Best STABLE OC on new retail 45nm CPUs?

[tet5uo]

Highest stable for 8 hours of stressing on my qx9650 has so far been 4125 mHz with 1.42v actual.

[Speederlander]

I'm starting to question the 4.6GHz+ number...I just don't seem to see people getting those numbers ever posting stability tests beyond SPi, except for a couple of people (really only 1 that I can think of off the top of my head). Maybe a lot of phase people are sitting on their results? Because that's not going to happen without phase.

[aoch88]

I'm just as curious and would like to see 4.6Ghz+ stable screenie as well. Many people are just giving baseless claims. I guess for normal usage, the max. these chips can do are only around 4.3Ghz.

It doesn't make sense to spend around $200 for only a Super PI screenie. I rather get a E6550/E6750 and overclock the hell out of it. Only a 400-500Mhz difference.

[Spyrus]

My cpu is not the best one i am trying to run orthos at 4050@1.45V but after 45 minutes it fails. But i prefer it over my previous very good E6750 i had @ 4.2 benchable because it is colder at same clocks and have a nice boost with 2mb extra cache so it is worth the money difference (10 euros in greece)

[aoch88]

My cpu is not the best one i am trying to run orthos at 4050@1.45V but after 45 minutes it fails. But i prefer it over my previous very good E6750 i had @ 4.2 benchable because it is colder at same clocks and have a nice boost with 2mb extra cache so it is worth the money difference (10 euros in greece)

10 Euros is ok but over here in the stores, price difference is quite a bit. Btw, you didn't have any issues with temp readings on your board?

[Dumo]

I'm just as curious and would like to see 4.6Ghz+ stable screenie as well. Many people are just giving baseless claims. I guess for normal usage, the max. these chips can do are only around 4.3Ghz.

It doesn't make sense to spend around $200 for only a Super PI screenie. I rather get a E6550/E6750 and overclock the hell out of it. Only a 400-500Mhz difference.

You have to weight Vcore, cooling, FSB and capabilities of your ram too

Heres 1.14vcore...8500ES



4.4G 8500 retail, it stays for 6 horas too...

[Quentin73]

CPU: QX9770
Stability: 24/7
Cooling: Air - Thermalright SI-128 + 120mm (1000rpm)
vCore: 1.31V
MP: 9x
FSB: 450MHz
Mobo: Asus P5E3 Deluxe with Bios 1001

Overclock: 4050Mhz

Remarks: Limited by air cooling setup, still looking for a better cooling setup.

[stone_cold_Jimi]

Picking up what chew* was saying (hi m8), back in the heyday of AMD s939, you had to show 8 hours of prime per core before you went in the list with a 'stable' or usable overclock. Standards kind of went out of the window with Conroe, with people showing fast single instances of SuperPi 1M and saying "lookit that!". Well, Core 2 is fast at single instances of SuperPi because of the cache architecture, granted. But stability is way way down from those speeds.

Dumo, your overclocks are always solid heh

[Leeghoofd]

If you can get 4ghz rocksolid I think you must be happy, if you can bench it at 4.5 under good air or water you are pretty lucky... we all know things can get better but if almost everyone can get 3.8-4ghz out of a retail CPU with moderate voltages what's there to complain ? Some users couldn't even get 3.6ghz at moderate volts a few months back...

I like benches too and max OC's for earthshattering results but do you really think there's much gain on day to day basis by boosting ya CPU from 4ghz to 4.5ghz in games ? Before you need that power we will be a year from now (at least) and retail CPU's will be around 4ghz maybe then...

If I can get 4ghz out of my qx9650 at around 1.35 volts I'll be pretty pleased as my Q6600 needed 1.55 volts for that and could not keep those speeds for long sessions... I'll prolly run that CPU at 3.6Ghz 24/7 for folding purposes, less heat output, less wattage needed... that's what's it all about for me... suicide screens are nice but not needed for daily users... and that is what 45nm is all about, not to gain another 500Mhz or 1ghz more...over ya current OC

People's goals are really far from realistic and are blinded by the experts' runs... if Hipro5 does a benchsession at 4.5ghz , many think they can do that too and use those speeds day in day out... not even thinking George doesn't even use a case, uses very advanced cooling, mods, ... way beyond their knowledge of hardware... they just cranck up the volts and their CPU/mobo flies... and then come back a few months later ( if it takes that long ) that this mobo is a piece of sh*t or that CPU's is totally bonkers...

4Ghz 24/7 mates is a feast, do not use more than 1.35-1.4 volts on these CPU's for 24/7 abuse... benches are fun but just like mobile phones were in the days, only to be used for a short duration of time...

