How well does the e5200 overclock?

[nyker96]

Originally posted by: VirtualLarry

Originally posted by: Foxery
12.5x 275 = 3437 MHz @ 1.30v (BIOS)
CPUz reports 1.31v under load (??)

So far so good, 3 hours in. Temps are still only 60C! (using RealTemp 2.7) The lack of transistors with the small cache make this thing very low-power and low-heat. I love it.

Sounds good. Maybe you can make it to 3.8 @ 1.4v?

I heard the E0s allow about 200Mhz more on same vcores at least it did that for the E8xxx series, so we can expect as much. I'd say yeah 3,8@1.4 is possible with E0. But I be honest 1.4 is as far as I want to go on a 45nm running 24/7. that's it, if it can't dance at that volt then I'll just live with the result.

 

[Acanthus]

Just checked in on the thread again.

You guys are pushing it with both volts and temps...

It is a cheap chip, but dont expect it to last you 2-3 years if your breaking 1.4v actual and temps are over 70C.

 

[VirtualLarry]

Originally posted by: Acanthus
Just checked in on the thread again.

You guys are pushing it with both volts and temps...

It is a cheap chip, but dont expect it to last you 2-3 years if your breaking 1.4v actual and temps are over 70C.

70C? Bah. These chips can take more than that 24/7. I don't know about 1.4v actual, I had heard that was safe. That's my limit though.

 

[demiurge3141]

Seems rather disappointing so far, I wouldn't bother getting one unless they can consistently hit 4.0+ at 1.35v.

 

[SonicTron]

Anyone have problems with E5200's not able to take higher FSBs? Mine tops out at ~3.5GHz but the max FSB I can hit is 310. So like, 310x11 works. But 315 x 10 doesn't work. 333x any multi no go. 400x any multi no go. This is pissing me off.

 

[MyLeftNut]

Yeah, I think there was another guy at XS that couldn't get very high FSB on these either, but not as low as yours. These chips pretty much don't go very high on fsb. How much VTT are you putting through it?

 

[SonicTron]

I tried everything 1.2 to 1.5v vtt, no change in ability to even POST. Its stable at 310, pretty much anything over it just blacks out on POST.

 

[Acanthus]

Originally posted by: SonicTron
I tried everything 1.2 to 1.5v vtt, no change in ability to even POST. Its stable at 310, pretty much anything over it just blacks out on POST.

yeah, the chip refuses to do even 400x6.

the same system with an e2160 has no trouble doing 400fsb.

 

[SonicTron]

So guys I think its pretty safe to say that Intel has crippled the E5200's ability to scale with high clock speeds correct? I haven't seen any confirmed reports of 400fsb....but tons and tons of confirmed reports of low fsb walls

 

[Foxery]

Originally posted by: SonicTron
Anyone have problems with E5200's not able to take higher FSBs? Mine tops out at ~3.5GHz but the max FSB I can hit is 310. So like, 310x11 works. But 315 x 10 doesn't work. 333x any multi no go. 400x any multi no go. This is pissing me off.

Er, who cares? They come with a 12.5x multiplier. 12.5 * 310 = 3875!

Originally posted by: SonicTron
So guys I think its pretty safe to say that Intel has crippled the E5200's ability to scale with high clock speeds correct? I haven't seen any confirmed reports of 400fsb....but tons and tons of confirmed reports of low fsb walls

It's the cheapest budget chip they produce. No intentional "crippling" is needed.

They probably gave it the low FSB because these chips couldn't be binned as 8400s, 7200s, etc, and would otherwise have been thrown in the trash. We get 3.5+ GHz for $80... everyone's a winner

 

[nyker96]

Originally posted by: SonicTron
So guys I think its pretty safe to say that Intel has crippled the E5200's ability to scale with high clock speeds correct? I haven't seen any confirmed reports of 400fsb....but tons and tons of confirmed reports of low fsb walls

I think Intel just left the lowerest binned 45nm dual cores to be E5xxx line, that's all. Just low quality that's preventing you from clocking it higher. But at the price they are selling, it is very hard to complaint, even though we all keep complaining about it anyways :]

But who knows maybe in another 6 months of process maturity/a new rev or two this chip would hit 4 for under 1.4, but that will mean the higher end lines will probably have to be able to do 5Ghz or something crazy by then for this to happen.

 

[Tempered81]

Originally posted by: SonicTron
So guys I think its pretty safe to say that Intel has crippled the E5200's ability to scale with high clock speeds correct?

