When thermal throttling fails

[Pinnacle]

Full story: http://www.theinquirer.net/?article=30863

[STEvil]

Some quick observation that the pins welded themselves to the contact points leads me to believe that the CPU "overheating" wasnt really the problem.

As i've said before and as i'll probably say again: LGA sockets suck.

[zabomb4163]

even with thermal protection off the processor will cut off at a certain point. it has an internal limiter. motherboard problem. not processor

[n00b 0f l337]

Looks more like corrosion.

[MaxxxRacer]

even with thermal protection off the processor will cut off at a certain point. it has an internal limiter. motherboard problem. not processor

Vey true. If hte CPU reaches 135C in the FP adding unit (assuming prescott), which is the hottest part of the core, it will immediately shut off without warning. Its an instant and absolute hard wired shut off. FWIW the rest of the core is around 100-115C when the FP adding unit is at 135C.

If you dont believe me, go read the Intel Tech doc on LGA Prescott CPU's. I did.


Also the reason the pads look like that is because he ripped the cpu out of the socket when the pads were soldered to the pins. This is very odd considering that Gold (the plating on the pins and pads) doesnt get anywhere near close enough to melt at 100C. There must have been some sort of chemical on the pads or pins that caused them to solder together in such a way.

[Celeron Gamer]

When the CPU touches the TDP it should automatically shut off. Happens all the time to inappropriate cooler mounts. But it certainly couldn't core, as the transistors hasn't been damaged. And the CPU still booted. Something is wrong with the mobo, as the wafer is damaged.

[[XC] MarioMaster]

definately something wrong there. i know my thermal protection works from the time i forgot to turn on my water pump. doh

[WeStSiDePLaYa]

this is complete BS. this is NOT a chip overheating.

first off, if the chip over heated, why would the pins solder themselves to the board, but yet the smd caps stay soldered, with no signs of the solder being remelted.

also, no way the pins could solder themselves to the board. it doesnt get anywhere near hot enough to do so.

also, why would the burn marks be on the pins ONLY? they would be the coolest part of the chip, with the underside of the core being the hottest(besides actual core)

this is just dumb.

[Bloody_Sorcerer]

pure bull. nothing on a CPU will melt at 100C, nor on the motherboard. really really low temp solder... maybe. gold pins/contacts? no. looks a lot more like corrosion, maybe a slow leak in his loop caused some corrosion and went unnoticed until his pump died.

[rhino56]

my best guess would be that he had water on the cpu and it shorted out. which could cause that type of heat from some high amps directly shorting out.

so i would say that entire story is bs and someone is looking for a rma.
if you mess up while overclocking and fry something be accountable for it and take the loss.
all too often i see rmas that are not valid. it costs me and you money.

unethical bastards

[nn_step]

Actually if I remember correctly we all can do that if we just connect the Thermal Diode pin to a Gnd Pin..
we should have a contest called how burnt your Prescott can make its self

[n00b 0f l337]

I've seen the same effect before with people using phase 24/7 and are not perfectly insulated. Corrosion is a killer!

*And even then the chip will work in a state like that*

[Formann]

Looks a bit like my old 3200+ claw .. sloppy mounting of evap and lots of volt ..

3-4 pins had marks like these:




BTW: This is after cleaning it with alcohol and a steel brush. It was much worse before.

[einCe]

well, thre are allot of speling errors in the articel last time i checked the inq didnt posdt wiht errs in their acrticles

[Celeron Gamer]

well, thre are allot of speling errors in the articel last time i checked the inq didnt posdt wiht errs in their acrticles

I just realized it was full of grammatical, vocab, and punctuation errors