HOLY SH@T 73C for 3000+A64????

[iamtrout]

I just built my brand new system:

A64 3000+
Shuttle AN51R

My temps are showing 73C for the CPU! I'm using the stock cooling while waiting for my watercooling stuff to come in. I've always had my temps below 50C, and above 50C I started to worry.

What's going on? Is this safe? Should I shut off my computer right now? The heatsink only feels warm.

 

[MegaWorks]

WOW!, shut it down and see if your heatsink is well installed on your CPU.

 

[aphex]

My old cpu used to run around 60C idle and 68-70C under full load.....

My new Athlon 64 3200+ runs around 37C idle and 45C under load.

 

[Nickel020]

If the fan is spinning either your monitoring chip is broken or the heatsink isn't properly installed.

 

[Jeff7181]

You didn't just set the heatsink on the CPU for testing purposes like someone else did, did you?

 

[iamtrout]

Hehe, looks like the thermal monitoring chip is fuXX0red.

Here's what happened:

Saw Anandtech posters tell me to shut my computer off, so I did.
Unclamped the HSF, only to find that it was stuck to the A64 with the thermal grease.
Pulled.
HSF came off, with the CPU STILL ATTACHED TO IT. Meaning it got yanked out while the ZIF socket was in the locked position.
Screamed.
Saw bent pins.
Screamed.
Carefully bent them back, telling myself that if the CPU is dead, I'll just say that it was DOA.
Took off old compound, put AS5 on CPU and HSF.
Remount.
Same old temperature.

Dammit. Anyone know if the AN51R has thermal monitoring issues?

 

[Mo0o]

AN51R.. is that an SFF? I thought shuttle SFFs had realy good cooling. My idle is around 35 right now

 

[iamtrout]

No, it's not SFF. And from all the pictures of the BIOS health monitor from reviews, it definitely looks like mine's got some issues.

linky to Shuttle AN51R

 

[iamtrout]

Just wondering... the thermal sensor for A64's is on the mobo, right, or the CPU?

 

[SplinterDB]

HSF came off, with the CPU STILL ATTACHED TO IT. Meaning it got yanked out while the ZIF socket was in the locked position.
Screamed.
Saw bent pins.
Screamed.


ROFL. Not at you I have been there.

 

[Nickel020]

Well the A64 has an on-die diode to monitor the temp but in order for that temperature to be read there's a monitoring chip on the mobo. That chip also measures the mobo temp itself (the mobo temp is what most people refer st as system). Dependingon where that chip is on the mobo the mobo temp can vary quiet a bit and therefore the sys temp reading is useless.
You could try a BIOS update; the chips sometimes get re-calibrated with BIOS updates. If that doesn't help then something has to be wrong with your monitoring chip.

 

[Jeff7181]

Originally posted by: iamtrout
Hehe, looks like the thermal monitoring chip is fuXX0red.

Here's what happened:

Saw Anandtech posters tell me to shut my computer off, so I did.
Unclamped the HSF, only to find that it was stuck to the A64 with the thermal grease.
Pulled.
HSF came off, with the CPU STILL ATTACHED TO IT. Meaning it got yanked out while the ZIF socket was in the locked position.
Screamed.
Saw bent pins.
Screamed.
Carefully bent them back, telling myself that if the CPU is dead, I'll just say that it was DOA.
Took off old compound, put AS5 on CPU and HSF.
Remount.
Same old temperature.

Dammit. Anyone know if the AN51R has thermal monitoring issues?

Say it with me... twist and pull... not yank.

 

[iamtrout]

Hehe, I couldn't twist because the damn plastic square thingy around the HSF was keeping it from twisting.

I just updated the BIOS, and the CPU temp in BIOS is now showing a cool 38C. BUT software temp monitors in Windows still show CPU temp as being 73C. I've tried both MBM and MBProbe. What should I do to get the CPU temp to read correctly? I have no idea what model of temp sensor I have.

 

[Nickel020]

Check the mobo list on the mbm site and check if your sensors are assigned right. Sometimes there are sensors that give random temps (sometimes even below 0 C). Maybr one of them is assigned to the CPU temp instead of the actual CPU sensor.

 

[iamtrout]

Eh, looks like the MBM site is down, not to mention they stopped development of it a few months ago I think. Looks like the only thing I can do is calibrate the temperature readings 34C BELOW what the sensors are telling me in Windows (so 73C-34C=39C, which should be about normal idling temp, right?)

The problem is that under load (running SETI) it only goes up to 40C after calibration..... basically it matches the chipset temp, and the chipset is passively cooled.

I'm also a bit confused with the temp readings.
There's a CPU temp, a MB temp, and an AUX temp. What the heck is the AUX temp? It's running 2C hotter than the CPU temp under load. I'm running an nForce3 250.

 

[cluy44]

Originally posted by: SplinterDB
HSF came off, with the CPU STILL ATTACHED TO IT. Meaning it got yanked out while the ZIF socket was in the locked position.
Screamed.
Saw bent pins.
Screamed.


ROFL. Not at you I have been there.

its a very traumatizing thing to have happen, i had 2 bent pins and luckily i managed to straighten them out.

 

[MrCodeDude]

MBM sucks with A64 boards. It said my chip was running at 70°C, while Gigabyte's EasyTune4 said it was running at something like 30°C

 

[imported_Nacelle]

You didn't put too much thermal greese on there, did you? If it's more than a real thin layer, it's acting as an insulator.

 

[DotheDamnTHing]

rma the board

 

[iamtrout]

The thermal grease was actually VERY thin. I've become pretty proficient at it over the years. I should RMA the board just for this? The BIOS displays the temp correctly, I think, it's just that for some reason MBM and MBProbe are adding ~34C to it.

What does AUX temp mean if I already have a chipset and cpu temp?

 

[JetBlack69]

I say as long as it's stable, who cares what the temp is. Once you start having stability problems or other problems, then check what the temp is.

Does shuttle have software to check the temp, or are you still using MBM? Since the BIOS seems to be detecting the temp correctly there shouldn't be a problem.

 

[iamtrout]

Ehhh... the problem is that it's TOO stable. Under load (Prime 95) the CPU temp increases by only 2C. If it's reading the temp increase wrong, then that would definitely be a problem because what if it was actually increasing by 15C? That would be bad.