nothing helps.

I can't think of anything else to do. I so wanted this board to be better than the DFI, but as it stands, the Asus is just plain worse.

Augie

 

[myocardia]

3.2 Ghz with 1.4v vcore is about average for E6600's. If you want faster than that, you're gonna have to give your chip more vcore. And changing motherboards isn't going to change that.

 

[Budman]

600 mhz overclock and still not satisfied.. were you expecting 4ghz when you paid for 2.4ghz ?

 

[NoStateofMind]

Well I got 3.2Ghz with my E6400, but I would still like to hit 3.5Ghz. Last time I tried it froze as soon as I chose "save and exit" in BIOS

I'm not unhappy at all though. If I had an E6600, I would shoot for 3.6. I'm starting to think maybe I'm too much of a noob at it or some people are getting lucky with thier mobo's.

 

[jiffylube1024]

Goes to show you that you really need to do your research on motherboards before taking the plunge. Especially when a new series of CPU's comes out; even with advertised "support" for the new tech, boards may not be as overclocker friendly as a more mature second-generation motherboard (eg the DFI Lanparty NF4 series, which was the fruition of a good NF3 series plus a boatload of improvements...).

With that said, if you can find a good board to give you 3.2 GHz at boot, then you shouldn't be unsatisfied. 3.2 GHz is still faster than anything AMD can do at the moment (ridiculous cooling aside).

 

[GuitarDaddy]

From what I've seen the 975x chipset seems to do better with conroes, while the P965 chipset works the best with Allendales. I haven't seen too many people having success with conroes on P965 boards.

And I agree that you will most likely need more than 1.4v to get past 3.2ghz

 

[augiem]

I do not get a good 3.2 bootable speed on the DFI. I said I maxed out at 3.0 because at 3.2 it refuses to boot 99% of the time unless I reinput the data in the bios and do a soft reboot. Then it will boot, but that's too much hassle to do all the time.

And it's totally stable at 1.45 vcore at 3.2, but again, this is not usable because of the booting issue. Even at 1.6vcore I cannot post at 1MHz beyond 360fsb EVER. This isn't the cpu suddenly flaking out -- it's the FSB wall.

>> 3.2 Ghz with 1.4v vcore is about average for E6600's.

I was running 1.4v on the P5B, which was not at 3.2GHz or anywhere close. It stopped at 315MHz FSB, which is like 2.8 or less. Vcore isn't the issue since I can run this chip at 3.2@lower voltages on the DFI but needs like 1.45 for full stability.


And to the responders who say just be satisfied with what you got... This is an overclocking forum where people who want to overclock should be able to go for help. If I didn't want to push the chip I wouldn't be here. I do not expect 4 GHZ, I wanted 3.5-3.6 tops. But my major disappointment is the lemon mobos.

THE CHIP IS NOT MAXED OUT -- it's the motherboards. The DFI just has an FSB wall, and I don't even know what the P5B's problem is.
Weird thing is, even at lower multi of 6x, this thing cannot reach 333FSB -- so it's not the chip unless it's just some weird incompatibility between the P5B and the 6600.... And I have heard the anecdote about 965's being better with Allendale than Conroe, but I see many people's systems on boards with 965's and 6600's going to 3.5 and beyond.

 

[Thor86]

Hmm, I know there are issues with Asus Conroe boards that do not boot over that wall you are reaching with certain bios memory settings.

Check out this nice C2D board tweaking thread:

http://www.xtremesystems.org/forums/showthread.php?t=115765

 

[imported_Buckster]

I was running at 3.15 gig ish - but anything over 3.05 gig I need to ramp up Vcore - so 3.05 is about 1.3875 volts, 3.1 is about 1.41 then 3.15 is about 1.4375, BUT if I run at 3 gig which is very close performance wise I can run at 1.36 volts = 1.33V measured by probe - so a lot less volts for about 3% less performance ....

Mark.

