Duron's temperature

[Priit]

I just wanted to know does 46-50 degrees celsius is bearable for Duron (700Mhz) or is it too hot and shorten's processor life ? It doesn't have any fancy cooling system, but do I really need one ? My K6-2+ was just 20 C max. ...

 

[Compellor]

As long as it doesn't run over 50 C, you're ok. Lately, my Duron 600 o/c to 800 has been running at about 46 C at idle. It's been solid as a rock for 3 months now.

 

[Subversal]

46C idle? Ouch... My Duron 700@1000-1.85v is running about that after some Prime95 and seti@home. Anyway yeah OC'ed they will run high temps (like 40-46) but at default speeds mine ran under 40C after the formentioned programs.

 

[Priit]

Damn, my duron's cooler doesn't look so bad (well, it's big, not form iron and with big fan @ 5000+ rpm), but processor still goes up to 51 C after and hour or two with Quake3. Only better cooler that I can get is Titan's model which costs about half of the processor's price I can't install any extra fans near the CPU into my case, too. According to AMD, Duron's death temperature is around 90 C, so I thought 50-60 C should be acceptable...

 

[Mikewarrior2]

What kind of heatsink are you using? how's the case cooling?



Mike
P.S. keep in mind these measurements can be significantly incorrect.

 

[Mikewarrior2]

You "SHOULD" be okay, but your thermistor may be under-reading the temp. Is there any way to find out what heatsink you are using?


Mike

 

[Priit]

Well, CPU-cooler is form A-Corp (as good as no-name), there's nothing else (e.g. model or sth.) written on the cooler/it's box. Case cooling pretty much sucks, too (no ventilation holes/place for extra fan near CPU). The mobo is ASUS A7V, I somewhy trust it's thermal detectors.

 

[Mikewarrior2]

Well, don't trust it. . See my sig for more info on why.

now that we know you're using an A7V, what bios are you using? For the most part, revision 1.02 boards(one with the socket-thermistor) tend to "approximate" temperature a bit warmer than actual core temp. This does not mean it makes for good comparisons, though, only for an individual to get an approximate temp.


Mike

 

[RoboTECH]

Mikewarrior, that's some very interesting information there. I just got a Duron + K7T Pro2a combo.

800 Duron + FOP32 + 1.85v "special BIOS" = 1-Gig for ~$135 (heh...)

Anyway, the bizsnatch runs (according to MSI's temperature readings in windows) @ 42 Celsius @ idle. BLAH! that seems high to me. Any idea about this? Is this a concern for me here?

I'm a spoiled little bitch with my cB0 P3. I couldn't hit 42 Celsius in the middle of the summer after running Prime95 + Q3 demoloop + Q2 demoloop + 3dMark2000 demoloop on the 700 when I had it @ 1012 @ 1.9v

 

[Priit]

Mikewarrior, I have board revision 1.02 with BIOS 1004D. Are newer BIOS'es more "accurate" when reading temperature ? Oh, I still remember the good-old days, when processor's temp was O.K. until you could touch the heatsink without burning fingers....

 

[Compellor]

<< 46C idle? Ouch... My Duron 700@1000-1.85v is running about that after some Prime95 and seti@home. Anyway yeah OC'ed they will run high temps (like 40-46) but at default speeds mine ran under 40C after the formentioned programs. >>



My CPU temp varies on how warm my apartment is. It's cold right now where I live, so, turning down the heat will take my Duron down to the high 30's/low 40's at idle. Ambient temp ranges from 22 C - 31 C. If I run Prime95 or Quake 3, it only goes up about 2 - 3 degrees more. I'm using a cheap Vantec Socket A cooler. A better HSF would cool it down some more, but my machine runs good enough for me not to worry about it.

 

[Mikewarrior2]

Robotech,

How the heck does the k7t pro2a read temp? I heard it does not have a socket thermistor...

Also, a duron at 800@1000/1.85V pushes roughly 54W of heat. A p3 @ 1050/1.9V pushes out about 36W. So its roughly 50% more heat for the duron versus the p3. Your temps are fine, but what are your full load temps?


Priit,

The good old days when chips didn't melt if the heatsink wasn't mounted properly... . Its kinda scary seeing times when socket a boards misread temps by 10 or more C too low(i've seen 28C too low once).


Mike

 

[Thor2]

I have a Duron 600 o/c to 1050 running at 1.8v. Temp at idle is 37 and under full load is 42. This is in Florida where room temp gets up to 28c.

 

[RoboTECH]

Mikewarrior, I honestly have no clue how it measures

I'm down to 1.750v tho, holding a steady 1-Gig

WH00000T!!!

 

[Mikewarrior2]

Thor2,

your mb is under reading your core temp by a lot. probably something like 10C too low. Not anything to worry about now, but something to maybe consider if you decide to push your chip farther.

RoboTECH,

Congrats!


