P4 D 805 - Safe Operating Temp?

[beersnob]

I just plugged in my new P4 D 805 with intentions to o/c, but wanted to verify first what the safe operating range is. Intel's site is useless...64 degreec C ? You've got to be kidding. I briefly get to 66 (at stock settings) just by running PCMark05 Pro. And I'm using the Zalman 9500.

This chart tells us what THW was able to achieve without torching it during testing, but are these temps safe for long term operation?

I'm normally an AMD guy, so I'm unfamiliar with the thermal idiosyncrasies of Intel chips. At least AMD provides a meaningful thermal max for their chips.

 

[pengwin]

65 is ok for Intels i would think. my old P4 gets down to 35*C on a zalman 7000-CU LED.

 

[beersnob]

Don't the operational thermal ranges differ from P4 to P4, depending on which core it is (among other things)?

Also when I ratchet up the FSB, heat production will also increase past it's current peak.

Asus Probe (I've got the P4WD2 Premium) starts to issue warnings when it gets above 60. I know I can tweak the threshold, but where should it really be?

 

[icbluscrn]

Ive taken one of mine to 80°c for some time during some testing, Tm2 seems to kick in at 80°

I think the 64°c is a spec they have set for nominal operation otherwise there would be noise issues with the stock cooler, everything i have read has led me to that fact.
itel's actual spec sheet for the d805 they dont give a certain temp° like most of there other cpus they referer you to there own thermal profile chart that goes by watts.

I like to have mine at 67°max after hours of dual prime95.

as far as your set up what voltage are you at? you should be at ~1.35v hitting 66° at stock 2.66ghz is high.

 

[jasonr49]

I picked up the Zalman CNPS9500 and while Idle, and overclocked, I'm at 40C.

When I exit a game I'm at like 46 approx, it's probably a little bit higher cause it takes a few moments to exit the game then open the Asus probe to get the reading. So give or take, maybe 50 tops. And like I said, that's overclocked to 3Ghz.

When I initially used the intel stock cooler, the temp was 50idle at 2.66Ghz. (My cooler came in 2 days after the mobo / processor)

So bottom line your temps seem high.

I also picked up some OCZ 5 ? I can't remember the exact name, but I put some of that on the processor and I believe that also makes a huge difference.

 

[Gavman42]

I'm running a couple stock at work. They seem to hover around 60 for me. They are at 50-80% load most of the day... ONly fans in the case are the cooler and the psu fan. Ive had no problems yet.

 

[FLA94FD]

The chip wont let you kill it because it will just throttle back. So over clock away. Also this info is pointed out in the article.

 

[Grimmy]

I think P-D CPU's can handle the heat, but take care on that.

I remember a thread on CPU forum, though I can't find it at this time, to where a water pump system on a P-D failed. There were pictures taken of the CPU, on how badly the chip was burned and fused to the bottom.

What basically happened from what I remember, it didn't throttle back, the guy let it run for a certain amout of time thinking that the feature would save it, and it didn't shut down from any thermo trip. Thing is, I can't remember what the top recorded temp he got at, or if the bios was properly set to shut off on a safe temp.

He took the system apart, and it was a burnt chip. Since the pins are on the MB, he said the CPU was fused on to the pins. He said the chip was broken after trying to get the chip off the socket, and that the CPU prolly would have continued to work, if he didn't take the CPU off the socket.

So I guess you need to be sure that everything is setup properly if your going to rely on that throttle back feature.

Edit:

Found it, though the story is INQUIRE based:

When CPU Coolers Fail

 

[icbluscrn]

I actually did that on purpose just to see what would happen with the 805. At the time it did not even cross my that i could of blown the chip. Removed the hsf then powered up, Id say less then 30 sec it shut the pc off almost burned my finger cause i had it on the cpu just to see. I never thought it would get that hot that fast.
Its been fine since then happy at 4040mhz.


When i get home ill try shutting the pump off when i am in windows to see what happens.
oh the joys of a cheap cpu i would not try such a stunt with my 3800x2 .

 

[Grimmy]

After re-reading the article, the temps I guess he saw "96 Degrees celcius and rising"

I edited it after you replied.

