X2 3800 Temps

[chaudx]

So, I overclocked my x2 939 a little 100mhz, and was looking to check the temps. What is thermal offset, and which of all those might be accurate. Ambient is ~27C, XP-90 cooling it, AS5, xp-90 came with almost a mirror finish, I touched it up. Had a 3000+ which ran ad ~34C under load, this x2 3800 gets to upper 40s.

Reseated it 3 times now to try and bring the temps down some....

EDIT: forgot picture..
http://img100.imageshack.us/img100/7026/whichul4.jpg

 

[MarcVenice]

trust speedfan/cpu-z for your vcore, ntune sais mine is running at 1.55v lol. I think it's the max of something, although I've also got a x2 3800+. As for temps, erm, the 39's I'm seeing are probably the right numbers, mine idles at 37-41 or so, depending on ambient temperatures. Under load, it would do high 40's or just touch 50. With higher ambient temps it will get to 55c under load.

Using arctic cooler 64 pro btw, with the pre applied grease, mx1 I believe it is.

 

[chaudx]

So what does that thermal offset mean?

I just used Clockgen to set everything back down to stock, and about 5s later the computer just turned off instantly. I've never had that happen before =\

 

[CTho9305]

So what does that thermal offset mean?

When software reads the temperature, it's supposed to add the offset to get the correct value (because there's chip-to-chip variation in the circuits used to measure temperature, so one might be +5 degrees off, and another might be -2 degrees off). www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/31410.pdf and www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/31411.pdf and www.amd.com/us-en/assets/content_type/white_papers_and_tech_docs/31412.pdf have some thermal info. Note that they say the diode itself, even after using the offset, is only accurate to +/-7C for some parts and +/-10C for others.

http://www.amd.com/us-en/asset...nd_tech_docs/26094.PDF (page 163) explains how to read the temperature for DDR1 CPUs, and http://www.amd.com/us-en/asset...nd_tech_docs/32559.pdf (page 177) explains it for DDR2 CPus.

 

[bryanW1995]

CTho9305, I know that you are not speaking for any company. However, if you WERE speaking for a company, do you think that we could get some inside penryn info ?

OP, my cpu idles at 37c on my zalman 9700 and, like marc's, creeps up towards 50 when under 100% load. Dual cores generate a lot more heat. I went from an fx55 that ran quite a bit cooler both idle and under load.

 

[chaudx]

Thanks CTho9305. Some good reading there.

bryanW1995 is yours overclocked any, and if so how much?

 

[bryanW1995]

not much, it's at 2.6 vs 2.4 stock. I got an asus a8v-xe mobo (read: cheap) b/c my old 939 mobo died when I upgraded from the fx55, and I didn't want to spend a lot of $$ on a place-filler mobo.
edit: oh yeah, my ram sucks, too. It was a very cheap system that I just kinda threw together over the years. I can't wait to relegate this thing to the file server room when my new cpu gets here (already have ip35 pro/crucial ballistix pc2 8500/ antec 900 for new system)!

 

[chaudx]

So, problems are still abound.
Shutdown temp is BIOS is 65C, and I think that is what is causing my computer to just turn off. I had it OCed 200 mhz and about 10 min into a game it just turned off instantly. Set everything back down to stock now, guess I cannot overclock for temp reasons?

EDIT: WOW. This is just getting worse. After ~15 min EVERYTHING in my computer at stock, everything updated, all newest drivers, newest bios, dual core patches installed, shutdown temp at 70c it just turned right off again. My firefox profile is now messed up, and is killing me in WoW.
I just received the processor Friday, so still can get a replacement from Newegg.
I failed to mention that this morning the computer would not POST, and only through a BIOS reset would it function properly.

What should I do...

 

[chaudx]

Bump =(

 

[MarcVenice]

Erm, you reseated it 3 times, I'm guessing you didn't reseat it correctly the last time. It's simply heating up to much.

If it still doesn't post right now though, then it's no longer heat, but something else. It could be that it doesn't post because perhaps it is NOT at stock settings, first things first, reset the cmos, I'm guessing you know how, right? Flip the battery out, switch the jumper, leave it like that for a minute, switch the jumper back, and flip the battery back in.

If that doesn't work, you might have a bad chip, mobo, or psu. This is a whole different ballgame, you will have to swap parts out for spare ones to see if one of them makes the difference and will post.

