Are P4s more stable at high temperatures?

Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
edited February 2005 in Hardware
People have stated that P4s are "hyper-stable" at temperatures that would cause an Athlon to crash. I've always known that this isn't true, and a lot of you also know that this isn't the case.

But, I set out to prove it nonetheless. So, I grabbed my 1700+ Palomino, and dropped it in the KX7 system that was sitting on my desk. I hooked up the digidoc5, with a thermistor touching the side of the CPU core so I could get the actual CPU temperature. The only cooling was a 2500+ HSF, with various fans. I put MBM5 and F@H on it, fired them up, unplugged the cpu fan, sat back, and watched the thing heat up.

Would you care to guess how hot the 1700+ got (it NEVER froze due to heat- it DID freeze with -forcesse, but it was doing that at 38*C, too. That's just an issue with some of the Palomino cores).

75*C?
No.

80*C?
Nope.

85*C?
*yawn*

87*C
*zzzzz*

Well, I'm actually not entirely sure what I hit- the DigiDoc 5 won't read past 90*C (trust me, I tried- with a Butane torch and a spare thermistor :D), but it was AT LEAST 90*C. And judging by the amount of time it took to come down to a temperature of 90*C or below so the DigiDoc5 could read it, I would guess that it actually hit something close to 100*C. While running F@H. Without crashing.

It spent a grand total of around 20 minutes at or above 70*C over the course of the test, without crashing. Ever. It then spent all night between 68*C and 53*C, with an average of 59*C (according to MBM5- remember that the actual temperature is 5-10*C higher than this).

So much for "the P4s are more stable at high temperatures", eh? ;D

Some pics/screenshots:

Comments

  • primesuspectprimesuspect Beepin n' Boopin Detroit, MI Icrontian
    edited May 2004
    geeky, you are certifiably nuts....


    But so are we :D

    Sweet!
  • mmonninmmonnin Centreville, VA
    edited May 2004
    I wonder what a duron or a Tbird could do as well. Those things are beasts. Many can run with chipped cored and all.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    Probably about the same. The Palominos are spec'd for a max die temp. of 90*C.
  • madmatmadmat
    edited May 2004
    Run both without a sink and see what happens...the P4 will shut down and the AMD will detonate.
    Sounds like there may be a grain of truth there to me.
  • ThraxThrax 🐌 Austin, TX Icrontian
    edited May 2004
    "Without a sink" is an invalid circumstance as all modern motherboards are preconfigured with overheat shutdown enabled.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    ^ what Thrax said. The CPU overheating protection was disabled on the KX7 for the test. It will shut down automatically if the cpu exceeds 60-75*C, depending on where you set it, when it's configured normally.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited May 2004
    My box, if left set to thermal protect, would shut down the P4s at 80-85 C Prescott never gets that hot at the OC I have in place. What I used to say what I said, was based on Barton that went unstable at under 70 C and progressively got worse at folding. Both the P4s and the Barton are (in the Barton's case WERE) running roughly 24\7 for months. And when I mean stable I mean stable over weeks and months at constant high temps. It is good that Palominos are more stable, though.

    Normal for the Prescott, though, is 66 C +\- one degree C at the OC I have it set at. I've had it running for weeks now, 24\7, at that rate.

    One thing about benching for video, Geeky, you might next try this:

    Load a video benchmarker, one instance. Get into the Task Manager, in XP Pro, and set process affinity to CPU0, Priority to High, leave it running but not actively benching. Benmchmark video with it set that way. NOW, open the Task Manager, reset the process to normal, check boht pipes, and see if benches improve. They should not change much, most video benchmarkers are 32 bit code, XP will let them use one pipe of CPU-- not both. IF you have a benchmarker that can be run in a window, you might be able to run two, get close to same rates from BOTH at once, versus running just one. Guess, the combined benchmarks (qty two, benchmarks from each added together as that is what the box is actually handling, TWO instances of bench at once), one per pipe, will be about 1.6-1.8X the single benchmark rate, because you are eating lots of RAM, you are using ONE video card, and the Prescott has a cache common to both pipes and does not have the anticipation strategies in the Itanium2 or even the Zeon hardcoded into it.
  • ThraxThrax 🐌 Austin, TX Icrontian
    edited May 2004
    Xeon. X-e-o-n.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    Uh? How did we arrive at video benchmarking?
  • madmatmadmat
    edited May 2004
    I'll put it this way, I've seen P4's survive things that would melt an Athlon...one of the guys I worked with ran one doing a virus scan in a notebook after he'd pulled the sink under the keyboard completely OUT as in gone, see ya and it ran without a hicup for at least 3 hours that I'm sure of and possibly longer until I walked by and noticed what was going on so I grabbed a small sink with no fan on it and set it on the CPU to at least help alieviate the problem and walked away, he came back and I pointed out what he had done and he said "no biggie, they thermal throttle so they won't get that hot" and grabbed the sink I had placed there, screamed and dropped the sink back into place.
    The sink was so hot he got a nice second degree burn yet the laptop was merrily churning along as if nothing out of the ordinary was wrong, and this was an older P4 laptop...not a P4M, or centrino just a 1.5 or 1.6 P4.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited May 2004
    We were talking back and forth about P4s versus AMD. What I wondered, was a couple things, and now that you mentioned it, I would be interested in temps on Prescott with two 3D benchmarkers running at once also.... In my case, not going to buy a hugely expensive video card to bench 2X simultaneously, so my benches would be low due to card capabilities. :D Figured might kill two birds with one stone, on a Prescott box since AFAIK you have one also.
  • Straight_ManStraight_Man Geeky, in my own way Naples, FL Icrontian
    edited May 2004
    madmat wrote:
    I'll put it this way, I've seen P4's survive things that would melt an Athlon...one of the guys I worked with ran one doing a virus scan in a notebook after he'd pulled the sink under the keyboard completely OUT as in gone, see ya and it ran without a hicup for at least 3 hours that I'm sure of and possibly longer until I walked by and noticed what was going on so I grabbed a small sink with no fan on it and set it on the CPU to at least help alieviate the problem and walked away, he came back and I pointed out what he had done and he said "no biggie, they thermal throttle so they won't get that hot" and grabbed the sink I had placed there, screamed and dropped the sink back into place.
    The sink was so hot he got a nice second degree burn yet the laptop was merrily churning along as if nothing out of the ordinary was wrong, and this was an older P4 laptop...not a P4M, or centrino just a 1.5 or 1.6 P4.


