E5200 Overclocking and performance comparison.

[Assimilator1]

That's an awesome o/c!

Only just noticed I didn't post back about my 2nd PC, well it was nothing special, actually turned out to o/c less than my brothers E5200!

Asus P5QL Pro (Intel P43 chipset)
E5200 M0 rev, VID 1.18v (def FSB Term. 1.1v)
Thermalright Ultima 90
2GB OCZ DDR2 800 ? (CL 4-4-4-15)

Ended up at, 3.5 GHz (333x10.5), 1.3v vcore & 1.2v FSB Term. (BIOS settings).
That passed a 24hr OCCT Linpack run, don't know what the running vcore for that is, but whilst running DPAD (both cores @ nr 100%) it's 1.264v.
And as others have noticed with these, the higher the clock speed the lower the FSB it can handle, although boosting FSB Term. helped a bit.

Faster speeds seem to need a lot more voltage, I tried 3.6 GHz (313x11.5) upto 1.3125 vcore & 1.2v FSB Term, that failed OCCT Linpack at just over 1hr (& interestingly still hadn't failed normal OCCT at 3hrs 38 mins!). Didn't bother any further with that setup as the FSB was so much lower & RAM slightly underclocked (785 MHz).
Also tried 3.56 GHz (339x10.5, RAM 814 MHz), vcore 1.3v & FSB Term 1.3v that failed OCCT Linpack at ~4 hrs, so would have needed a little more voltage to be 'properly' stable.

I could get it stable at faster speeds but I don't know if its worth it really. I might some day try just out of curosity .
Anyway, a 1GHz o/c is nice .

 

[Assimilator1]

Revival!

A little update on my E5200, although I did further tweaking on my CPU back then I didn't come close to nailing it down until more recently.
Currently testing at 3.6 GHz (343x10.5), RAM 824MHz, vcore 1.325, FSB Term down to 1.1v (didn't help afterall), tRD7,NBv 1.2v.

Vcore may have to go upto 1.33125v to pass a 24hr OCCT run, & I might be able to get the RAM stable at 860MHz. But I'm at long last getting near to it's final settings .

So is anyone else still playing with these (or any E5x00 CPUs), or are they 'has beens'?

 

[Muel]

I just overclocked my brothers E5300, cracking little chip! For some reason the vcore refused to change though...

That said, I got some 1066mhz RAM, plugged it together, locked all the settings off then just turned the vcore straight up to 266mhz giving 3.46ghz

I was hoping for more like 3.8ghz, but the problem with the vcore not changing limited it. It was almost stable at 3.6ghz, but that had the ram running at 930mhz, so I decided to drop the fsb back down and the ram multi back up so it was running at 3.46ghz and 1066mhz.

 

[Assimilator1]

Yea sounds like a good compromise, & a good chip to do that on def vcore .

Btw what do you mean the vcore refused to change?, setting wouldn't change in BIOS or vcore actual readings didn't change in windows software?

 

[Muel]

Readings didn't change in windows or in the BIOS health screen either. Maybe it was just reading funny.

3.4ghz for mild gaming and web surfing is fine though.

 

[nikzGfx]

Nice, got my e5200 today, but also e5400 arriving next few days, will try to overclock both to their potential then decide which one to keep. Hoping they don't hit a FSB wall acompared to the e2160 which can't reach 3ghz!

 

[Assimilator1]

A FSB wall will just be a problem if your stuck with or choose a low multiplier, otherwise it's just a clock speed wall. No problems with the E5200 or E5400 with multipliers , I lowered it on mine & my bro's.

Let us know what you get out of them.

Good to see people are still tweaking these

 

[pastymuncher]

Got rid of mine ages ago. Got a E6300 to play with which could hit 4Ghz but needed 1.42v to do so. Upgraded my main rig to the Q9550 and am now running the E8500 in the second rig.

 

[Assimilator1]

4GHz from an E6300 that's awesome! (65nm core right?)
And 4.13GHz is a damn good o/c from a quad too , what vcore does that take?
I'll buy it from you when you come to sell it (ah I've just noticed it's water cooled, mine's on air).
What did the E8500 o/c to?

 

[pastymuncher]

4GHz from an E6300 that's awesome! (65nm core right?)
And 4.13GHz is a damn good o/c from a quad too , what vcore does that take?
I'll buy it from you when you come to sell it (ah I've just noticed it's water cooled, mine's on air).
What did the E8500 o/c to?

Wrong E6300. I meant the newer Pentium dual core based on the Wolfdale just like the E5200.