[KTE]

Problems with all this stable business is nothing but temps. Penryns do oc high, we know that but for 24/7, there are way too many factors to consider. Most of the Penryn temps are complete BS, a known problem. So you have no idea what temps you're running at whilst stability testing because the probes don't function accurately and this is one of the most crucial factors before calling something stable. I have my own opinion on this for my own system which I reserve since I only use it in my own systems or the work systems I manage, i.e. stay below 55C load at 22C ambient air to keep in mind low noise/high ambients throughout summer, but that can still be more lenient since there is headroom. The E8400 I had was reading 16C idle at 22C ambient with a heatsink and no fan. Still waiting for the next one.
The only one working sample I've seen out of the last batch we got delivered at work was an E8400 ES 3GHz 1.1V my boss had and it idled 34C and load was 50-51C on a Tunic Tower 120 with 22C heatsink ambient running Prime95 25.6 after 3 hours (Orthos gets it to 47C). And 1.4V 3GHz was topping 60C idle, so I know what to expect when I see these "stable" labels. Things are just too clear with IC physics; voltage will throw the TDP/heat up by the square and frequency will throw it up high all alone too.

IDK about the new Penryns since I've not played with them much at all but for a few minutes, but with the new AMDs, there is no way 8hrs P95 stable is actually stable. Game it heavily, bench it, load cycle it, reboot it a few days at those settings after 12-24hrs stability testing and you have it stable. Most of the ones that passed P95 8hrs with Phenom can't even game because it's idle->load->idle fluctuations (dynamic load) which shows its unstable rather than static load which is easy for it to do.

[brenanh]

E8200 max stable (12 hour Orthos small FFT) OC 4.2ghz (8x525).

Are the temps on these 45nm CPUs really a problem? At the end of the day they may be reported incorrectly, they may be high, but surely if you're 35 degrees off TjMax then you're OK? Surely it's set that high for a reason.
I ran a Prescott P4 at 85 degrees for 3 years without a problem, doubt that was as far off TjMax...

[Leeghoofd]

I really think temps are an issue but surely CPU voltage will degrade their performance more rapidly then their 65nm counterparts.... and really I don't see the point in running Prime for 12 hours as I have done this before and a game rebooted the pc in a matter of minutes while prime didn't show any errors... I really do some benches, play some games verify my temps regurarly and then when I'm not on the pc it folds on the cores available... if it survives all that then i call it stable ...

Prime and co is good to have an ida but you still can encounter instability with several apps...

[KTE]

Three rules:
Max safe silicon volts is always lower the lower process node you go.
Max safe silicon temperature is always lower the lower process node you go.
Silicon degradation incl. electronmigration is much higher the thinner the metal gate and the lower the process node you go (with higher current/volts).

So, at 45nm, you are going to have to stay lower especially with regards to voltage. That's why max Intel specified is 1.45V for 45nm and there have been many dead CPUs even under cold with volts just above this.

Staying 10C below 70C fully loaded in winter/summer is fine, but that's if you really are below it and depends on if you're one of the lucky ones with a working set of core probes. Most aren't and read unrealistically low values, unfortunately.

[brenanh]

I really think temps are an issue but surely CPU voltage will degrade their performance more rapidly then their 65nm counterparts.... and really I don't see the point in running Prime for 12 hours as I have done this before and a game rebooted the pc in a matter of minutes while prime didn't show any errors... I really do some benches, play some games verify my temps regurarly and then when I'm not on the pc it folds on the cores available... if it survives all that then i call it stable ...

Prime and co is good to have an ida but you still can encounter instability with several apps...

True, Prime is not the be all and end all of stability testing... But it's still a good test. I run games on it all day anyway, so I'd notice any instability there.

Sure, 45nm chips are more sensitive to voltage than 65nm, hence 1.45v max safe voltage instead of 1.55v. I can't see 1.45v degrading it much, given that 1.55v doesn't seem to have significantly degraded 65nm's.

Still don't think temps are as much of an issue as people are making out. They hardly change at all when overclocking, so it's not like the CPU is getting a massive increase in thermal load. Oh noes my poor IFX-14 with 3x 120mm fans can't keep the CPU <68 deg. stock, underclock it quick!

[brenanh]

Three rules:
Max safe silicon volts is always lower the lower process node you go.
Max safe silicon temperature is always lower the lower process node you go.
Silicon degradation incl. electronmigration is much higher the thinner the metal gate and the lower the process node you go (with higher current/volts).