    Had that happen a few times, then I started using a thermal sensor first or letting box cool for half an hour first, or blowing chip\component cooler fluid on the HS before working on HS, and 5-8 min of that flow also worked. Using a Data-safe vacuum\blowers blower end also works.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    *snicker*

    But how fast and how hot was it running? There are no desktop P4s that can handle being run at 90*C at full speed. The clock throttling kicks in before then. It could very well have been maintaining a surface temperature of about 70*C (the max. rated die temp for the older P4s is 65-75*C), which is more than enough to give you a nasty burn. It also wasn't running at full speed.

    But, the fact that the P4 can run at 80*C or whatever is irrelevant in this case, because, as I've already proven, the athlon can, too, and it can do it at its rated speed. The P4 has to underclock itself to keep from cooking.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    Oh, and Ageek, spraying some of that chip cooler on a hot cpu heatsink is an extraordinarily stupid thing to do. Rapid changes in temperature like that are Not Good for silicon. You're just ASKING for a cracked CPU core, you realize that, right?

    //Edit
    Yeah, I have a prescott system. I may give that a try. After I find said prescott, which is MIA, and get a working board for it.
  • madmatmadmat
    edited May 2004
    I want to know what generated that graph...it's way off whatever it was as it never even breaks 70c...that's one temperature monitoring proggy I'd stay well away from.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    The graph is from mbm5, just dropped into excel. If you look at the note in the screenshot, or the post, you'll note that I mentioned several times that the KX7's readings are low (as measured at the cpu core using a digitaldoc 5 with a flat thermistor with as much of the protective plastic stripped away from the thermistor as possible) by as much as 10*C, more in some cases.

    The KX7 has an in-socket diode. It's reading the back of the CPU package, not the actual cpu temperature. That graph was for the overnight run, which averaged 59*C according to mbm5; the actual temperature was closer to 65*C. Look at the photograph of the digidoc 5, and the screenshot above it for the maximum temperature results.
  • madmatmadmat
    edited May 2004
    yeah, that's what i was going by, still mbm shows 82c in that top screenie yet the graph never broke 70c, that's why i was wondering what made the graph...
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited May 2004
    Oh, I see. I misunderstood. The reason for that is because the MBM screenshot is from the "maximum temperature" run; the graph is from the interval log that I had it run overnight. :)
  • madmatmadmat
    edited May 2004
    Ah, that makes sense. And for the record I've never said that either was more stable at high temps but I have said that O.C'ed Athlons are more temperature intolerant than an equally O.C.'ed P4 which has been my personal experience...your mileage may vary.
    I've seen customer rigs with stock clocked AMD's that were like ovens inside when they ran that never faltered but on the flip side I've seen O.C.'ed AMD's that were very touchy about temps, 2 or 3c the wrong way and they'd refuse to run stably.
  • Geeky1Geeky1
    kicks thread... someone should've stickied this a while back. The cooling form mod is falling down on the job again. :rolleyes: University of the Pacific (Stockton, CA, USA)
    edited February 2005
    kicks thread... someone should've stickied this a while back. The cooling form mod is falling down on the job again. :rolleyes:;D
  • MedlockMedlock Miramar, Florida Member
    edited February 2005
    Why was an Athlon ever tested? I'm assuming, based on the title of this thread, that you were testing the stability of a P4 at a higher temperature and comparing to the stability at a lower temp. :confused:
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited February 2005
    You assumed wrong. :p First line of the thread dude. :D
  • shwaipshwaip bluffin' with my muffin Icrontian
    edited February 2005
    TheGr81 brings up a valid point though. The title makes me expect some sort of comparison, or at least data on a P4. Perhaps they mean at 105* C...or at 110.
  • Geeky1Geeky1 University of the Pacific (Stockton, CA, USA)
    edited February 2005
    The P4 won't operate at that temp period. It will shut itself off before it gets that high... no mobo intervention required. :p
  • shwaipshwaip bluffin' with my muffin Icrontian
    edited February 2005
    Geeky1 wrote:
    The P4 won't operate at that temp period. It will shut itself off before it gets that high... no mobo intervention required. :p

    Regardless, "Athlon XPs are stable at high temperatures" would be a more fitting title :p
Sign In or Register to comment.