Q9550 vcore is 1.325v for 4.13Ghz. It does 3.65Ghz on stock volts. Don't intend to get rid of it anytime soon. Not planning on going with Intels or AMD's next round of cpu's at the moment.

It's the same E8500 i used for the benchies so 4.5Ghz.

 

[Assimilator1]

Ah I see makes sense, still 4GHz is good.
Damn good Q9550 you've got!

Been struggling to get my E5200 stable at just 3.6GHz today!, turns out it needed a lot more vcore to go from 3.45GHz to 3.6GHz. Currently testing at 1.35v.

Wierd thing is it ran OCCT normal test OK for over 20hrs before an error occured (with 1.325v), suggesting it was very near the right voltage, yet on OCCT linpack it errored in 3mins!!
I do wonder if the linpack test runs the CPU unrealistically hot, but I can't bring myself to ignore it either! lol

What do you guys reckon about OCCT linpack?

 

[pastymuncher]

For a quick stability test i do 20 runs of linx and then when i have settled on a clockspeed i do a 18-24 hour run of prime. My Q9550 does 20 runs of linx at 4.2Ghz but fails prime after almost 15 hours.

 

[Assimilator1]

Fair enough, can't say I've know of Linx.

I don't use Prime95 anymore, I found I could pass it for 24hrs & then my rig would crash (EUE) F@H within 10mins! I've not yet had that problem with OCCT, so OCCT is definetly a better stress test than P95. I guess that level of stabilty only really matters if you run DC,play MMORPGs, or some sort of production though?

I just can't decide whether I should use the normal OCCT test or it's Linpack test .
I think what I'll do is once my 2nd rig has passed it's current test I'm going to drop the vcore 1 step to see when/if the normal OCCT test picks up any errors.

Atm my E5200 @3.6GHz has passed OCCT linpack for 19hrs , I always go for 24hrs as I have had errors upto 22hrs on the rare occasion! This rig runs DC too btw.

 

[MikeTimbers]

Ha! Lucky sods. My R0 E5400 in a Gigabyte G31M-ES2L needs 1.45v to do 333x10.5 for 3.5GHz. Any less voltage andone of the cores fails Prime although the other one just keeps on going. Since it needs to be Prime-stable to fold properly I'm stuck having to use high voltage.

 

[Assimilator1]

Ouch that's high vcore!, I'm pretty sure it's even been mentioned early on in this thread that over 1.4v(actual I assume) on 45nm CPUs can kill them early, within a yr IIRC.
Or is that gonna be your good excuse for getting a better CPU?

Btw did you read my previous post about Prime95 & stabilty for F@H? P95 doesn't cut it.

 

[MikeTimbers]

Yes, I did and know well that F@H is harder on chips than Prime95. With over 16million points I've been around F@H for a while!

I don't get a lot of these posts about chips doing huge overclocks with little or no extra Vcore. In my experience overclocking since the C300A (I don't count when I used to change the crystals on 286-based motherboards!) if you're running any form of Distributed Computing, you need extra voltage or the chips will simply not produce good science.

This E5400 is possibly limited by being in a G31M mobo with PC2-6400 as the dividers mean that the slowest RAM speed for a 333 FSB is well over spec (889). Still, compared to blowing 9800GX2s, an E5400 dying on me isn't going to be a major blow!

 

[Assimilator1]

Ah your an old timer with F@H then , nearly 10x more points than me! .
Anyway,if you stabilty test with OCCT instead of P95 you're very unlikely to then need to boost vcore further for DC. And much less likely to get EUEs for F@H.

And yea large overclocks usually do need more vcore, goes without saying.
I was just pointing out the reckoned safe limit, if you're aware of that shortened life above that limit & are happy with it, that's cool.
And losing a 9800 GX2's is an expensive death!, was that from folding on it?

Ah you meant RAM dividers holding you back, got ya.

 

[Assimilator1]

Further to my comments on OCCT on my 2nd rig earlier, turns out that even with the std test the E5200 still needed 1.35v for 3.6GHz. It just took the std OCCT test a lot longer to find that out (although at lower temps) than the OCCT linpack test.
For example OCCT std ran for over 10hrs @1.34375v before crashing, whilst OCCT linpack errored afer just 16mins at the same vcore!

 

[nikzGfx]

Installed the e5400 along with the CM hyper 212, not fully had a chance to play with it but got it sitting on 3.8ghz at the mo.

Will play with it over the weekend. Might see what I can pull out otherwise I'll be happy to stick with 3.8ghz or lower at a lower core voltage. Feels fast enough for my needs anyway.

 

[MikeTimbers]

And losing a 9800 GX2's is an expensive death!, was that from folding on it?

It only cost me £55 on the bay but still annoying.