E5200 - Overclocking Thread

[W4LNUT5]

Hey guys.

I have been running this E5300 OC'd to 3.83Ghz with voltages around 1.43v for a little over a year now.

Recently I have been seeing some stability problems. Not sure if its the motherboard or the chip itself, but I do think its related to the voltages I run. I have had to step it down to 3.66 @ the same voltages to remain stable.

I know there used to be two camps on this issue, those who believe 1.45v is the max to run, and those who believe 1.36v is the max safe voltage.

I figured on a $60 chip, I'd push the 1.45v as my max. I don't really care if it dies, I'm going to update this system to a quad soon anyways. But I do want people to know these results. 1.45v may be a little much for a fairly 24/7 computer like mine.

Later
-walnuts

[nemaca]

Hello,

About 50 pages back or so, I posted about my e5200 and my motherboard.
My ASUS P5KPL-CM is one of the most troublesome motherboards when it comes to OC. It allows a decent amount of OC and it COULD do better but ASUS chopped its BIOS and mislabeled the memory ratios. Which makes one learn by heart the settings and actually end up trying ALL the possible combinations.

Which I did, increasing FSB by 1(one) unit at a time and testing it in the max/min area with all the multipliers. In the same time, i tweaked the DDR latencies on all(actually 2) the ratios available: 1:2(labeled 800MHz) and 3:5(labeled 667MHz). I removed also my pci network card, audio card to exclude having them interfere, since the pci ratio is locked down completely.

I keep my RAM at 2.0V and my VTT_CPU at auto, since it doesn't matter what i set(auto/1.2/1.3), cpuz reads only: 1.240, 1.288.
My best rock solid OC is @3.3GHz, with nights of tests and prime 24 hrs smallfft.
I now keep it @ 3.5, fails prime95 instantly at first core and after 10 seconds or so at next core. It fails the OCCT cpu test too, and I'm not sure.. is it the fact that @3.5GHz I should have upped the vcore? That's what I think. Also, 3.68GHz is the max I could do with this voltage at this mobo, and I ofc can't figure out if it's the voltage or the motherboard cap for cpu frequency.

What I haven't tried is the direct cpu pin mod, and I wouldn't want to, unless there is nothing else left for me to do, because some people got it right, some got it extremely unstable, so it's a confusing result at best.

A little screen shot from google on my BIOS:
http://media.photobucket.com/image/p...kpl-cmbios.jpg

And what my RAM can do on my mobo, tested a bit under 3 hrs with memtest86 boot cd 3.5 version.


So... problem is... I cannot go beyond 3.68GHz on cpu on stock (1.288) voltage or it is simply mobo cap limit at 3.7GHz, that with a 290 fsb. And if i manage to get the fsb up to 315, the best multiplier that lets me use is 10x, which is a sign of voltage problem, imo. At this config mem testing goes perfectwith memtest86 boot cd, while cpu test failed prime after a bit over 1 hour on one core and I stopped it.

Any ideas? I'm stuck.

P.S. lapped cpu and sink to near mirror face with different wool roughness textures pinned on a flat surface:P I don't have the grains, lol, and I bet these guys haven't took wool in consideration. Worked.

[nemaca]

Oh, how do I use IBT? I press start with Stress lvl at Maximum, at 32 bit(I have winXPx86), and Threads on Auto.
Somehow it finished the test in 0.22 seconds(?!?!!?)with the Success! note in the end.
Am I running the right program?... in the wrong way?

L.E. I put Thread on 1 and it did 5 rows of test. HOw many Threads and how many Times to run to put to get the ideal 100 loops?

[nolonger]

Originally Posted by nemaca

Hello,

About 50 pages back or so, I posted about my e5200 and my motherboard.
My ASUS P5KPL-CM is one of the most troublesome motherboards when it comes to OC. It allows a decent amount of OC and it COULD do better but ASUS chopped its BIOS and mislabeled the memory ratios. Which makes one learn by heart the settings and actually end up trying ALL the possible combinations.

Which I did, increasing FSB by 1(one) unit at a time and testing it in the max/min area with all the multipliers. In the same time, i tweaked the DDR latencies on all(actually 2) the ratios available: 1:2(labeled 800MHz) and 3:5(labeled 667MHz). I removed also my pci network card, audio card to exclude having them interfere, since the pci ratio is locked down completely.