So, at 45nm, you are going to have to stay lower especially with regards to voltage. That's why max Intel specified is 1.45V for 45nm and there have been many dead CPUs even under cold with volts just above this.

Staying 10C below 70C fully loaded in winter/summer is fine, but that's if you really are below it and depends on if you're one of the lucky ones with a working set of core probes. Most aren't and read unrealistically low values, unfortunately.

OK so 45nm need less voltage, but they're getting less. I seriously doubt 1.45v is going to kill my chip within a year or 2, just like 1.55v won't kill a 65nm.

They need less heat, but they run cooler too. Have you seen the puny heatsink they come with? I'm sure my lapped IFX-14 can do better. And if they run cooler, why is TjMax 30 degrees higher than 65nm?

My 65nm ran at 45 degrees with 1.55v and I seriously doubt this is running any hotter with 1.45v. In fact if you go by TjMax - CoreTemp I'm 35 degrees off max, where as with 65nm I was 32 degrees off. Sounds about right!

[Marios]

That's why max Intel specified is 1.45V for 45nm and there have been many dead CPUs even under cold with volts just above this.

Many dead CPUs???
Any links about this?

[Speederlander]

Many dead CPUs???
Any links about this?

Well, I don't know about his comment, but I've seen several of the bigger guns around here mentioning they killed chips. If I come across discussion I will post it. It does seem a legitimate issue.

[brenanh]

Well, I don't know about his comment, but I've seen several of the bigger guns around here mentioning they killed chips. If I come across discussion I will post it. It does seem a legitimate issue.

But has anyone killed a chip with =<1.45vCore?

[jaredpace]

well i can get to 4.5G and start up prime

it will crash at anywhere over 15 mins to 1h 20 mins

I'm still trying to tweak the vcore & mch




http://picasaweb.google.com/jaredpace/45ghzPrimingOnAir

[kiwi]

But has anyone killed a chip with =<1.45vCore?

Probably not. The ones I heard of all run LN2 and 1.9V+, could be even ES not retail

[kitfit1]

Mine is 8hr stable. Going for 4.7 at the moment.

[fwhomeboy]

The best I could do was 3.8 and I had all the settings you had. Like said, Lucky You !!!

My cpu is not the best one i am trying to run orthos at 4050@1.45V but after 45 minutes it fails. But i prefer it over my previous very good E6750 i had @ 4.2 benchable because it is colder at same clocks and have a nice boost with 2mb extra cache so it is worth the money difference (10 euros in greece)

[Xvys]

E8400 Q740A555T* Note the "T"* So far, nobody has been able to find out what the deal is with the "T" All I can determine is that some XS members from Canada and Australia have this letter at the end of their FPO/Batch#. Any more info would be appreciated
Pack Date: 12/20/07 VID:1.1125
Running 24/7@4005 445x9 Vcore:1.344v Gigabyte P35-DS3R

I got the exact same batch as you, retro...And running at the exact same FSB and vCore: 445FSB @ 1.344v (1.365v in bios). I have had mine up to 4600MHz @ 1.545v, but only SuperPi stable.

I'm curious about the "T" designation in the batch code, also. What does it mean, and why do some cpu's have it, while most do not? I have seen two posters from Italy and Greece with this T also...Perhaps T for Terrific?

[nama]

Not 8hrs,but atleast a screen shot.Took the last week off of overclocking this set up.Maybe try again this weekend.

Lets see some screen shots .

http://www.xtremesystems.org/forums/...&postcount=425

dude can you post you settings? i cant even get mine past 10 minutes

[Retro]

I got the exact same batch as you, retro...And running at the exact same FSB and vCore: 445FSB @ 1.344v (1.365v in bios). I have had mine up to 4600MHz @ 1.545v, but only SuperPi stable.

I'm curious about the "T" designation in the batch code, also. What does it mean, and why do some cpu's have it, while most do not? I have seen two posters from Italy and Greece with this T also...Perhaps T for Terrific?

http://www.xtremesystems.org/forums/...&postcount=799
799
Seriously though, you found two other countries with the mysterious "T". They are probably all scattered around worldwide.
Maybe somebody from XSF will shed some light on this. Seems that we have a very early Batch Code though.
"T for terrific", that would be good. You were able to hit an impressive speed there
Mine has been completely stable for a few days now@4005. At a reasonable vcore too.
Seen a lot of complaints about the temp readouts on the forums on these E8400's. How are yours? I have been completely satisfied with my "T's" readouts, idle and load temps in Coretemp seem very realistic. The cores only vary by a degree or two, no "frozen" temp readouts either.
Glad I have a Mr. "T" chip