 

[chaudx]

Well, it idles at about the same temperature each time, so reseating did not make any real improvement. Cleaned heatsink and IHS on the CPU each time with 91% alcohol (Best I had on hand at the time) and tried applying the thremal paste 2 ways. The first time I just put a rice grain or so size drop in the middle of the IHS, set the heatsink on and gave it a slight twist while pressing down as I put it on. That got ~39-40C idle temp in bios, ambient at that time was 22C. Heatsink was warm to the touch, and I can hook up my thermometer to get a temp of it this afternoon after work.

I watched the temps last night, at all stock settings while playing wow, coretemp/speedfan hovered around 50-54C. That seems to be rather warm, and I don't know what to do to get them down. Every time after opening a game around ~10-15 min later the computer will just turn off instantly. This even occured with shutdown temp at 70C in the BIOS.

I wasn't very clear last night, it was that day's morning that I couldnt get a POST, a long cmos reset and just checking that everything was securely connected finally did it.

What can I do to get the temps down enough for stability at stock, and then room for a 200mhz OC....The xp-90 is mirror finish, using brand new AS5, I don't see what I am doing wrong.

 

[MarcVenice]

You need to apply more grease I'm guessing, not a rice grain size drop, but more. My arctic 64 pro came covered completely in grease at 1-2mm I'm guessing. Btw could it possibly be that your fan is blowing at very low rpm's, or perhaps blowing the wrong way, or maybe hot air stacks up in your case and can't leave ?

 

[chaudx]

Allrighty, Ill reseat it this evening with more.
Fan is a Panflo "medium" speed, don't remember the exact CFM, but it is plugged in to a 4 pin power connector, so running at one speed. No difference with the side off of the case, it is a big case with 2 intake and 2 exhaust fans, not much dust due to both of the intake fans having a filter, some cables messing with the airflow, but not enough to cause an issue.

How much should I squeeze onto the IHS.

Thanks for your help =)

 

[chaudx]

I am pretty much fed up. Reseated and got a whole 1C at idle. Did it again and it was still about the same. That time it took a cmos clear to post. Now the power just turns on and then off, staying on for half a second. My 3000+ works fine, and the x2 3800 will not even let the mobo turn on. I checked all the pins, ect and everything appears to be fine...
RMAed to Newegg, shipping it out this afternoon. Hopefully this time I get one that functions. That and having a reasonable difference in temperature between cores, no 11-15C difference.

 

[chaudx]

And....now it won't boot with either CPU.
x2 3800 power will just flicker on and off.
3000+ will give me 3 red LEDs, ram error.
Cleared cmos for an hour, tried other method of reset, moved ram around, tried 4 different sticks...


Help =(

 

[MarcVenice]

How old is the mobo ? Can you swap out for another ? What about the PSU ? My money is on the mobo though. I guess it couldn't get stable voltages to your x2 3800+ and no longer can get stable voltages to your 3000+ either. Hold up rma-ing the x2 3800+, because it might not be the chip after all. If your mobo died though, you're better of rma-ing the 3800+ and sell your ram and go am2/c2d.

 

[GeezerMan]

I use just about 3/4 of a BB of Arctic Silver 5 when I seat a HSF. Their web page says the same thing IIRC. That amount works fine. I have my personal 939 board in the for sale section right now, should yours be bad.

 

[chaudx]

Order - 7/24/2005 7:46:38 PM
Mobo:
DFI LANPARTY UT nF4 Ultra-D Socket 939 NVIDIA nForce4 Ultra ATX AMD Motherboard - Retail

PSU:
OCZ ModStream OCZ45012U ATX/BTX/PCI Express/SATA 450W Power Supply - Retail


I have no other S939 mobo. I could try the PSU on my old socket A computer that I am using right now. (With Vista and decent performance to top it all lol).

I have ~10 days to ship the x2 3800 back before I have to make a new RMA, and a total of maybe 20some days left. It was OEM from newegg. (I can only return it for a refund for the next few days...after that it is replacement only. 15% restocking fee hit as well as shipping)

Selling the RAM/CPU/PSU would net me about how much do you think?
AM2 prices are looking nice, and DDR2 as well.


What about going to a store, buying a 600+ watt quality PSU, and trying to boot with it, and if it does not work, return the PSU.

Also, found some posts about this problem, swapping in a new CPU and having this error. Even a person with the same CPUs and board as me. Many have had the problem solved by the long reset, or a new PSU. I am trying this procedure tonight...