Mike

 

[Balael]

Heya, just one quick question here for anybody. A duron 600 at 600 with the default core voltage of 1.5v should run about 8c higher then room temp right? So then, i have my D600 @ 963 with 107 x 9 @ 1.625v, and it runs at 34C, which is about 12-14c above case temp. This is accurate as far as i know, i'm running a tk7raid with fop38 in a fongkai 320 with 120mm front, 90mm back case cooling. This all seems about right, right?

balael

 

[Balael]

sorry, double post.

 

[Mikewarrior2]

Hmm.. to run only 12-14C over ambient case temp would require a heatsink rating of .28C/W(which doesn't exist in a socket a format).

As typical of kt7's, your socket-thermistor is &quot;under-approximating' cpu core temp. See my sig for more info.

What bios revision are you using? If it is UL, what heatsink are you using? Chances are fairly high your cpu core temp is in the 40-45C range.

And as a note, no socket A mb is accurate. Some &quot;Can&quot; be good at approximating core temp, but accuracy is an entirely different issue.

 

[Balael]

I'm using the newest version of the kt7raid bios with a fop38, not sure what it is right now, wt i think? and if my reading was correct for the duron 600 @ 600 with 1.5v, why would it not be correct at duron 600 @ 963w1.625v? especially pointing to the fact that the temperature readings are correct when the duron 600 is running default voltage and speed, but when i increase the speed, and only increase .125v that it would be reading 34c idle, which would be according to your findings off by 10c then? Sorry, but that doesn't make sense at all, even if the readings were off after the increase in voltage and speed, i don't think they would be THAT off. oh well, just my two cents.

 

[Mikewarrior2]

Are you comparing full-load temps(the only temps that matter) or idle temps? If those are your idle temps, try getting some full load temps and reposting those.

Either way, the readings arent' correct at default clock either. Truth be told, they are never accurate on a socket-thermistor setup. I've seen UL Bios Kt7's under-read cpu core temp by 20C.


Mike

P.S. If you're idling at 34C, what is your load temp at 963?

 

[Balael]

load temp, playing sacrifice for an hour or two or so, can get up to 37 or 38, but just depends on how cold the room is. Also, if the chip was running in the mid 40's range, i'd at least be able to tell a little bit by how warm or cold the air coming out the back is, but the air coming out the back of my case is always cold. Now if i run the chip at 1000 with 10 x 100 at 1.75v i think it is, the idle temp goes up, but i'm not sure what it goes up to, and then i can tell actually tell that the air blowing out of my case is warmer. I don't know, i'm just confused by the fact that my chip is supposed to run 8-9c above room temp when it's default speed and voltage, and that's what my thermistor reads, i just don't see how it goes from being accurate when it's reading my 600 readings, to VERY inaccurate when it's reading oc'ed readings as some people claim, I would totally understand if it was reading the wrong temp at default, but it's not. Well that's my four cents now.

balael

 

[Mikewarrior2]

How do you know it is accurate at default clock? How do you know it is &quot;accurate&quot;. Does your thermistor magically show full cpu core temp changes? It doesn't, therefore is not accurate. It &quot;Approximates&quot; temp well at default, but like all socket A platforms, temp changes are &quot;compressed&quot;. Look at my sig. The 12C core temp difference between the PAL6035 and the C-ORb in anandtech's review is compressed into a 2C reading. Same with your case, the temp change from idle to full load is not being show fully, due to thermistor-reading compressions that are very common to socket a setups.


Sacrifice stresses subsytems + cpu. Try a cpu-only stresser, like Prime95.



Mike

 

[BW]

Two days ago i upgraded from my sig below to a Abit kt7raid and have already done the voltage mod. Ive got a duron 700 at 1gig now.seems as fast as my t-bird on my old Asus k7m. It runs now at 2.05 volts and at idle the temp is 45c and after i played rune for 1 1\2 hrs it went to 56c.Never a hiccup thuogh.Im using a chrom orb on it.

 

[Balael]

::laughs:: So it magically compresses it 5 fold when the temp increases about 6c? So you mean to tell me that EVERY single motherboard, and every single chip is the same, and responds &quot;magically&quot; the same way to overclocking measures? I don't think so. Oh sure i see the reasoning in your results, and all the facts about mobo's temp readings being off, but every single setup is different. Btw, full load is 38 or 39, on top of that this little crappy duron 600 i have can oc to 1100 with max volts, so maybe, once again that beautiful word, just maybe it &quot;magically&quot; runs a bit cold when it's only running at 963. Oh well, every setup's different, and when i go out and buy an external source of reading my temps, i'll be sure and post the results, but hey maybe i'm wrong, but then again, my motherboard might just be right.

balael

 

[Mikewarrior2]

a full load at 38-39 is close enough to 42 or whatever C that it is a close enough approximation.

NO, it isn't a 6X Compression all the time. But the nature of measuring from a secondary heatsource ensures a compression of results, period.

External readings likewise are LOWER than core temp.


Mike