E5200 Overclocking and performance comparison.

[pastymuncher]

Woah... Are all those CPUs just lying around?

I have the E8500 in my main rig but i did'nt have it when i did these tests. The E5200 is in my second rig clocked at 3.6Ghz. The E6600 was replaced by the E8500 so is soon to go on the bay as prices have increased for them. The Q6600 is long gone. Sold for more than i paid for it too!! The E4300 is sitting around waiting to go in a general purpose rig someone has asked me to build them. I also have a E2140 sitting around now. At £10 i just could'nt resist it.

 

[pastymuncher]

Excellent review there mate. I bought an E5200 on the basis of this for a build for a mate. Got 3.0ghz straight away and after a bit of "burn in" now running at 3.2ghz in an antec 300 so cooling isn't a prob but I'm still using the stock heatsink so thats gonna be replaced with summat better. Again, very infornative info mate, well done.

Great job buddy, surprised that you didn't get your E5200 further at those volts though. I have mine running at 4Ghz on 1.475v and have seen even higher clocked ones at those voltages.

Thanks for the comments guys.

1.475v is too much for a 45nm core. I could probably get mine stable with that voltage but there is no way i am killing this cpu. It's not worth it for 400 mhz. I killed a E8500 when they first came out and that was just with 1.52v.

 

[pastymuncher]

Hi Pastymuncher - what's next ?

(as one who follows a lot your work, both with admiration and gratitude)

cjph

Finding a source for pasties in Scotland!!

This was just a one off really. I did this at the time that they came out as a lot of people were undecided on what to go with. I thought i may be able to help people decide by doing some real world test's. I also wanted a new toy. Any test's i do with my current cpu would'nt be comparable as i don't have the GT anymore. I really don't want to start from scratch again as this took several days to do and benching is really boring. I won't be buying any new hardware for a while now as my rigs do everything (and more) that i ask of them and i am very happy with them.

I am planning on making a Mountain Mods type case from scratch to house the main rig in. Going to be made from 1.5mm aluminium. Currently in the planning stages at the moment. I will be building a new radbox for my window mounted radiator at the same time. The "other half" wants something that does'nt stand out so much. I will be taking the opportunity to fit a new rad, Probably a PA 120.3/Feser 120.3 so that i can use lower cfm fans.

 

[pastymuncher]

I clocked one for a mate to 3.75 GHz, 1.400v BIOS, CPU-Z showing 1.385 and temps 59/58 under 100% load. I'm under impression 1.4v is ok for 45nm, or is that still too high?

If so ill back it off to 3.6 GHz at 1.375v BIOS.

If these are similar to the E8400 (architecture, specs, etc) then I read that 1.4v-1.45v Vcore is about the max they'll take. i've a few posts from pl that they start to suffer electron migration above those voltages irrespective of cooling and any overclocks will start to degrade pretty sharpish. I reckon you're safe at a max of 1.45v Vcore, but any higher and degradaton might set in.
If I can get a decent cooler on this E5200 on me mate's, I'll try 1.35v Vcore and go for 3.6ghz. Incidentally, whats the highest fsb that these cpus can get to? I'm wondering if bandwidth performance would improve a fair bit.

Bang on. They are inded 45nm Wolfdale cores. 1.45v should be the max applied although i will not take mine past 1.4v. It's not nice having a chip die on you.

Mine actually hates high fsb's. I tried that while testing to try to run the ram at 1066. It will not go past 336-340mhz. I have seen someone do 420mhz but that was with a stupidly high 1.9v and phase cooling. When i was doing the test's i dropped the multiplier at 3Ghz (i think) and increased the fsb and with the extra bandwidth it did much better in the artificial benchies.

As you correctly say, it does'nt matter about what cooling you use, it will still end up dying. These things run pretty cool anyway. I only have a Freezer 7 Pro on mine and it's running nicely at 3.6Ghz at 27 degrees. I was playing with it earlier and i can now run it at 1.3v compared to the 1.325v it needed when doing the test's. That was a nice surprise. Maybe it will go further now. Might have a try over the weekend.

 

[pastymuncher]

Electron migration is not really affected by temprature (my understanding) though, it will happen even at subzero temps with high voltage.

Correct. *Temps do not make the slightest difference.* It's the voltage that counts.

Balls, you are only 0.045v away from the voltage that killed my first E8500. That was only run at that voltage (1.52v) for short periods and still degraded in less than 10 days.

