Need some help w/ a 3600 OC

Rocksaurus

Member
Aug 30, 2004
48
0
0
Hey Guys,
I'm normally a "Hot Deals" only poster, but I recently built a system for my friend around a Brisbane 3600 (a very hot deal!) with the intention of OCing it. Well, here I am... I've read lots of threads about it, and have been toying with voltages, RAM speeds, HT multipliers, etc...

I read in a post that these new Brisbanes don't correctly report their temperatures... Is this true? When running Prime95 and SuperPI to stress test the computer, it will shut down after around and hour or less of full load. Here's what it's currently at (I've had it higher but have been backing it down due to the shut downs):

2.375 GHz (250 Bus)
1.34v
RAM is DDR2 800 currently running slightly under its spec (396)
HT is at 4x so it's running at full speed (1000).
Using the new Biostar nForce 7050 board

Temperatures at idle in BIOS however are reported as 67C... But the CPU isn't reporting correctly, right? I'm sure the "random" shut downs during stress testing are due to the mobo "detecting" a temperature at its shutdown levels...

So, yikes, sorry for the long post, my question is- how do you guys who have successfully OC'd a 3600 do it if it doesn't report correct temps? Should I just adjust the "max shutdown temps" in my BIOS? Or is this stuff I'm hearing about Brisbanes not reporting correct temps not true? I'm kind of confused and out of ideas. Thanks!
 

ItsAlive

Golden Member
Oct 7, 2005
1,147
9
81
I had this same problem when overclocking, lucky I had a DFI board which read correct temps on 2 different chips (1 brisbane, and 1 windsor core). I have the Biostar 7025 board, and it would report temps like 50C idle and 75C load.....which I knew to be incorrect. So while testing on the DFI, I just compared temps using core temp. It seems like core temp reports about 10-15C lower temps tho. So you will want to add 10-15C to the temp that core temp shows, and you should be ok.

You can download Core Temp Here

Also I had to increase the shutdown temp in bios, otherwise the board will shutdown because of the misread temps.

Hopefully, a bios update will be released soon that fixes this issue.
 

Rocksaurus

Member
Aug 30, 2004
48
0
0
Thanks for the tips... I went ahead and turned *off* the shut down temp as it's max was only 10C higher than what it was set at default... And since this thing reports to my BIOS about 15C higher than it is, that's not really all that helpful...

The alarming thing now, though, is that I've got it running at a calm 2.49 GHz and doing two instances of the Prime95 stress test, and Core Temp, after adding 15C is giving me temps ~67C... That's pretty hot, no? What can I do to get this temp down? I'm only running it at 1.33v, and I'm not pushing it nearly has hard as most people on this forum push their's on stock cooling... Did I just get a less than optimal piece of silicon?

Edit: to clarify, I've already reseated the CPU, and this thing doesn't have a case fan, just a CPU fan...

Double Edit: Now I'm running it back at stock speeds to see the "temperature" core temp reports... Adding the 15C it's now, at stock speeds, it's ~64C... Not much of a reduction. To be honest I feel like I can't take any of the temperatures I read seriously...
 

ItsAlive

Golden Member
Oct 7, 2005
1,147
9
81
I hit around 45C at 2.8ghz using an Arctic Freezer 64, and I have pretty good airflow in my case. What temps are you getting in core temp at stock speeds?
 

ItsAlive

Golden Member
Oct 7, 2005
1,147
9
81
Wow something doesnt seem right....what kind of heat paste are you using?

EDIT: Since we are running different boards, core temp might be more accurate on your board....

There are 2 bios updates for your board, one says it adds support for rev. G CPUs, which is the brisbane.....you might want to try it. T7050 bios
 

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
I've got an MSI K9N Platinum with the 1.5 BIOS (v1.6 seems to be worse for OCing) and the temps are correct as far as I can tell. At 2.6 GHz with 1.424v being fed to the core, the chip idles at around 36C and loads at around 53C on a lapped Freezer 64 with MX-1. That should give you an idea of what temps to look for.
 

Rocksaurus

Member
Aug 30, 2004
48
0
0
Wow how did I not see those updates? I looked earlier. Thanks, I'll try that and report back.

On the topic of temps -

IN my BIOS, the temps are insanely high, ~60 idle, and coretemp is around 30 idle and lower. So I just don't know what to think!

OKAY: Edit time.

I updated the BIOS, no change in BIOS, it's actually reading 70C even idle in BIOS, not 60ish. I guess I'll just go with Core Temp's read outs... It's currently idling at 34C according to Core Temp, which is likely correct. Guess I'll just go with the Core Temp and the no shut down temps. Thanks guys!

(any additional advice is still welcome, though.)
 

Rocksaurus

Member
Aug 30, 2004
48
0
0
Ooookay so after awhile of running Prime95 on both processors my friend started using the machine casually while running Prime95 still and the screen goes wacko... Is this a common thing...?
 

rmed64

Senior member
Feb 4, 2005
237
0
0
I would use orthos, its prime95 but multi threaded. And I dont think your screen should go wacko.

My stock cooled 2.8GHz 3600 idles around 37c, load 65-70c with orthos depending on cool or hot temperature in room. Ran orthos for over 12 hours with no errors or shutdowns.
 

Rocksaurus

Member
Aug 30, 2004
48
0
0
Well I do have two instances of Prime95 running, one on each core... Could the screen issue be Prime95? That'd be odd...


Also, if you don't mind telling me, what are all your settings on your 3600? ie HTT multiplier, RAM speed, vcore, etc... Since the screen situation I've knocked the proc down to 2.3, but given all I read online that seems pretty weak...
 

lyssword

Diamond Member
Dec 15, 2005
5,630
25
91
umm, did you put artic silver on heatsink? or thermal compound? remember it has to be really thin layer too.
 

Rocksaurus

Member
Aug 30, 2004
48
0
0
Originally posted by: lyssword
umm, did you put artic silver on heatsink? or thermal compound? remember it has to be really thin layer too.

Just using stock everything. Heatsink comes with thinly applied thermal paste on the bottom... Is everyone who's running on "stock" HSF using different thermal compound and I'm just silly?
 

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
Well most people who are overclocking ditch the stock heatsink because it's pretty inadequate once the voltages get increased to beyond stock...and with the X2 3600+ being a 65nm chip, you really shouldn't push the CPU temperatures into the high 60C or 70C range or the vCore beyond 1.5V. That's why most people who overclock elect to use an aftermarket heatsink like the Freezer 64 or Tuniq Tower. If you don't want to shell out the extra cash for one of these heatsinks, at least remove your stock heatsink and apply Arctic Cooling MX-1 Thermal Compound to the base...it works better than Arctic Silver 5, but make sure you apply only a very thin layer (rice-grain size to center of CPU, then spread) or else the thermal conductivity will be negatively affected.

Link to AC MX-1 ($6.99 + ~$3 shipping if using priority mail or first class mail)
http://www.svc.com/acmx1.html