 

[The-Noid]

There is sometimes a wall on P5B where you have to go to say 401 mhz fsb. Your ram probably can't take that. And the 965 chipset has problems with patriot ram. Try to lower the multiplier to 6x on the p5b, put 2.1 vdimm or 2.2 vdimm (don't know how far the -e goes, just the deluxe) change timings to 5-5-5-15. PCI-E = 101, northbridge = +.1, then boot at 401 mhz fsb. will be 32xx mhz but may have a better chance of booting. Also use the 1:1 or even 5:4 ratio on memory.

 

[augiem]

Originally posted by: Yoxxy
There is sometimes a wall on P5B where you have to go to say 401 mhz fsb. Your ram probably can't take that. And the 965 chipset has problems with patriot ram. Try to lower the multiplier to 6x on the p5b, put 2.1 vdimm or 2.2 vdimm (don't know how far the -e goes, just the deluxe) change timings to 5-5-5-15. PCI-E = 101, northbridge = +.1, then boot at 401 mhz fsb. will be 32xx mhz but may have a better chance of booting. Also use the 1:1 or even 5:4 ratio on memory.

I tried 315, 333, 375, 400, 401, 420, 450 all with the lowest divider so the memory wouldn't be running too hard. I seriously don't think it's my memory because it runs very well at 800+ MHz speeds on the DFI board. I also did try 420 and 450 at the 6x mult, all with boosted vcore, vdimm, PCI-E at 101, and the slowest mem timings I could go... The only thing is I have no northbridge voltage setting to play with.

Augie

 

[Accord99]

augiem, what's your setting for the Refresh Cycle Time? On my P5B-Deluxe with Mushkin EM2-6400 memory, if its set to something other than the highest value of 42, the system will not post. Even if the RAM is underclocked.

However, once set to 42, I can boot up no problems at 400MHz FSB and the RAM can be overclocked to 900MHz.

 

[augiem]

Originally posted by: Accord99
augiem, what's your setting for the Refresh Cycle Time? On my P5B-Deluxe with Mushkin EM2-6400 memory, if its set to something other than the highest value of 42, the system will not post. Even if the RAM is underclocked.

However, once set to 42, I can boot up no problems at 400MHz FSB and the RAM can be overclocked to 900MHz.

Hey! Weird but encouraging news! Your suggestion about the Refresh Cycle Time got me messing with more of the memory timings... I already was on 42, but some strange combination of 103MHz PCI-E bus combined with a 5 tCL actually BOOTED at 402 MHz x 8 mult! I was trying 6 tCL, but that refuses to boot even though it's looser.

It's unstable in orthos, but I'm trying to track down whether it's the memory or Vcore. I'm wondering about vcore because I'm setting it to like 1.45 and it's giving me readouts in CPUZ and Asus mobo monitor and even ASUS bios mobo monitor of like 1.38v. It's either REALLY undervolting, or it's just giving me a low reading. And the temps are way higher than on the DFI, making me think this things sensors are screwy.

First sign of hope in a long while! Thanks!

Augie

 

[Accord99]

Originally posted by: augiem

It's unstable in orthos, but I'm trying to track down whether it's the memory or Vcore. I'm wondering about vcore because I'm setting it to like 1.45 and it's giving me readouts in CPUZ and Asus mobo monitor and even ASUS bios mobo monitor of like 1.38v. It's either REALLY undervolting, or it's just giving me a low reading. And the temps are way higher than on the DFI, making me think this things sensors are screwy.

First sign of hope in a long while! Thanks!

Augie

No problem. As for the orthos problem, what's your Static Read Control setting? This is in the North Bridge configuration section of the BIOS. When it was on Auto for me, Prime95 would throw an error in a couple of minutes when the FSB had risen to above 300MHz. Once it was disabled, the Prime95 problems disappeared.

 

[augiem]

Originally posted by: Accord99

No problem. As for the orthos problem, what's your Static Read Control setting? This is in the North Bridge configuration section of the BIOS. When it was on Auto for me, Prime95 would throw an error in a couple of minutes when the FSB had risen to above 300MHz. Once it was disabled, the Prime95 problems disappeared.

SRC is set to disabled.

If the DFI boards voltages were correct, this Asus P5B-E is seriously undervolting the chip.

