Idle temps: P3-S 1.26Ghz Tualatin

What's a good range of idle temps for this chip? I've got a
Thermalright SK-6 w/ 5800rpm 60x60x15mm Delta fan,
running a P3-S 1.26 @ 1.33 (140FSB). Idle temps are
case + 16C (e.g. right now 27C case 43C cpu), and I'm
wondering if more airflow will lower this number substantially
(system is in a living room so noise _is_ a consideration), or
is this about as good as it gets with air cooled Tualatins?

The other curiosity is load temps -- this chip gains only 4-5C
after running CPUBurn for 30 minutes. Is this normal?

Thanks for any info.

 

That sounds about right, my P3-S 1.4s idle at case +10C - but I'm
measuring CPU temps with a thermistor jammed up against the heat
spreader (old board, doesn't know how to read on-die temps). I would
expect to read at least 5C higher on-die. I'm running 1U solid copper
coolers with 60x60x10mm fans @ 5000rpm.

I doubt increased airflow will make much difference - I had a big case
fan blowing directly on my Coppermines (dropped their idle temps by 10C)
and initially left it in place after upgrading, but it made almost no
difference to the Tualatins so I disconnected it. You might gain a small
improvement by lowering Vcore - mine run on 1.35V.

Yup, mine only gain 2C under load, again measured externally.

P2B

 

temps are normal /IMHO HLT softCooling doesn´t work properly with
Tuallies on most boards when idle-ing/; you have a hell of a cooler,
try to run it @ 5V !
see my site under comp/benchmarks about temperatures ..

 

Thanks all for the info.

Spajky, instead of running the fan at 5V or 7V, I'd like to put
an adapter on the SK-6 and use a 70 or 80mm fan at a lower
speed. Do you know if anyone makes such a thing for the
SK-6? Or how would I go about making one of my own?

Also, I'd like to go higher than 1.33GHz but this silly Asus
motherboard (TUSL2) automatically throttles back memory
timings and drops AGP to 2X when FSB is set to anything
over 140MHz. No way around it other than to use a custom
bios, which from what I've read lots of people have had
problems with.
http://www.x86-secret.com/articles/tweak/i815twken.htm

 

Fan adapters, in general, don't work, or at least not as one expects,
because the narrowing of the hole diameter means the airflow has to
dramatically change direction (axial fans blow the air radially outward,
not 'downward') and that increases back pressure which counters the
supposed benefit of the larger fan.

The SK6 sping clip mount makes it particularly difficult to mount anything
other than the supported fan sizes.

 

Interesting. Well I tried the existing fan both ways, and there
was virtually no difference in idle or load temps. With the fan
blowing outward, temps were 1C higher than with it blowing
down into the heatsink.

Yeah, those wires are a royal PITA. I had to modify one of
the two included sets to accomodate a 15mm fan. Sounds
easy enough, but it took forever -- _real_ easy to bend them
into an unusable mess.

Thanks for the response.

 

...for making it your own, you would not need some strong
material, IMHO 1mm hard cardboard paper could support bigger fan w/o
problem /I´ve seen such solutions/ ...
...but even if you use your existent fan should not be problem
running it (if starts w/o problem) @ 5V (still should run around
3000rpm!); my fan is smaller than yours & at 2400rpm & my Tuallie runs
no problem really full load @ even more than 60°C from die measured.
Surely does not need more than 7V even if machine is settled in very
warm inviroment if airflow thru case is normal ... If does not start @
that voltage (deltas can be problematic sometimes), there is a simple
circuit (kick-on) on electronics section on my site too ...
Heat is really NOT an issue with Tuallies if are really not
pushed up too much when OC-ing....

The MoBO is excellent (the best for OC stability!)

...may be that first your Ram can not handle higher
frequencies (if having it more than 2 sticks or more than 256Mb if are
generic PC133 ones. (my actual 2x128 sticks can not go higher than
142MHz, but goes at 140 cas222!) . Cpu itself IMHO could go higher if
trying to lower Fsb/ram ratio in bios, but that I do not recommend you
(machine would be slower!) The problem to OC further your CPU is that
you @ default have already 133fsb & for better OC you need a heel of
Ram (Mushin PC150 or something - too expensive & hard to find these
days ...) Maybe would be better than trying to run higher than 140MHz
to "squeeze" Ram latencies instead it it can handle it!

Well about that AGP2x problem it is not a real problem, since
in practical life having 4x you would not gain more than 5% !!! If you
woud like to reflash bios, surely DO NOT do it from Windoze !!! Use
instead this:
http://users.volja.net/jerman55/FlashBiosFD.zip

But if you have an nVidia video card, try a reg. hack:
/you can copy/paste onto NotePad & save with *.reg
extension ------------

REGEDIT4

[HKEY_LOCAL_MACHINE\Software\NVIDIA Corporation]

[HKEY_LOCAL_MACHINE\Software\NVIDIA Corporation\System]
"EnableIronGate4X"=dword:00000001

 

I'm trying to keep temps under 50C at full load, since the
place we're living in now is not air-conditioned and gets
brutally hot during summer/early fall months. That's why
I opted for an older P3 instead of a P4 or Athlon system.
I don't know what the upper operating range will be until
July or August.

Also, this system is in a living room so I'm trying to keep
noise to a minimum as well.

Last year we were dealing with ambient temps over 90F
inside this apartment, during afternoon/evening hours.

It's good quality Infineon PC133 CL2, which I've run as
high as 150fsb CL2 on another system. Also, from what
I've read, dimm slots 2&3 on the TUSL2 share power,
and having all 3 slots populated can cause instability. So
I opted for 2 x 256MB (512MB dimms are supported
on this board, but they're too expensive).