No, just lower bin, bottom of the barrel. Not a "select" chip, for E8600 status, also not of the E0 stepping.

Many reasons, one of them is not to "purposely cripple", although that is what many might believe. Imagine they have an assembly line where chips are shot out, and then tested in a motherboard or socket in intel's labs, they are cranked up to a high speed & measured in terms of temperatures & required voltages. They are then classified according to their performance("binned"), as no chips perform identicle and highest speeds are a luck of the draw scenario. If any individual 45nm dual core penryn is a horrible performer on intel's scale, it is immediately binned into E5xxx status, or the lowest speed SKU (in the future maybe E3xxx). If it is highest, it is binned E8500. If the wafers produce chips in the E0 stepping, then the lowest performers are currently binned E8400's, the highest are binned E8600's and beyond into future sku's. E5200's are fabbed in Costa Rica & bear the Pentium dual core moniker, E0's are fabbed in Malaysia. The OC is the luck of the draw, this is why many pro overclockers purchase dozens of E8600's to test each of them, keep the best one, and ebay the rest. They also notate which batches produce the best performers, even down to typically which chips perform the highest based on which area of the wafer they are cut from.
It's insane. these 4.8, and 5.1, and 6.77ghz Penryn's are obviously top shelf, highest binned, E0, E8600, and cut from the center of the best batches, etc...

A bad E5200 might reach 3.4ghz at 1.40 volts, while a good E8600 could reach 4.6-4.8ghz at 1.40 volts

 

[brencat]

So based on all of the comments so far, it looks like the avg OC for these is 3.3 - 3.6ghz with reasonable voltage. Bah, not worth the hassle to upgrade from E2180 @ 3.2, IMO.

 

[faxon]

well i finally got the new system set up, i had to bring the comp from my buddies house to home and remove a crossbar in the case with a dremel and a drill so the xigmatek cooler would fit properly. im still working it up to its full OC potential, currently sitting at 3.25GHz at 1.36 core voltage (CPU-Z 1.344), but i havent upped the voltage since i started. will post some updates as i go as sort of a log for myself

ed: ran into a ram barrier at 3.5GHz (280FSB), had to unlink the ram to keep it stable. system would post but would instantly reboot upon loading into windows

ed: wouldnt work with them unlinked for some reason, reset FSB-RAM to 3-2

ed: finally got it to boot at 3.5GHz with 5:4 FSB-RAM ratio after a good 25 minutes of tinkering

 

[Shmee]

nice oc so far, faxon. I am not sure how good that ram is, may be why I was having trouble with it when it was with the other ballistix, cuz i know its a newer batch. when I get some new memory, ill be able to put my good sticks in the system, the ones that will reach over 1100 with ease. for now, what did you have them on when unlinked? cuz they seem to have had trouble running at 1066, may just want to set them to 800 4-4-4-12. thats what I had them at b4 you left with the comp. anyways, good to hear you got the oc up there allready.

anyways, when you bring it back, we can play with it somemore, mb get some better ram.

 

[faxon]

already consulted with Shmee on AIM, but yea i agree with him on them having trouble running at their stock settings stable. with the FSBRAM set to 5:4, the ram is running at 896MHz at really loose 5-6-6-18 (mobo auto set). i think im also having trouble with the motherboard in that there is no way to disable Vdroop on a 680I without doing hardmods, so the voltage i give it is getting altered by the bios to something lower than what i told it. from what shmee tells me, very few mobo's actually make it easy to disable vdroop, so im probably going to leave it at 3.5GHz with 1.408 core voltage (CPU-Z) and call it a day after i tighten the ram latencies down. will let you know what my prime95 results are tomorrow

ed: core 1 failed prime95 outright, core2 is still.....nvm the comp rebooted

 

[demiurge3141]

Originally posted by: faxon
already consulted with Shmee on AIM, but yea i agree with him on them having trouble running at their stock settings stable. with the FSBRAM set to 5:4, the ram is running at 896MHz at really loose 5-6-6-18 (mobo auto set). i think im also having trouble with the motherboard in that there is no way to disable Vdroop on a 680I without doing hardmods, so the voltage i give it is getting altered by the bios to something lower than what i told it. from what shmee tells me, very few mobo's actually make it easy to disable vdroop, so im probably going to leave it at 3.5GHz with 1.408 core voltage (CPU-Z) and call it a day after i tighten the ram latencies down. will let you know what my prime95 results are tomorrow

Somewhat disappointing oc as I had high hopes. You think it's the chipset? How would it go on a P35 board?