***Edit*** Just to clear things up. As most people will realise, high temps are bad. What i was referring too was it does'nt matter even if you use phase cooling and the cpu is -10 degrees or whatever ridiculously low temps they give. If you still have a high voltage it will kill the cpu. Just to make myself clear so that no other smart alec can nit-pick. It's obvious that we meant it did'nt matter how low your temps are as people were stating low temps.

 

[pastymuncher]

Reading some previous posts I take it that a higher FSB with a lower multiplier will be faster than a lower FSB and high multiplier with this chip?

Yes but these cpu's are fsb limited as they really do not like high fsb's.

 

[pastymuncher]

I bit off topic, or maybe not.. Can someone show some test results or tell me what fps you guys get in games (Crysis, 1920x1200, v.high/high) with your e5200@~3.6-4GHz?
Just for comparision, I'm at the point of buying new cpu and I'm very close to decide to buy this little fella....

Not instead of your E8400? That would be a big backward step. You certainly won't get a E5200 to 4Ghz either.

 

[pastymuncher]

I'm asking for results - yo did a lot of tests on these, but do you have any for Crysis in high res and details?

No need to get shirty. You are asking for results due to buying a new cpu and settling on one of these. In your siggy you have a E8400 so i presumed it was to replace that. We have already had one person swap to one of these from a E8400. Even at the lower end the E5200 would'nt match a E8400. Crysis does'nt particularly matter as it is a gpu limited game not a cpu limited one. It's a poorly coded game anyway.

No i don't have any results. Try the Crysis benchmark thread to see if anyone is running one.

 

[pastymuncher]

I beg to differ, this article at least shows that voltage and temperature affect CPU life.
E8500 review

(Note I'm not arguing the fact that seriously over volting a CPU will kill it quick).

Hopefully I'll be buying a E5200 soon for my 2nd PC, I look forwards to seeing what it'll overclock to , & curious to see if their is a low FSB wall......

What i meant was it did'nt matter how low the temps were. It is obvious that high temps are bad.

 

[pastymuncher]

Fair enough

Correct. Temps do not make the slightest difference. It's the voltage that counts. - pastymuncher

Pieman109
The point being made by you & pasty was that temperature makes no difference to cpu life - not true (apparently).
Extra heat is naturally also a by product of increased clock speed, although extra vcore has a much bigger impact (which you did allude to).
I think your main point originally was that overly high vcore kills quickly, totally fair enough & I totally agree , I was just saying that temperature does have some impact to CPU life too.

Btw I have a 9 1/2 yr old Celeron 366 @550 that still works , the beauty of that CPU & many of it's family was that it only needed a tiny increase in vcore to hit that speed (def 2v ,increased to 2.05v). Of course that rig is only any use as legacy PC these days, & light internet use at a push .

You have taken us out of context. People were saying thing's like, i am running at 1.4v but my cpu's nice and cool so still have some headroom left yet. It was then that myself and pieman posted about temps. It was obvious that we meant it did'nt matter how low the temps were and that the voltage would still kill them. These are 45nm Wolfdale cores and i have first hand experience of one degrading due to voltage despite being on high end custom watercooling with a window mounted rad. Temps never went above 40 degrees even at load but the cpu still degraded in less than two weeks.

Incidentaly, a cpu should throttle long before it cooks itself anyway. Unless you turn thermal throttling off in the bios that is. Personally i would'nt want temps over 55 degrees on any cpu, air or water cooled.

 

[pastymuncher]

Any of you guys running these babies with a GTX280? Wondering what overclock I need to maximise the new GPU effect and if I should ditch the E5200 altogether. Loathed to upgrade CPU again with Core i7 out - though I wont be touching that until you can build a reasonable rig for £500 (say 2010).

My setup should get the best out of my chip tho (P5Q-Deluxe, Lapped TRUE with 2 P12s and some decent general airflow).What kinda 24/7 stable o/cs might get on such cooling?

ps3ud0

People who buy these tend to be on the budget side (tight git in my case) and not team them up with a 280. Mines paired up with a 3850. Other's are more likely to be with a 4670/9800Gt or similar. There was a guy further up though that ditched a E8400 for one of these. I believe he had a 280. Not phinix though.

Clocking wise it's a bit hit and miss. Mine will only do 3.6Ghz prime stable as i will not give it more than 1.4v. I can boot into windows and bench at 4ghz but it fails prime within a hour. Tried it in two boards as well as water and air cooling. Some people are stuck at 3.2ghz and other's are at 3.75ghz with good volts. There's no way to predict what anyone will get. Does'nt seem to matter what steppings you have either.

