Sign in with
Sign up | Sign in
Your question

Overclocking E6400 stuck @ 2.93 (366*8).

Last response: in Overclocking
Share
November 24, 2006 1:00:25 AM

OK, I've read several guides, and have had some success overclocking, but I'm hitting a wall, and it's lower than it should be from what everyone says, so if anyone could give me some pointers, that would be great.

Current setup
CPU: E6400
Mobo: P965-DS3 (edit - forgot to mention - bios version F7)
Corsair XMS2 TWIN2X2048-6400C4 2GB 2X1GB PC6400 DDR2-800 CL 4-4-4-12
Zalman CNPS9700LED
Sapphire X1900XT 256MB
Ultra X-finity 500W dual-rail
Aerocool ExtremEngine 3T case

Windows XP Pro SP2

I've gotten it up to 2.93 GHz (and run dual prime95 for hours and hours problem-free) with the following settings:
all the extra features the guides say to disable are disabled

FSB: 366 MHz
DDR multiplier: 2.5
RAM timings: 5-5-5-12
vCore: 1.300 (and I might be able to back this off further, just haven't tried lower yet)
DDR voltage: +0.3 (2.1)
FSB voltage: +0.2 (1.4)
MCH voltage: +0.1 (1.55)

Idle temps in the low-mid 30s with the Zalman and case fans on low. Leaving the Zalman and case fans on low during 100% load results in 52-53C, cranking all fans to max results in a noisy 45-46C.

I can get it up to 2.99 GHz (though I didn't torture-test that setting yet) but if I try to go past that, Windows just won't start.

I've fiddled with the voltages, timings, and just about everything, but it refuses to do much better. Not that I'm totally dissatisfied with the current increase, but I would like to hit 3.2, it seems everyone else with these parts can do it.

So, did I just get unlucky, or can someone point me in the right direction?
November 24, 2006 2:45:21 AM

Nobody cares, huh?

Anyway, I played with it some more, still no luck at all. It posts just fine, but always freezes up loading Windows. Loading safe mode, it always seems to fail at "giveio.sys".
a b à CPUs
a b K Overclocking
November 24, 2006 3:49:45 AM

How high have you taken the voltages for vcore? Might need some more juice. Remember, just because someone can hit 3.6Ghz at stock voltage doesnt mean you can, everyones chip is different, as is their mobo and every other part in the system. You did read wusy's guide? Oh and I'm happy to see that you have done research since there are so many people on here that want help when theyve never read anything. GL :wink:
Related resources
Can't find your answer ? Ask !
a b à CPUs
a b K Overclocking
November 24, 2006 3:52:10 AM

Oh and people do care, its just they're too busy helping those who wont do the research they need to do and are afraid of blowing something up (and cant speak english :)  ).
a b à CPUs
a b K Overclocking
November 24, 2006 3:54:35 AM

And one more thing (this is really helping boost my post count 8) ) dont use noob in the title when talking about OCing or it immediately suggests a lazy idiot who has no idea of what theyre doing, no offense.
November 24, 2006 4:22:46 AM

Well, I am a noob, in that I am new to overclocking, This system is my first overclocking attempt. If that's not noob, I don't know what is. I wouldn't say I have no idea what I'm doing, but I lack experience. I guess I should take the "noob" out of the title (a lie of omission :lol:  )

I'd be lying if I said I wasn't a little afraid of blowing something up - who isn't? But it's a risk I'm willing to take.

I've tried vCore a little over 1.4, with no better results. Next step: 2.something! - Kidding.

I've read wusy's guide several times before making my first attempt (I mean, c'mon, it's freakin' stickied in this very forum - who could miss it?). That got me right to 2664 with no trouble. I've also read some others, like this thread, and looked over a lot of people's settings in this thread at extremeoverclocking.com

I guess, in the end, I'll just have to keep trying until I'm too afraid (or too sensible) to go further.
a b à CPUs
a b K Overclocking
November 24, 2006 4:34:19 AM

You'd be surprised how many people dont read wusy's guide, thats why I asked. Even tho you lack experience, you did the research so you have some head knowledge, which you need b4 you get experience. There are so many people who ask for a quick guide to OCing so they can get straight into it, never works. We just have to keep saying, "do the research, read wusy's guide and google until you know what your doing"

Well it looks like vcore is not your prob, yet. Try using a RAM divider just to see if the RAM is holding you back, which it shouldnt be, but its worth just making sure. Since I've never OCed a C2D I cant help alot, lets hope someone who has joins soon. At least if we keep relpying it'll keep bumping this thread.
November 24, 2006 4:52:30 AM

Well, when I try anything higher than my current stable 366 FSB, I set the divider to 1:1 (or the multiplier to 2.00, as it is listed in my bios settings) and, it doesn't go any lower.