Make sure that all 4 power connectors are plugged into the board(3 on the Expert)

# 24pin ATX connector.
# 4pin P4 or 8pin EPS +12v connector, usually near the ATX connector.
# Floppy power connector, under the CPU socket.
# Hard drive power connector, on top the chipset fan.(not used on the Expert)

Remove power from the rig by pulling the plug or switching the PSU off.

Move the PC Speaker jumper to the ON position.

Remove the Battery.

Move the CMOS jumper to CLEAR.

Press the Start Button on the case or motherboard to drain the capacitors.

You may leave the board in this condition for as long as it takes to clear the CMOS. If a short 30 second clear or a 15 minute clear doesn't work, try an 8 hour clear then a 24 hour clear.

Plug in a PS2 Keyboard.

Put one stick of RAM in the top orange slot marked DIMM2.(may vary on Infinity boards)

Replace the Battery.

Move the CMOS jumper to Normal.

Press and hold the Insert Key on the Keyboard.

Apply power to the rig by plugging the power in or using the switch on the PSU.

Press the Start Button to power the rig up.

When you hear the BEEP, release the Insert Key and press the Delete Key on the Keyboard.

Once you enter the BIOS set the DATE and TIME then Load Optimized Defaults.

Save and Exit.

Enter the BIOS again and set it up the way you want for your particular rig.

NEVER load any saved CMOS settings from CMOS Reloaded once you have a new BIOS installed.
Includes all current motherboards produced by DFI

If it does not work tomorrow morning, where do I go from there, seeing as I have limited time to return the x2 for a refund, and not to mention I need to do WoW arenas.

 

[CTho9305]

Was your BIOS up to date?

CTho9305, I know that you are not speaking for any company. However, if you WERE speaking for a company, do you think that we could get some inside penryn info ?

Wrong company, and no. I happen to like my job .

That and having a reasonable difference in temperature between cores, no 11-15C difference.

Well, for what it's worth, (assuming I'm reading these documents correctly, which I may not be) two CPUs at the same temperature could report up to 14C difference according to the PDF on socket 754 CPUs (I can't find the number for socket 939 CPUs)... and on top of that, the 3800+ looks like it burns more power than a 3000+ if it's clocked down but not completely idle (i.e. if the OS isn't putting it into the C1 halt state), according to this document. I'm not sure what conditions are required for the OS to / to not halt the CPU (are flash ads on web pages enough to keep it out of C1? Is an mp3 player like WinAmp?). Keep in mind that the PDFs spec max power - any given die could potentially leak less than the maximum limit.

Of course, something else could have been wrong... I guess there's no way to know at this point.

edit: Oh yeah, this might help with your Firefox profile issue.

 

[chaudx]

BIOs was newest, updated a week or two ago.
Why would they differ so greatly though, with no apps besides coretemp running. I even shut down non essential services.

Does the whole buying a PSU and testing with it sound like the best route to go then, at least for now?

BTW, AS5 isn't conductive right...so it wouldn't matter if some was near the CPU socket and got wiped up...

 

[VirtualLarry]

AS5 is slightly conductive (capacitive?). Be careful getting it on the chip package.

 

[myocardia]

Before you do anything else, try booting with each stick of RAM, one at a time. If it boots with one, you've found your problem. Also, your shutting down in the middle of a game is what XP does, if the RAM "dumps" while in use, unless you know how to tell it not to do that. Anyway, your problem is either one of your sticks of RAM or your power supply, but only one of them is free to test.

 

[MarcVenice]

"Cleared cmos for an hour, tried other method of reset, moved ram around, tried 4 different sticks..."

It's what I thought myo, but apparently he tried it. I wonder if he tried booting with 1 stick though. If you haven't you should definately try, like myocardio suggested!

 

[chaudx]

Defintly am trying with one stick, even borrowed some from my old computer to test with, One stick in any socket does not work, put it in orange now to try and start, according to DFI forum suggestion. CMOS long reset will be completed again this afternoon, as well as new power supply test.

 

[chaudx]

So, got a new quality 550W from Circuit City (Expensive =( ) and the results are below...

New Power Supply:
x2 3800 = 4 Red LEDs
3000+ = 4 Red LEDs

Old Power Supply:
x2 3800 = Powers on and then off
3000 + = 3 Red LEDs and long beep


What is broken, and where do I go from here?