That might be true on some boards. However on the TUSL2
Asus claims AGP4X must be initialized in the system bios
before Windows' video driver takes over, or else 4X is not
enabled under any circumstance. Not that 4X is any great
leap of performance -- I know it's only ~5% but I do a fair
amount of gaming, and with the older CPU I need all the
help I can get

Anyway thanks a lot for the info!

 

Rick wrote:

[snip]

You opted for the right CPU - I wouldn't be worried ;-)

Our cottage gets brutally hot too. The PC there has a P3-S [email protected]75Mhz
on a P2B-S (150Mhz FSB), 768MB of [email protected], 64MB GF4 video, and 10K7
SCSI drives. I've seen heat-spreader temperatures up to 65C on hot days
(which equates to at least 70C on-die), but the system shows no signs of
instability.

Mine lives in a mid tower case I scrounged at work, and has only CPU,
video card, and PSU exhaust fans. No case fans. Sure it gets hot in
summer, but it doesn't crash and it's quiet

P2B

http://tipperlinne.com/p2bmod

 

Interesting that it did that well exhausting upwards because there's no
shroud. Maybe because it's all copper.

Was that under load?

Btw, exhausting upwards give an 'adapter', if made right, the best chance
of working... essentially as a plenum. The problem is most of them aren't
made that way and it would need to be taller than is normally convenient.

The countervailing problem in that case is that axial fans don't 'suck'
well so you're back to six of one or half dozen of the other.

Yeah, I've got one and had to do some interesting 'bend and shape' mods
when I broke one of the dern things.

 

than put the fan on 7V (schematics are there on my site even if you
still wanna have RPM readings w/o problem); you will loose not much in
cooling, but much on noise!

What about some circuitry to have fans @5V during all the year & only
when amb.temps go over 25°C you svitch them to 7V
(this kind of "hot summer" switch i have in my sistem) ...

mounting an additional front case intake fan would benefit the system;
especially caps & other heating parts on MoBo & HD too!

try it, maybe will work! ...

 

yeah, thats true .. )

if thats correctly measured (calibrated sensor!) thats IMHO too much
/can not touch the HS for more than a second/ (on die there should be
with your frequency around 75°C onDie); IMHO the case ventilation is
not good enough or the paste has vanished ...
When I heated my machine simulating hot summer with a hair
dryer to 35°C I did not encounter HS base temps ˛ full load more than
55°C even with fan @ 5V, so it would be better not to go higher of IHS
temp than 60°C IMHO long term ... inserting @ 7V front case fan would
help & if mounted on cutted hole w/o screws (see my way under mods)
won´t contribute noticeable on cumulative noise figure ...

 

It gets that hot because the case ventilation is practically
non-existent (PSU exhaust only, as I said). The case to CPU temperature
differential remains close to constant regardless of ambient temperature
- and has not changed since I built it, so the paste should be OK.

Obviously it would run a lot cooler in summer if I added a case fan or
two, but that means more noise in the living room.

My point is the Tualatin remains stable at those elevated temperatures -
and I worry a lot more about premature failure of the 10K SCSI drives
than the CPU.

Actually, I figure that system is more likely to fail in winter since
there's no heat while we're away. The place is usually around -20C when
we arrive on Friday nights, and we proceed to light the woodstove and
warm it up to +25C as quickly as possible. This regular thermal cycling
cannot be good for the PC

P2B

 

I don't think cycling between -20 C and 25 C when not powered up will have
an effect on the CPU or any of the electronic components. Just think of
all the exotic cooling reported in this newsgroup that cause much larger
temperature swings. Many posters here would LOVE to have the problem of
what to do with cooling air at -20 C.

Phil Weldon

 

In the summer, try a box fan, about 20 to 24 inches rather than a side
panel for your case. Run it on the lowest speed. Or you could try a paddle
fan, a 52" Hunter original moves thousands of cu. ft. per minute, even at
the lowest speed (when it is practically silent); this fan will also keep
YOU a lot cooler. A Hunter original is expensive, but will last 75 years or
more. If you get a cheaper one, make sure the blades are not metal.

Phil Weldon

 

yes especially if are newer biggest ones (expensive) ...

the problem could be only if the electricity goes away (some snow
storm, etc.) for more than few hours; fans & especially HDs don´t like
temps under zero when started cold ... if running continuesly, no
problem ... ...

 

They are only 18GB, but 6th generation - fast and quiet. Very expensive
when new, still not all that cheap used.

Power is off at the panel when we are away - I try to stop the kids
powering on the PC until the place warms up....

I have plenty of spare parts for my P2B/Tualatin systems, so no big deal
if something fails - it's kind of an experiment in thermal abuse, I
guess

P2B

 

LOL, thermal abuse; thats what I did when I covered my machine with a
card box together with runnin hair dryer running 4 two hours
(simulating hottest summer) full load to test stability ... )))
... I gotta use this your term on my site´s article for torture of
PC parts ... ) ...
[... after a mont later all not perfect caps on MoBo revealed
bad ... & replaced them !] .. fast test of quality ... )

 

do you have them on Raid0 PCI hardware controller?

can you benchmark them (you can mail me too) with this:
http://users.volja.net/jerman55/SiSoftSandraMin.zip
/see inside Read_me file; 460kB stripped, no install/

I ´m really curious if are faster than WD Raptor76´s in Raid0 ...

 

I hereby place the term 'thermal abuse' in the public domain. You are
therefore free to use or abuse it as you see fit

P2B