How do CPU manufacturers know if you have overclocked?

[FreeStream]

Doesn't the Core2Duo have a memory that stores the highest temperature reached and the highest FSB? I seem to recall some post from somewhere that said they did, but I could be dreaming.

Well that could just be 400x6 for my E6600 which is near stock as it is

 

[WJA96]

Must have been dreaming m8

Maybe - but I'm fairly sure I went to a demonstration or a conference and someone from Intel was banging on about being able to read usage information from processors because of the high rate of claimed DOAs they experienced after overclockers fried processors. Now, whether he said they could do it or they wanted to do it is what I don't remember clearly.

 

[ted34]

because of the high rate of claimed DOAs they experienced after overclockers fried processors.

You mean to say that some of us lie to big companies about stuff we've broken - shocking.

 

[WJA96]

You mean to say that some of us lie to big companies about stuff we've broken - shocking.

Some of us lie to OcUK about stuff we've broken too if the posts on here are anything to go by!

 

[Durzel]

Overclocking doesnt really harm your CPU, too much voltage does.

That's not strictly true. CPUs have an 100% failure rate, they will all degrade over time to the point of failure. Obviously if they are run at their intended speed and voltage this period of time could be many years, but overclocking will reduce this time, and increased voltage will reduce it even more. That said even with overclocking most of us will have outgrown the processors long before it becomes a real issue.

 

[FrankJH]

I wonder if there is a way of "engineering" a weak link in a cpu which would break if something was tweaked too much - I realise Intel / AMD and the rest would be only too happy to find a way to do this so its unlikely (or at least not worth the cost to develop) - ie something to prove it could have only happened by OC'ing too far (even if this is only to the level just below the next cpu speed up and not an actual cpu failure like pushing too many volts through )

 

[Jokester]

You could easily build in a fuse I would imagine that would blow if you exceeded a set current through any part of the core, and it would only be possible to do if you increased the CPU voltage/clock rate beyond a reasonable amount.

I suspect in reality though that the losses from people clocking up would be more than offset by the increased costs of RMAs to the chip company.

Jokester

 

[FrankJH]

You could easily build in a fuse I would imagine that would blow if you exceeded a set current through any part of the core, and it would only be possible to do if you increased the CPU voltage/clock rate beyond a reasonable amount.

I suspect in reality though that the losses from people clocking up would be more than offset by the increased costs of RMAs to the chip company.

Jokester

Im sure you are correct - its not exactly as though CPU's in general (with correct cooling applied ) arent reliable, just one of the miriad of thoughts that popped into my head

 

[fornowagain]

Fudzilla: Intel knows if you overclocked

A few big retailers and etailers have confirmed that Intel actually can check if you overclocked your CPU. When you burn your CPU Intel asks its retailers and etailers to return the CPU's back and in three to five weeks Intel gets back with the answer to whether the end customer gets a new chip or not.

If Intel figurate out that if you load a CPU with too high frequency or too high voltage it might deny the RMA. This is a very unfair way of doing business, as the company doesn't oppose overclocking, at least not officially.

Yes, we are saying that if you overclock your CPU and fry it at a very high frequency, then there is a big chance that Intel will deny your RMA.

Retailers and etailers said that they are not sure in Intel's methods of judging who gets a new CPU, but they said that Intel gets back to them if you overclocked too much and simply refuse to RMA the part. This is more than disturbing news to all overclockers.

If you are one of these customers who didn’t get its RMA'd CPU repleaced because of overclocking, please let us know.

What's this about a black box, a permanent register of the voltage/frequency wtf?

 

[Cyber-Mav]

Thanks guys; I was just curious to know when people overclock there chips how it might affect there warranties because manufacturers state that any overclocking will void it.

Would the same thing apply to graphics cards (no way of knowing apart from using an electron microscope ) ? Or would the on-board memory act as a sort of black box/user history?

your first post makes it sound as your cpu is fried from overclocking.
if it is dead then i hope you did not get it from ocuk and try to send it back to them, the proof would be here.

iv seen this happen a lot of times to people.

but yes there is no feasible way of knowing a cpu was overclocked.

 

[rapid64k]

yes this correct Intel do have chip of sorts on the CPU that they get info from.

 

[drak3]

When you overvolt beyond reasonable limits, you actually short circuit the CPU since electricity will jump over to the next lane. This will betray you if you attempt to con them, but then again you can claim faulty capacitors (filters) caused this or any other sort of propagating burst in your whole circuitry. Unless it will help in a murder trial they wont bother checking I belive

 

[drak3]

yes this correct Intel do have chip of sorts on the CPU that they get info from.

It is possible, but I dont think intel architecture use any kind of non volatile writable memory onboard (since they do general purpose computers, which you use with software to adapt them to the task). Only special reconfigurable processors use flash based on board memory for the instruction code (in order to change it) and then again keeping a log in expensive memory inside the CPU to catch clockers would be a waste.