I realise that this is a 805 vs 840 system, but it is the same feature?

He's a quote of what the user discribed on his P-D:

I recently experimented with WaterCooling on a Development 840 System I had on the bench, all was fine until one day when I was transfering some data between hard drives while the system was unattended , the Water Pump failed, no biggie, the system would heat up to a point and shut itself down , Right.. ??

Wrong.. , I stepped back into the room and noticed a slight smell of burning, nothing too alarming until I dialed up Probe on the desktop and noticed the Temp was at 96 Degrees celcius, and rising.., btw this system would sit comfortably around 35 degreee when not under laoad, realising something had gone drastically wrong, I pulled the plug..

I let the system cool, thru on a HeatPipe Cooler, and it kicked back into life ..., Phew... I thought.., The system seemed fine, but it was running way hotter than it normally did on air, so I puuled it down to check out if there was some issue with the Cooler installation.., weel imagine my shock and horror when I tried lifting the CPU out of the Socket and realised that it was a bit tougher than it should have been, Why, because the Bloody thing had litterly welded itself to the farking pins..., I kid you not.., of course onremoving the CPU, I destroyed it and the Motherboard... Argghhhhhh.

Now Intel have basically told me to go jump , ASUS are basically telling me to have relations with myself , and I am out of pocket arounf $1200.00 AU.

 

[beersnob]

as far as your set up what voltage are you at? you should be at ~1.35v hitting 66° at stock 2.66ghz is high.

Having tweaking nothing yet, I'm at 1.36v. I just turned the system on and loaded WinXp for the first time last night. When I ran PCMark05 I had turned on QFan in "Performance" mode. I had an idle temp that in the low 50's. I noticed the CPU fan jumping all over the RPM scale and the CPU temp differing by as much as 5-7 degrees C ever five seconds.

I've now turned off QFan and turned all my fans on full throttle, and have an idle temp of 45. I'll run PCMark again to see where I end up.

By the way, what CPU burn-in programs do you all use? I want freeware, of course.

 

[beersnob]

The chip wont let you kill it because it will just throttle back. So over clock away. Also this info is pointed out in the article.

I'm not concerned about suddenly frying the chip. I'm concerned about prolonged exposure to undesirable heat causing a slow, premature death. Naturally some risk is involved when you o/c, but why not take a careful, measured approach? I'm just looking to draw from others' experience.

 

[SidVicious]

Prime95 is a popular tool for stress testing, just follow These instructions.

Once you are satisfied with your system's stability, feel free to join the THGC Folding@Home team.

 

[pengwin]

u should be fine i ran my old northwood at 75-81*C for like weeks then i got the zalman so its fine.

 

[beersnob]

Here's my update with stock settings:

Idle temp: 45 C
Max temp during PCMark05: 64 C briefly during multi-tasking tests, mid 50's otherwise.

Prime95 (Thanks, SidVicious) for 35 min. with all fans wide open: 63-65 C

Prime 95 for 35 min. with rear fan (SILVERSTONE FM121 120mm) reduced from ~2500 to ~2000 RPM: 64-65 C (this cut noise in half or better). Basically same temps, with a lot less noise!

My temps basically parallel THW, with Zalman 9500 (and the ambient temp in my office has been between 82 and 85 F during my testing).

It's Friday night, so off to have some Friday fun I go....

 

[SidVicious]

Was that running one or two instances of Prime95 ? Did you follow the instructions to run one instance of Prime95 on each core ?

BTW, a system can be considered Prime stable only after six to twelve hours worth of testing, still, some maniacs (like me) won't settle for anything short of 48 hours.

 

[beersnob]

Was that running one or two instances of Prime95 ? Did you follow the instructions to run one instance of Prime95 on each core ?

BTW, a system can be considered Prime stable only after six to twelve hours worth of testing, still, some maniacs (like me) won't settle for anything short of 48 hours.

It was two instances per the link you provided, and I realize 35 minutes isn't thorough enough. I just wanted to play around a little tonight before I had to leave. I will be more thorough this weekend.

I'm happy to announce I'm not a maniac. Not yet anyway.