Intel 45nm Quads: "Safe" Voltage Range

[Foxery]

edit: Not fried after all. Yay!


I seem to have damaged my shiny new Yorkfield CPU, and I wonder whether voltage tolerances are a hidden difference between the Xeon models and Desktop models, or if I just pushed my luck too far.

A few forum searches show other folks hanging around 1.25V. I ran an X3320 (Q9300 equivalent) around 1.35V with a heatpipe keeping temperatures nice and chilly. It ran great for 6-7 weeks; now the box crashes every few hours at any speed, any voltage, including stock. I dropped my old dual-core back in over the weekend, and it runs fine. Boo, $300 paperweight.

So, per http://processorfinder.intel.com/
The Desktop versions have a whole range of 0.85-1.3625
The Xeon versions are rated for a single voltage, 1.212
(This is true for the 9300, 9450, and 9550)

I knew I was on the high end - hence why I wouldn't push any higher - but didn't expect to burn it out this fast! Your thoughts?

 

[TC91]

not convinced its the cpu that damaged, did you stress test it (like quad p95 small fft's 24hr) when you had it oc'ed/overvolted? check the ram with memtest86+ too, it could be the cause of crashes.

 

[bastula]

1.35v? it should be fine up to there, but no more than 1.4v honestly...

 

[Foxery]

Originally posted by: TC91
not convinced its the cpu that damaged, did you stress test it (like quad p95 small fft's 24hr) when you had it oc'ed/overvolted? check the ram with memtest86+ too, it could be the cause of crashes.

Stress tested with Prime95, OCCT, and primarily Folding@Home for weeks on end
I checked the RAM with HCI's Memtest for ~3 hours before swapping in a different CPU.

Identical system except for the CPU has been humming along fine for 3 days, also with a mild overclock. I'm positive it's the CPU.

edit: Just in case anyone asks, the motherboard is an IP35-Pro with the most recent official BIOS, which explicitly supports all 45nm processors. The system was quite happy for the first few weeks.

 

[Foxery]

Upon closer inspection, I am an idiot. The retail box for my Xeon says "1.25V max" on the sticker with all the bar codes and serial numbers.

Anyone else care to chime in whether their Xeon models say this, and whether desktop (Q9xxx) models say something else? Maybe I've found a tangible difference between these "equivalent" processors. If only it hadn't cost so much, eh...

 

[Tempered81]

if the cpu fries because of overvolting, it wont post, boot, and let you run benches

 

[graysky]

@op - what was the VTT you had it set to?

 

[Foxery]

I left the VTT at stock for most of its life. I tried bumping it up a few notches when the system started crashing, (tried up to 1.39v after reading a few threads here about it) but at that point it didn't make any difference where I set it. Ditto for FSB voltage - if I changed it at all, it was barely above stock for the first month of operation, and boosting as high as 1.56v didn't bring me any more or less stability once things started to go downhill. Ditto again with GTL REF settings.

 

[Idontcare]

Sorry to hear of your loss Foxery. Sounds like you might have had a chip that was on the edge of its stability all along (would have died prematurely even at stock) and just some slight OC'ing caused it to have accelerated lifetime issues.

 

[M1A]

My E8400 boxed processor says 1.225v max but I am above that and many others are also. Been at 467 fsb since Jan with no problems.....
lots of people are running them 1.36,1.38 etc with no problems but I am not sure the quads will take the voltage like the duals.....

 

[M1A]

1.56v is getting up there I guess, I just seen that in your last post.

 

[gramboh]

m1a what kind of cooling do you have? That is a badass OC on the 8400.

 

[Tempered81]

m1a, what stops you from going to 475x9. ~ 480x9.? Is it > 73C load temps, or > 1.400 Vcore, or both that are holding you back?

Also, i think the op said 1.56v on his mobos FSB voltage.

 

[M1A]

As you see on my sig its on air, and not even true or other high end cooler its an older Zelman 8000. The lapping helped as did the bios 16 for the motherboard. And after the mix of artic silver and 7 diamond paste its been stable without issue since. I got an early chip in Jan and have been running daily since. Dominator memory has also helped get there and I have 4 sticks of ballistic just sitting on the shelf.

 

[M1A]

Acually both kind of. I will not run at 1.4v core and want to stay below 70c on air. Maybe in time and if I see others doing high volts I will push it more but for now its good here. Also we are getting into higher room amb here in Pa and I just want to be safe. I feel bad for the people that are stuck below 4ghz but i think a lot is the memory or other issues than these wolfdale chips but I could be wrong.

 

[Markfw]

My X3350 was @1.3125 for weeks 100%load@3576, I just put it back in service on another motherboard, and now, even at stock speed, re-boots every few hours ! But F@H doesn't think there is anything wrong with it. and I have the memory thing un-checked to auto-reboot. Very odd.....

 

[Tempered81]

Originally posted by: m1a I feel bad for the people that are stuck below 4ghz but i think a lot is the memory or other issues than these wolfdale chips but I could be wrong.

+1 my thoughts exactly

 

[aigomorla]

just for sakes, can you run memtest on the ram?


My first guess is F@H killed your ram after many weeks on load.