I'm running it at 3.0GHz at 1.425v BIOS set which is turning out to be a 1.368v reported vcore (Asus MB, ASUS util, and CPUZ) with no Orthos errors at all. (almost exactly what I needed on the DFI to run 3.0 stable)

But at 3.0GHz at 1.3625 BIOS set, it reports 1.288-1.312v and does have errors in orthos within 2 minutes.

Another strange issue... Because I trust the reported voltage more than the voltage I set in BIOS, I ran at 3.25GHz trying to reach the 1.48v required by the DFI to be stable, so I set the bios to ~1.525v, but when I run the tests it fluctuates between 1.448v and 1.48v reported voltage. In fact, it runs at 1.47-1.48 until I start the orthos blend test, which causes the vcore to drop to 1.448-1.46v. This is very surprising considering I have a top-notch 600w Enhance power supply and very few devices in the system.

It looks like as long as I leave the RAM timings at 5 TCL and all my other settings the same, I can boot at any FSB I want. But further tests prove that to boot windows at 3.4GHz requires quite a heavy boost in vcore and my temps are already uncomfortable at 3.2GHz. (reported ~5-6 degrees higher on the P5B than on the DFI).

I suppose now that I managed to get all this working, I can safely say my chip is now the limiting factor.

Augie

 

[Accord99]

Originally posted by: augiem
Another strange issue... Because I trust the reported voltage more than the voltage I set in BIOS, I ran at 3.25GHz trying to reach the 1.48v required by the DFI to be stable, so I set the bios to ~1.525v, but when I run the tests it fluctuates between 1.448v and 1.48v reported voltage. In fact, it runs at 1.47-1.48 until I start the orthos blend test, which causes the vcore to drop to 1.448-1.46v. This is very surprising considering I have a top-notch 600w Enhance power supply and very few devices in the system.

It shouldn't be the PSU. All the Asus MBs have this difference between the set voltage and the actual reported voltage. The voltage drop under load is an Intel specification.

It looks like as long as I leave the RAM timings at 5 TCL and all my other settings the same, I can boot at any FSB I want. But further tests prove that to boot windows at 3.4GHz requires quite a heavy boost in vcore and my temps are already uncomfortable at 3.2GHz. (reported ~5-6 degrees higher on the P5B than on the DFI).

What temperatures are you getting from Core Temp or the Intel Thermal Analysis Tool under load?

 

[augiem]

Temps... Checked them on CoreTemp.

3ghz --- 35 idle shows up as 42 idle in Asus.
I was running it at 3.25GHz load and coretemp got to 55c and Asus was reporting it as 66c.

I recall my DFI readings were something similar -- 34c or 35c on coretemp. Can't be too sure. 60c on Coretemp would be too high I think.

Augie

 

[Zap]

Originally posted by: augiem
Dang it, I just can't win. "The great Conroe overclock" rush has just been bad luck for me.

That's the YMMV part of overclocking. Bunch of people swinging their e-shlong about how high of an overclock they got while the ones without that high of overclock stay quiet because there's nothing to say, so everyone starts expecting the higher overclock and many become disappointed.

I saw it happen with Opterons. "OMG, did someone put down Opteron overclocking?" Some people act as if every single socket 939 Opteron made will hit at least 2.7GHz "guaranteed" but that just isn't true - I can prove it because I had one that didn't even come close. Of course the typical response was that I wasn't using the "right" components, or that I was doing something wrong. Well, I had multiple Opteron 144 chips and one of them couldn't get near the speed of the other two using the exact same parts and settings. A couple other people then also came forward and mentioned that their Opterons weren't hitting those "guaranteed" speeds either. As for Core 2 Duo, I know someone with an Allendale that won't even POST at 3GHz on an Asus 975X chipset board. He got all the "right" parts and the BIOS settings seem right, but around 2.7GHz the CPU starts needing more and more voltage to stay stable - sure sign that it's near limits.

 

[SparkyJJO]

Ever think that maybe your particular chip is no OC king? There are always good OC chips and some that just don't do so well. Maybe you are at your true OC potential but just assume that it can go higher because some others have? Doesn't always work that way.