 

[pastymuncher]

Excellent post, must have taken you ages to do all those benchmarks. Thanks!

E5200 is pretty high on my list for a new build now

Yes it did. They wanted me to update it with full E8500 benchies too.

Luckily i don't have the 8800GT anymore.

 

[pastymuncher]

You got me really concerned here. I was thinking about getting a dude's beat up case with wc attached to it which seems to keep his q6600 at reasonably low temps.

However, would it be worthy of pairing up watercooling with e5200 for 24/7oc for 1, 1.5 years? from what I've read, it serves me no good to have the temps low as long as long as voltage is high (safe would be <1.4 ?).

Thanks.

As long as the voltage is at 1.4v or lower you should be ok. All of my rig's are overclocked from the day i build them. My main rig in siggy was built and clocked to 4.5Ghz within 48 hours. It is high end custom watercooled and temps are extremely low. Never above 40 degrees on the cpu even under load. Even the 4870 which is quite a hot running gpu under normal conditions never goes above 35 degrees.

You have to remember that these are 45nm Wolfdale cores, even the E5200, and are very sensitive to high voltages. The point we were trying to make above was that some people were running their cpu's with 1.5v+ but saying it's all right as the temps are nice and low. It makes no difference if you have it under phase cooling and it's running at -30 degrees. The high voltage will still kill it. If you stick below 1.4v it should be fine.

Incidentaly, mine is now doing 3.6GHz prime stable with only 1.3v compared to the 1.325v it needed when i did the test's. The only change i have made is to flash to a newer bios on the Gigabyte P35 DS3R that it is now in. Maybe it has a bit more overclocking headroom to come now. may have a go this week.

 

[pastymuncher]

I'm only at 3.2Ghz at the moment having been pretty short of time recently, thinking of shooting for 3.6GHz tonight. Wish me luck.

Good luck. We need a smiley with a thunbs up.

I hope you make it. Been losing a bit of faith in these lately with people being stuck on 3.2Ghz and can go no further.

 

[pastymuncher]

Pasty
Awesome o/c on the 8500! , that's gotta be right near the top end of what's achieveable isn't?

Gurusan's does more than mine. I do believe it's got more in it though as 4.5Ghz was easy. I just can't be bothered to try. I like using the pc for gaming rather than keep going for that little bit more. Maybe one day i shall try.

 

[pastymuncher]

Just noticed the E5200's have gone back up in price again. The price cut did'nt last long did it?

 

[pastymuncher]

Thanks for the reply. For the moment, I think I'll get a Zerotherm fz-120 to cool the e5200.

As for the board, I'm choosing between the cheaper Gigabyte EP31-DS3L and the more expensive MSI P43 NeoF. Would the first be OK?

The Neo-F boards are abysmal. I have had a the misfortune of having both P31 and P35 Neo-F boards (until they were rapidly returned). Very cheap and nasty bug ridden pieces of junk.

The Gigabyte EP31 DS3L on the other hand is an exceptional board. It will put many more expensive boards to shame. It overclocks briliantly. I had a E2180 to 3.6Ghz on one. The only downside is that you can't use more than a pair of single sided memory sticks.

 

[pastymuncher]

I'm now at 3.8GHz with 1.2V Vcore, 2 hours Orthos stable.

Pushing for 4.0GHz later on.

1.2v. That's fantastic!! You have a 4GHz jobbie i reckon. Got to be 8 hours for absolute stability though. I have had cpu's fall over after 6 hours before.

 

[pastymuncher]

D'oh! Correction, it was actually 1.25V. I'm still really happy with it.

I'd normally give it 8 hours to test, but I'm feeling wreckless and want to go straight to 4.0

Even 1.25v is fantastic for 3.8Ghz. Mine needs 1.3v for 3.6Ghz.

 

[pastymuncher]

My Neo2-FR has served me brilliantly for well over a year now

The Neo2-FR is a completely different board from the Neo-F's. The Neo-F's are basic, cheap and nasty rubbish. Just take a look at them on MSI's website.

P35 Neo-F

P45 Neo-F

For some reason they only list them as Neo on the website yet the boxes say Neo-F.

Looks like she's a good'un!


5 and a half hours and still going strong.

Very nice. Do you reckon it's got any more to give? Maybe a 4.2-4.5Ghz? Temp's are getting high though.