Second guess is degrading board because of a loaded quad. It holds duals but not quads. If you have another board, now would be the best time to test it on another board.

 

[Foxery]

Originally posted by: m1a
1.56v is getting up there I guess, I just seen that in your last post.

That was my highest attempted FSB speed. The CPU would never survive that.

Originally posted by: Markfw900
My X3350 was @1.3125 for weeks 100%load@3576, I just put it back in service on another motherboard, and now, even at stock speed, re-boots every few hours ! But F@H doesn't think there is anything wrong with it. and I have the memory thing un-checked to auto-reboot. Very odd.....

This is exactly what I'm seeing. Either a spontaneous reboot, or the screen goes black and the system stops responding. (Windows is set to not auto-reboot on BSODs.) I'd be curious to hear if yours gets any better or worse.

Which motherboards did you switch from -> to?

Originally posted by: aigomorla
just for sakes, can you run memtest on the ram?

My first guess is F@H killed your ram after many weeks on load.

Second guess is degrading board because of a loaded quad. It holds duals but not quads. If you have another board, now would be the best time to test it on another board.

I used HCI's version of MemTest (Link) twice - with the RAM overclocked, and at its stock settings - and it passed several hours each time. I can steal a floppy drive to run the classic Memtest version if necessary. I pondered the motherboard too, but figured a stock quad and an overclocked dual should be close enough that it's unlikely.

The same RAM and mobo have been running F@H for 4 days straight now with no interruptions, and did so for several months before I bought the new CPU.

I called Intel for shits'n'giggles. (I'd feel like a dick for RMA'ing one I broke, but...) Because Abit's web site does not show this mobo supporting any model of 45nm quad, they aren't interested until I test it in another mobo which they can verify as "compatible." Oy. It made zero difference to them that it ran just fine for 2 months.

There's a slim chance that I could actually do that. However, customer support that barely speaks english only encourages me to abuse the warranty. /sigh

 

[Markfw]

Originally posted by: aigomorla
just for sakes, can you run memtest on the ram?


My first guess is F@H killed your ram after many weeks on load.

Second guess is degrading board because of a loaded quad. It holds duals but not quads. If you have another board, now would be the best time to test it on another board.

This DQ6 was running a Q6600 B3 for months, and up until I sold the B3, and moved the X3350 over, it was fine 24/7.

The old motherboard was a DS4, and its running fine in that same box the DQ6 was in.

And 2 different sets of memory that run fine in other boards ??? Both bad ?

Something is up here....

 

[Markfw]

update ! I updated the BIOS, and it stopped re-booting ! The F7 bios was support for 9770, but I suspect all yorkfields get a boost.

 

[aigomorla]

Originally posted by: Markfw900
update ! I updated the BIOS, and it stopped re-booting ! The F7 bios was support for 9770, but I suspect all yorkfields get a boost.

QX's are C2 steppings. :\

 

[Markfw]

Originally posted by: aigomorla

Originally posted by: Markfw900
update ! I updated the BIOS, and it stopped re-booting ! The F7 bios was support for 9770, but I suspect all yorkfields get a boost.

QX's are C2 steppings. :\

I meant, whatever they did affect all the 45 nm chips, it sure fixed my problem, I didn;t tweak anything in bios, and it wouldn;t even stop rebooting at stock before, now its almost 24 hours@3520 full load, no reboots !!

 

[Diogenes2]

Originally posted by: Markfw900
My X3350 was @1.3125 for weeks 100%load@3576, I just put it back in service on another motherboard, and now, even at stock speed, re-boots every few hours ! But F@H doesn't think there is anything wrong with it. and I have the memory thing un-checked to auto-reboot. Very odd.....

oops.. Nevermind ...

 

[bryanW1995]

Sorry I missed this before, foxery.

I noticed a huge vdroop on my ip35pro with the x3350 when I first started oc'ing, it was 1.41 bios/1.27 load in fact at 3.6. However, with the 16 bios it was 1.41 bios/1.35 load, so I've dialed it down to 1.34 bios/1.28 load now. My 9450 is in an ip35 e, but I only run it at 3.40 b/c it has a zalman instead of a tuniq. I use the 16 bios on that one, 1.305vcore bios, 1.24 load. I should upgrade to bios 17 on the ip35e I know, but it's been 24/7 stable with bios 16 so I've left it alone.

There are several beta bioses available for the ip35 pro, you might need to download one of those. Abit has been pretty adamant that they don't officially support 45nm quads on the ip35 series, maybe you just got lucky and brought out one of the flaws that hasn't been fixed yet. try 17b04 and see if that helps any.

http://forum.uabit.com/showthread.php?t=122393

edit, also, here's the official ip35 pro thread at abit-usa

http://forum.uabit.com/showthr...hp?p=966478#post966478

abit says that they don't officially support 45nm quads YET for their ip35 series, but the 17 bios is supposed to fix that on the ip35 pro. the 16 bios offers "enhanced 45nm compatibility" and I've been golden with it, but it's also only been a couple of months so we'll see how it goes for the long term. I am pretty comfortable with my voltages, however, and seti isn't as stressful as folding, so hopefully I'll be worry-free going forward...(crosses fingers)