You'd think that 400*2 would fit nicely with DDR2-800, but for some reason, I can't even get Windows to load past 375. even 380 is too much for it.

Here's a question: with my FSB at (or near) 400 and my divider at 1:1, should I have to relax my timings, or should it work with the 4-4-4-12?
November 24, 2006 7:09:50 AM

Hint: What speed is your PCI bus, Intel SATA interface, & other south-bridge connected hardware running at ?

History:
At 366 FSB the PCI bus will be near 37 MHz because of FSB/PCI divider, this will cause instability.
SATA signals tend to go spastic past about a +/-10% deviation in clock.
Try locking PCI bus to 33.333 MHz, or using a 400 FSB (If you can lower CPU multiplier, raise voltage / heat output safely, or get CPU working at rather insane speed).
I think most ppl above overlooked that, and the Gigabyte BIOS does support locked PCI clock speed, which will keep SATA and everything else South-Bridge connected within spec.
Also try +0.05 Voltage on FSB and MCH (Instead of those ridiculously high values), and use 1.375 V for CPU at circa 3.00 GHz with a 333 or 334 FSB.
- A Tabris:D arkPeace post

Questions:
With a 366 FSB surely it is booting at 3.3 GHz right, which at 1.30 Volts on a E6400 is unlikely to work (speed binning). Are you changing the multiplier of the CPU ?
What Power Supply Unit (PSU) are you using ?
Which BIOS is the board running ?, Is it the GA-965P-S3 or GA-965P-DS3 ? (Worth double checking IMHO).

Recommendations:
1.375 Volts, 333-334 FSB (as PCI/SATA ratios will be within spec), 3000 MHz, Core 2 Duo = about 100 watts draw, enough to cool with the stock cooler.

As above, back off the RAM to 2.05 V (+/- 0.05 V), relax timings [5-5-5-15 comes to mind, then work down with *heaps* of testing], and reduce clock speed / increase ratio of FSB - RAM.

Remember that it is quite likely 1333 FSB (333-334 quad-pumped), with RAM at dual-channel 667 - 800 will works fine, esp with 'slow' timings. My money is on the PCI and/or SATA bus not being within spec.

You may need to replace any corrupted files if you've overclocked 'to far', this could involve re-installing Windows XP Pro, or MD4+MD5 hashing a lot of files to isolate the corrupt ones and 'fix' them. If you can't get into Windows and/or the registry is already slightly corrupt you're ****ed anyway and will need to re-install OS at stock settings. (MemTest86 overnight before installing OS).
Consider: Install OS at 'stock' settings then use Ghost 2003 [DOS] to image it while in stock settings, as chances are something critical will get corrupted again until you find the sweet spot. This way you can restore a known good (registry is not even partically corrupt, all driver files have OK integrity) OS within minutes, then try different settings.

"Overclocking 101: Corrupting files / binaries (code/data) on disk, or having them go corrupt once loaded into memory is *very* easy when overclocking. - It is 'un'corrupting that is the hard part. - Once corrupted on disk it is a right bitch to isolate / fix without re-installing the OS" - Tabris:D arkPeace
November 24, 2006 11:28:27 AM

I'm not sure you've read what I've got, and what I'm having problems getting.

Quote:
Try locking PCI bus to 33.333 MHz, or using a 400 FSB (If you can lower CPU multiplier, raise voltage / heat output safely, or get CPU working at rather insane speed).
I think most ppl above overlooked that, and the Gigabyte BIOS does support locked PCI clock speed, which will keep SATA and everything else South-Bridge connected within spec.

I can only find a PCI Express frequency, and I've had that locked at 100 all along.

Quote:
Also try +0.05 Voltage on FSB and MCH (Instead of those ridiculously high values), and use 1.375 V for CPU at circa 3.00 GHz with a 333 or 334 FSB.

The board doesn't seem to have that fine of adjustments - everything seems to be in 0.1V increments, except for cpu vcore. And it already fails if I back off the FSB voltage a notch.

Quote:
With a 366 FSB surely it is booting at 3.3 GHz right, which at 1.30 Volts on a E6400 is unlikely to work (speed binning). Are you changing the multiplier of the CPU ?