Sorry, but some just do better than others. And you possibly did not get a great OC chip. 600MHz OC isn't a bad OC anyway.

 

[GuitarDaddy]

3.2ghz is sweeet! 300mhz over a stock x6800 won't set any records but who could complain?

 

[TanisHalfElven]

Originally posted by: Zap

Originally posted by: augiem
Dang it, I just can't win. "The great Conroe overclock" rush has just been bad luck for me.

That's the YMMV part of overclocking. Bunch of people swinging their e-shlong about how high of an overclock they got while the ones without that high of overclock stay quiet because there's nothing to say, so everyone starts expecting the higher overclock and many become disappointed.

I saw it happen with Opterons. "OMG, did someone put down Opteron overclocking?" Some people act as if every single socket 939 Opteron made will hit at least 2.7GHz "guaranteed" but that just isn't true - I can prove it because I had one that didn't even come close. Of course the typical response was that I wasn't using the "right" components, or that I was doing something wrong. Well, I had multiple Opteron 144 chips and one of them couldn't get near the speed of the other two using the exact same parts and settings. A couple other people then also came forward and mentioned that their Opterons weren't hitting those "guaranteed" speeds either. As for Core 2 Duo, I know someone with an Allendale that won't even POST at 3GHz on an Asus 975X chipset board. He got all the "right" parts and the BIOS settings seem right, but around 2.7GHz the CPU starts needing more and more voltage to stay stable - sure sign that it's near limits.

its true. my s939 opteron 165 can even reach 2.6 on stock volts

 

[augiem]

Originally posted by: SparkyJJO
Ever think that maybe your particular chip is no OC king? There are always good OC chips and some that just don't do so well. Maybe you are at your true OC potential but just assume that it can go higher because some others have? Doesn't always work that way.

Sorry, but some just do better than others. And you possibly did not get a great OC chip. 600MHz OC isn't a bad OC anyway.

As I have said, I conclude now that the chip really can't comfortably go higher than 3.0, but there was no clear indication of that before. The 360FSB limit WAS the DFI motherboard -- NO QUESTION. I can post at 420 and beyond on the Asus. I can even post at 3.6GHz now, but the chip cannot run stably it without massive voltage which I don't want to pursue.

There is no way in the universe you or I could tell how far this chip could go until I bought this 2nd mobo and did these tests. That is a fact. Hindsight is 20/20, so you can see that my chip is no oc king now, but if you were to say that before would have simply been a guess.

Augie

 

[lopri]

What's your Vmch and Vdimm? Intel platform is totally different from A64 platform. Vcore isn't the only factor of CPU clocking. CPU, MCH, memory are all tightly co-related so you can't use 'process of elimination' here. For example on my system,

@266FSB: DDR2-800 CL3 takes 2.2V for memory
@400FSB: DDR2-800 CL3 takes 2.35V for memory

Memory frequency and timing hasn't changed but voltage requirement changes. Strange, huh?

What is the max vMCH and vDIMM on your board?

 

[tr1kstanc3]

Have you considered trying a different PSU?

 

[augiem]

Originally posted by: lopri
What's your Vmch and Vdimm? Intel platform is totally different from A64 platform. Vcore isn't the only factor of CPU clocking. CPU, MCH, memory are all tightly co-related so you can't use 'process of elimination' here. For example on my system,

@266FSB: DDR2-800 CL3 takes 2.2V for memory
@400FSB: DDR2-800 CL3 takes 2.35V for memory

Memory frequency and timing hasn't changed but voltage requirement changes. Strange, huh?

What is the max vMCH and vDIMM on your board?

Max Vdimm is 2.1v and I don't see any Vmch setting... There is something called FSB Termination Voltage which goes up to 1.45v.

I could boot at 3.5GHz 1.5vcore 2.0vdimm 1.4 FSBTV, but it begins to error immediately in Orthos. One time it rebooted.

After your post, I tried again at that setting with 2.1vdimm 1.45 FSBTV, but I get the same issue.

Trying 3.6GHz with all the max vdimm etc and about 1.52vcore does not boot windows.

Augie