 

[faxon]

yea its probably the chipset, shmee replaced his 680i because it was giving him trouble OCing his Q6600 as well. looks like im going to have to tone back the OC a bit since it rebooted when i started testing it in prime95. like i said before, the vdroop was giving me a hard time getting the voltage where i wanted it, so its possible it wasnt getting enough voltage and i wasnt seeing that in cpu-z. either way, a good p35 DFI board would probably solve any OC problems im having right now hehe. once i build my new system for at home to replace my a64 4000+ i will try dropping this chip into it for a bit and see if i can get it to go higher before i drop it back on the 680i and put a quad core in it

ed: core1 failed prime95 again at 3.4Ghz. test ran 4 minutes. im going to see how long core 2 can go for, just to get an idea of the difference in stability between the 2

 

[Shmee]

well, I would hope it would do better with a p35, but I think the key may be either a vrop mod, or you may have neared the chips limit. hard to say, mb you just got a not so good chip. still, I think 3.4 is a pretty good oc, also, you can try playing around with bios settings a bit more if need be. also, keep in mind we are currently using an EVGA bios, due not being able to find the latest from xfx.

BUT, just a question, not sure if this would work, or even help, but does anyone know if the p33 bios works on a 680i LT? keep I know the p31 did, and mb the p32...

right now it is running the p08 bios, the latest one for the LT board.

I know some bios flashing can unlock better features, such as better V control. I would hope an nvidia mobo expert could help us here.

also, a vdrop guide might be nice

 

[faxon]

eh i had to drop it to 3.3 to get core 1 to not fail the prime95 small FTT test. both cores currently going strong at an hour and 15 minutes this time, going to let it go overnight and on for another. im betting i just didnt get the best chip of the batch. no biggy though either way, 3.3Ghz should hold fine with an HD4670 and 2GB of ram for gaming in XP, and if the system needs more ram in the future, DDR2 is going to hit rock bottom soon as DDR3 goes mainstream.

 

[Zap]

Originally posted by: SonicTron
Anyone have problems with E5200's not able to take higher FSBs? Mine tops out at ~3.5GHz but the max FSB I can hit is 310. So like, 310x11 works. But 315 x 10 doesn't work. 333x any multi no go. 400x any multi no go. This is pissing me off.

Has anyone tried a BSEL mod on it? Worth a shot.

 

[faxon]

both cores prime95 small FFT stable 13 hours 0 errors 0 warnings at 3.3GHz (264x12.5) with 1.408 core voltage (CPU-Z). highest recorded load temp was 64C, lowest was 56c, room temperatures ranged from 65F to 80F during the test.

ed: running intel burn test now

 

[Shmee]

sounds good, though idk if you need that much vcore for 3.3 once you get it stable, which it sounds pretty good now, try backing it down a tiny bit.

 

[faxon]

i needed that much vcore to not get it to BSOD on loading windows lol. it was okay at a lower voltage before, but once vdrop started kicking in the actual voltage going to it was much lower than what i was originally using, so i had to up it to get it stable. i think it was at like 1.39 or something in bios even though i had it at 1.41xxx

ed: looks like it failed intel burn test (1/10) after running for 1271 seconds. temps were stable at about 61c the entire time.

 

[VirtualLarry]

Originally posted by: brencat
So based on all of the comments so far, it looks like the avg OC for these is 3.3 - 3.6ghz with reasonable voltage. Bah, not worth the hassle to upgrade from E2180 @ 3.2, IMO.

Well, I'd say 3.5-3.75 instead, but you know what the funny thing is? Running SeventeenorBust (a distributed computing program), it gives me 6.6M per core running on my friend's 3.75Ghz E5200, and it gives me 7.0M per core running on my E2140 @ 3.2Ghz.

Figure that one out. So all in all I agree with you, I don't see a major need to upgrade. ESPECIALLY if you are running apps that benefit from a high FSB speed, that you can achieve on a 65nm chip, but not on an E5200.

My results are even stranger, because the E5200 has twice the L2 cache as the E2140.

Edit: I've had one report that Intel's web site is wrong, and that the E5200 does NOT have VT extensions. Can anyone else confirm or deny? He said his BIOS showed that it didn't, I just wonder if he had an older BIOS revision or something.