It's a E6400, the CPU multiplier is locked at 8, resulting in 366*8=2928 MHz - all in the original post (and the title). This setting runs very smoothly.

Quote:
Which BIOS is the board running ?, Is it the GA-965P-S3 or GA-965P-DS3 ? (Worth double checking IMHO).

All my stuff is listed in the original post, it's a DS3, running the latest available bios (F7).

Quote:
1.375 Volts, 333-334 FSB (as PCI/SATA ratios will be within spec), 3000 MHz, Core 2 Duo = about 100 watts draw, enough to cool with the stock cooler.

That would be no problem (heck, 366 MHz is no problem). Cooling is not an issue (at this point) I have a CNPS 9700, and it works great. But it's a 6400, the multiplier doesn't go over 8, so 333*8 would get me 2664 (and I've already done that using wusy's guide, problem-free, first try).

Quote:
As above, back off the RAM to 2.05 V (+/- 0.05 V), relax timings [5-5-5-15 comes to mind, then work down with *heaps* of testing], and reduce clock speed / increase ratio of FSB - RAM.

Can't set in 0.05V increments, only 0.1. The RAM is rated for 4-4-4-12 @ 800MHz and 2.1V. I have been trying 5-5-5-15 when I shoot for a high speed.

But thanks for your advice and insights.
November 24, 2006 2:20:23 PM

OK, another user PM'd me, and helped me out.

I reduced FSB and MCH voltage to normal, set RAM to 1:1, and FSB to 400 MHz, vCore is currently 1.35, and it got into Windows - which is way the hell further than I'd gotten at that speed before. Seemed to run fine, but I don't have time to really test now, will leave it off until later.

Tabris, thanks for the advice, particularly about OS-backup. I will probably put it back to stock and reinstall and use Ghost to be safe.
November 24, 2006 3:26:02 PM

Quote:
How high have you taken the voltages for vcore? Might need some more juice. Remember, just because someone can hit 3.6Ghz at stock voltage doesnt mean you can, everyones chip is different, as is their mobo and every other part in the system. You did read wusy's guide? Oh and I'm happy to see that you have done research since there are so many people on here that want help when theyve never read anything. GL :wink:


Yeah, i was pleased to see that someone actually did research before asking a question. Hopefully this is a sign of things to come, but then again the chances are it isn't :( 
November 25, 2006 7:37:47 AM

(Q1) Any chance of getting a (nearly) complete SiSoft SANDRA dump (System Report, etc) of the system at 400 FSB ?


http://www.sisoftware.co.uk
http://www.sisoftware.co.uk/index.html?dir=dload&locati...
and one at 367 FSB if it will boot at that speed.


They'll ZIP / RAR heaps small, and most ISPs have FTP space for users to upload to, failing that é-mail it over to me (Please). [Thought the voltage mod on MCH/FSB was overkill]. 8)

Just want to see what SANDRA thinks the PCI bus + other South-Bridge components are running at.

Make sure North-Bridge / MCH stays cool, ditto for MOSFETs / VRM. Just because they might be stock voltage and all. They'll get much hotter at 400 FSB (1600 FSB post QDR).

:oops:  PS: Kept thinking E6400 was 2.4 GHz, not the other one, 64 24. (I run a E6600 - 2.4 GHz stock at 3.0 GHz but the new part numbers are not the first thing I commit to memory - they just change to damn often :p  ) :oops: 

I want to know what PCI divider does every +33/34 MHz on base FSB clock speeds past 333 MHz - Well past 266 MHz to be honest, but every-tech knows them up to 333-334 FSB. Beyond that it isn't documented very well.

As in:
266 / 8 = 33.333 PCI
300 / 9 = 33.333 PCI (but not personally tested or observed)
333 / 10 = 33.333 PCI
366 / 11 = 33.333 PCI
400 / 12 = 33.333 PCI

But:
399 / 11 = 36.2727 PCI - (Which is +8% to +11.5% on some parts, since some manufacturers calibrate 33.333 MHz to 34.132992 MHz, and the difference multiplies out to be quite wide at higher speeds).
- Even with the PCI/SATA speed locked, it doesn't always work in Async at high overclocks. If possible I try to use sync FSB/PCI/SATA speeds when overclocking. (So my next stop is 366 - 367 FSB if I want to try for 3.3 GHz, but may need to re-jig memory ratio myself :p  )

Good to hear you got a 'heaps' higher FSB too. Hope the hardware holds together over the test of time. :D 
November 25, 2006 8:19:43 AM

Heh for 3 years at least that cpu clock of 2.93 is what i would be happy with, unless it was who can get the fastest clock.
November 25, 2006 8:35:20 AM

I didn't have the time to read through the whole thread. I don't know if your problem has been solved.

