High Temps

  • Thread starter Thread starter Willi
  • Start date Start date
W

Willi

Playing around with my xp2400, 266 bus model, on a new motherboard, AN7
to see what it can do. I've got it running a 138x16 2.2 Ghz - xp3200
speed. I've run Prime 95 over night with no errors. However, it's gotten
to 70 degrees (which I know is too high).

A few questions:

However, if it's over heating shouldn't I have gotten some errors?

Are Abit EQ temps accurate?

I thought the multipliers were locked on these chips, why can I change
the multiplier in the bios?

It seems that with the cooling I have, I'm overdoing this chip, any
suggestions?

Willi
 
Willi said:
Playing around with my xp2400, 266 bus model, on a new motherboard,
AN7 to see what it can do. I've got it running a 138x16 2.2 Ghz -
xp3200 speed. I've run Prime 95 over night with no errors. However,
it's gotten to 70 degrees (which I know is too high).

I know what you mean by "xp3200 speed" but don't expect XP3200 performance.
The 3200 runs on a 200Mhz FSB and has a 512KB L2 cache over your CPU.
A few questions:

However, if it's over heating shouldn't I have gotten some errors?

70°C is within the design specs of the CPU so, no, it doesn't mean you'll
get errors. However I wouldn't be confortable running a CPU at above 55°C or
so (diode temp) constantly.
Are Abit EQ temps accurate?
Pass.

I thought the multipliers were locked on these chips, why can I change
the multiplier in the bios?

What chips? Is it a Tbred or a Thorton? The multi is locked on the later,
not on the former. Overwise they're essentially the same CPU.
It seems that with the cooling I have, I'm overdoing this chip, any
suggestions?

What vcore are you running to get this speed stable? It's the vcore that'll
give you the heat. Try backing off the vcore until Prime95 gives you errors,
then going back to the last stable setting.

What I'd do: Take the side off your case, mark where the centre of the CPU
fan lines up, take it to a panel shop and get an 80mm hole cut. Fit side to
case and measure throough the hole the distance from side to CPU fan. Fit an
80mm case fan to inside of case, you can use a wire finger guard and even a
pretty LED fan such as the three-colour one my finace has in the side of her
[email protected]. Subtract the thickness of the fan from the measurement you got
before, then subtract a further 4-5mm for clearance.. Go to the supermarket
and find a tapered plastic glass with a top diameter of approximately 80mm.
Cut it to the length you came up with (A hot knife is good, I used a
chisel-tip in my soldering iron as I don't have a hot knife) and hot-glue it
to the inside of the case fan. You should now have an excellent cooling
solution that sucks cool, outside-case air and ducts it directly to your
CPU. This should drop your CPU diode temp (IME) anything between 8 and 20°C
depending on how efficient your case ventilation was before.

If you in fact have a Tbred CPU (as I do) that is unlocked then get some
PC3200 RAM and run it at 200 x 11 or thereabouts (if your mobo supports
200Mhz FSB). This will greatly improve your system performance as opposed to
running a 133Mhz FSB. My XP1800 Tbred, using the cooling described above,
with the standard AMD HSF, runs at 2.1Ghz all day every day at 200 x 10.5,
benchmarks better than an XP3000+ and sits at around 52°C full-load (SETI)
with a vcore of 1.8v. It won't do 2.2Ghz on a 200 FSB but will do over
2.2Ghz on a 133Mhz FSB. This is because the CPU is actually doing a lot more
work on the faster FSB than the slower one. It's not about CPU speed alone.

BTW, I've crossposted this to alt.comp.hardware.overclocking where it really
should have been in the first place. You'll get more answers there.
 
Playing around with my xp2400, 266 bus model, on a new motherboard, AN7
to see what it can do. I've got it running a 138x16 2.2 Ghz - xp3200
speed. I've run Prime 95 over night with no errors. However, it's gotten
to 70 degrees (which I know is too high).

A few questions:

However, if it's over heating shouldn't I have gotten some errors?

You might get errors, crashes, etc, etc.

Was it at 70 while running Prime 95? That Torture Test is about
as severe a condition as the CPU is likely to face, providing
ambient room temp wasn't lower during that test.

Are Abit EQ temps accurate?

Don't know, but 70 does sound kind of high, did you raise CPU
vcore (how much if so?) and does case have suitable airflow and a
relatively good heatsink? In other words, do you have reason to
suspect that is accurate or inaccurate temp? A high-end heatsink
(mounted properly), good case airflow, etc, would make it very
hard to get chip past 60C unless the vcore was too high or room
temp was over 100F.

I thought the multipliers were locked on these chips, why can I change
the multiplier in the bios?

Because it ignores the bios setting (if the cpu is locked, that's
what it would do, else your cpu is earlier model that wasn't
locked)?

Run a windows(?) utility like WCPUID (Google will find it) to see
if the bios settings are actually changing multiplier or being
ignored.
It seems that with the cooling I have, I'm overdoing this chip, any
suggestions?

Only the obvious 3 things:

1) Run chip cooler with better heatsink, speed reduction, and/or
voltage reduction.

2) Try reseating heatsink and increasing chassis airflow.

3) If it ain't broke, don't fix it.
 
I posted the below some time ago:

Playing around with my xp2400, 266 bus model, on a new motherboard, AN7
to see what it can do. I've got it running a 138x16 2.2 Ghz - xp3200
speed. I've run Prime 95 over night with no errors. However, it's gotten
to 70 degrees (which I know is too high).


Based on advice here, I moved up the FSB as high as I could and adjusted
the multiplier. I'm now running at 197 X 11, which is much faster than
what I was running before. Also without changing the voltage or
cooling, it is running much cooler. It idles in the high forties/ low
fifties and only reaches the low 60's when running Prime 95.

Stable now for over a week running all kinds of stuff. Sweet!

Thanks Guys,

Willi
 
Back
Top