I keep my RAM at 2.0V and my VTT_CPU at auto, since it doesn't matter what i set(auto/1.2/1.3), cpuz reads only: 1.240, 1.288.
My best rock solid OC is @3.3GHz, with nights of tests and prime 24 hrs smallfft.
I now keep it @ 3.5, fails prime95 instantly at first core and after 10 seconds or so at next core. It fails the OCCT cpu test too, and I'm not sure.. is it the fact that @3.5GHz I should have upped the vcore? That's what I think. Also, 3.68GHz is the max I could do with this voltage at this mobo, and I ofc can't figure out if it's the voltage or the motherboard cap for cpu frequency.

What I haven't tried is the direct cpu pin mod, and I wouldn't want to, unless there is nothing else left for me to do, because some people got it right, some got it extremely unstable, so it's a confusing result at best.

A little screen shot from google on my BIOS:
http://media.photobucket.com/image/p...kpl-cmbios.jpg

And what my RAM can do on my mobo, tested a bit under 3 hrs with memtest86 boot cd 3.5 version.


So... problem is... I cannot go beyond 3.68GHz on cpu on stock (1.288) voltage or it is simply mobo cap limit at 3.7GHz, that with a 290 fsb. And if i manage to get the fsb up to 315, the best multiplier that lets me use is 10x, which is a sign of voltage problem, imo. At this config mem testing goes perfectwith memtest86 boot cd, while cpu test failed prime after a bit over 1 hour on one core and I stopped it.

Any ideas? I'm stuck.

P.S. lapped cpu and sink to near mirror face with different wool roughness textures pinned on a flat surface:P I don't have the grains, lol, and I bet these guys haven't took wool in consideration. Worked.

Sounds like not enough vCore. If you can get it to boot at 315MHz at a different multiplier it's not your motherboard, it's the CPU that's unstable. Most E5200's can do 3.25-3.5GHz on stock voltage.

[nemaca]

So... this is the max I can get out of it, with unmodified vcore(stays at 2.42-2.96max). The processor is just fine, but needs more voltage. Any idea how I can get it? The problem is the motherboard as in...

1) It does not have a vCore setting, it has a VTT_CPU setting which has NO effect, I actually maxed all voltages allowed everywhere: effect 0.

2) The motherboard simply doesn't allow CPU's with frequencies higher than 3.7GHz (that's what i read on some forums, early this week, but unfortunately I lost the source).

The RAM allows a FSB of 315 at 3:5 ratio(the best ratio available), which lets me OC the CPU with a multiplier of x11.5 maximum, resulting in the frequency shown in the image below. With the max 12.5 multiplier, would come to a 3.937GHz. I have pushed only once without stressing the RAM to a 320 FSB, booted and stayed with it for about 30 minutes of browsing and it didn't crash or flickered in any way.
So my aim is obviously 4GHz, which, if I could slide more vcore in the CPU, would be easy to get, then ofc, remaining the stress testing.

.

[nolonger]

E5200 1.325V VID

This guy had his E5200 at 1.325V with a VID mod, you could try that, but I recommend reading a little bit on the matter before linking anything on your processor.

[PizzaMan]

Like nolonger said, more vcore. Mine takes 1.47v to keep 3.7Ghz stable. It was one of the first e5200s when they were released and was well binned to be an e5200. A lot of the newer e5200 and e5300 could have prbly been e7200s or higher but with a high demend for e5xxx series, that's what they become.

[nemaca]

Thanks guys,

So it's all clear then. No improvement except hard-modding. I'm going to document on this very well and take my time to get tools and patience. I'll repost after, hopefully, from a 4GHz cpu pc.

Best regards!

[nolonger]

Originally Posted by nemaca

Thanks guys,

So it's all clear then. No improvement except hard-modding. I'm going to document on this very well and take my time to get tools and patience. I'll repost after, hopefully, from a 4GHz cpu pc.

Best regards!

Good luck mate!

[nemaca]

Well, I've done the mod, both on vcore and the fsb. And I polished a bit more the heat sink so now its even shinier.


The problem is... When i restart my PC from software or from BIOS, it doesn't boot/post, just tries to start and keeps failing, like a car without gas. What would be the cause? The power supply i think is more than enough for it, so it couldn't be that. It's not the RAM either, cause when I first booted, i had them set at 533 speed, from their normal 400, and without any sign of kick-up on a short 3 round prime95 blend test, done more from reflex.