Your RAM shouldn't be the limiting factor unless you overclock it above 400MHz FSB, above 400MHz if it fails, loosen the timing to CL5, and increase the Vdimm, shouldn't have problems hitting somewhere around 500MHz with CL5 timing on that module.

The most obvios problem I see is, DDR Multiplier, you should set it at 2.00 instead. At 366MHz FSB, you're running your RAM at 915MHz, which is above the 800MHz stock rating. Besides that, C2D's should be run at 1:1 (2.00x) for maximum performance.

So now what you want to do is, have your RAM timings set to 4-4-4-12 at stock Voltage (check your RAM's stock Voltage settings, mobo's default is 1.8V, increase accordingly), RAM shouldn't fail at stock ratings, till you overclock above 400MHz FSB at 2.0x Multiplier.

If your system is failing while RAM running below stock ratings, it's most probably your CPU limiting you. Try increasing Vcore. Symptoms of unstable CPU usually varies from not being able to boot into Windows, freezes, fails to boot up at all, or even rebooting itself. Make sure you have a good enough PSU supplying enough power. When you overclock, observe all the readings, temperature, Vcore, etc ... from there you should be able to diagnose your problem ...

Anyway, just to let you know, your target of 3.2GHz seemed likely to be achievable. I hit 3.409GHz on a E6300, and I don't see any problems with E6400 hitting 3.2GHz. I've helped at least 7 people achieve above 3GHz on E6300, where most of them hit at least 3.2GHz. You can read more about it over the link at my siggy, regarding the Budget C2D System for $631.

If you need more assistance, you can PM me, I'll give you my MSN or so, I'll try to see if I could help you out personally, probably achieve a much higher overclock than your target of 3.2GHz. :) 
November 25, 2006 7:42:08 PM

Things are going fine.

I'm pretty sure the problem before was the voltages on MCH and/or FSB. Once I backed those off to sane levels, it went straight into windows. I didn't do any testing, and after all those un-backed-up attempts, I've reinstalled Windows (brand new machine, so no loss) and made a ghost backup, all at stock.

Then I plugged back in the settings that got me to 3.2:
FSB @ 400 MHz
CPU multiplier @ 8
PCI-Express @ 100 MHz
1:1 on the RAM, 800 MHz @ 4-4-4-12, 2.1V
1.35V vCore - I'm sure I can back this off a bit edit - it works just great at 1.3000
All other voltages normal

I'm currently running memtest on it. Ballpark figure, how long should I be running things like memtest and prime95 to be certain of stability? I know we're talking hours and hours, but what, like, 6 hours? 12? Is memtest massively important right now, given I'm not running the RAM above what it's rated?
November 26, 2006 6:06:15 PM

Memtest for 10 passes. P95 for 12 hours(my preference is Orthos. More user friendly). If everything passes, then you're good to go. And easing back the Vcore isn't a bad idea. I have my E6300 @ 3.22GHz 1.3 volts stable. These things don't need a big bump in voltage unless you pass the 3.3GHz mark.

http://sp2004.fre3.com/beta/orthos_exe_20060420.cab
November 27, 2006 4:46:46 AM

HAHAH...sux for u...my 6300 goes till 3.2 without any upage of the voltage
November 27, 2006 4:54:56 AM

How does it "sux for me"? I haven't tried a lower voltage than 1.300 yet, but that doesn't mean it won't work.
November 27, 2006 9:38:22 AM

1.3 is stock for C2D. And who cares what your running it at. He solved his issue and is pleased with his O\C. What are you, a 12 year old?
November 27, 2006 10:54:42 AM

Quote:
HAHAH...sux for u...my 6300 goes till 3.2 without any upage of the voltage


LMAO lewbaseball07, and you're considering yourself having a great rig?
Try beating my 3409MHz E6300 @ 1.2875V (below stock Vcore) ... and best yet, STOCK COOLING ... and the whole thing costs me $631 ... Anyway, don't tell ppl of who sux or who rox ... it's mostly on luck what parts you got ... the same parts might not work as well on other people's configurations ...
November 27, 2006 10:56:45 AM

Quote:
1.3 is stock for C2D ...

Stock Vcore for C2D's are actually 1.325V ... :) 
1.30V for overclocking is considered pretty good, as it's still below stock.
!