Probably a daft question

[franco_22]

I was just wondering. This is probably a really daft question but anyways:

Why doesn't INTEL and AMD sell their CPU's already overclocked?

For example ,if most i3s and i5s reach 4GHZ, why does Intel sell them at 2.66ghz ? Is it because of temperature issues? People not wanting to pay more for an after market cooler?

Same goes for AMD, if the 955 easily reaches 3.8GHZ, why is it sold at 3.2ghz?

The same for graphic cards, if most graphic cards are easily overclocked an extra 100-200mhz, then why dont companies just do that in the first place?

 

[Big.Wayne]

Why doesn't INTEL and AMD sell their CPU's already overclocked?

I think they work out the max speed and them "underclock" them and release them . . . slowly after time when people get bored with a number like 920 they clock them up and re-release them with a new number like 930 or 950 as a "new" product! . . .

nVidia got burned for doing this but somehow people turn a blind eye when its a processor . . . I think the trick is to be obvious about it rather than making up a new name! . . . i.e hidden in plain sight!

 

[thingemajib]

all to do with speed binning. As you may or may not know, each individual chip has its own performance characteristics. Some will hit 4.2ghz on air, others will struggle to get past 3.8ghz.

at the fabs, each silicon die is screened, and its designated to a certain chip (930/950/960 etc) depending on how well it performs, which products are currently in strong demand etc.

you've got to keep in mind, the majority of the chips will end up in some generic dell/hp, that will most likely be sitting in an office doing nothing more than word processing. If intel sells all its chips factory overclocked and overvolted, then they will need bigger coolers, which means more cost which means they are les attractive to the big OEM retailers.

Also, the higher the base clock speed (ie, if they were all sold at 3.8ghz), there will be less dies that will make this speed comfortably, hence more dies thrown out, hence loss of money.


edit:
sorry if that seems garbled. Just trying to get all my thoughts out...

 

[chipps101]

it is so they can still provide warranty even for cpu's that will be in the worst case scenario for them and not fail.

for example an i3 530 @ stock speeds with the stock cooler will hit the high 50's (under full load) in a poorly ventilated case add a few months dust to that and it will add a few more degrees and so on.

the warranty on those cpu's is three years so you can imagine in a poorly ventilated case that has never been cleaned if the cpu was running at a higher frequency it would overheat within the warranty period.

thats what i think anyway.

 

[thingemajib]

sorry for the DP

nVidia got burned for doing this but somehow people turn a blind eye when its a processor . . . I think the trick is to be obvious about it rather than making up a new name! . . . i.e hidden in plain sight!

I do see what you're getting at, but i have to disagree ... ish. The way Nvidia did it was stick the same chip (half the time with the same clocks) into a new card, and pass it off as a new generation along with the marketing hype to go with it, when there was actually no benefit to the end user at all. 8800gt > (older 65nm)9800gt spring to mind in particular.

with the processors, at least Intel/AMD are sticking to the same product families. I can easily see that an I7 930 is a tad faster than an I7 920 at a slightly higher price, or a althon 245 is a tad faster than a 240.

again, im probably confusing everyone, but i hope you can see what im getting at

 

[mmj_uk]

The TDP would be too high I suppose.

Both Intel and AMD's high-end parts have a TDP of 130W, so I guess that is about the upper limit without extreme cooling measures.

Even if they sold them with kick ass coolers and a TDP of 200W it's still not enough because they'd need to guarantee that everybody had a good case airflow, otherwise the heat would just build up inside until it overheats.

 

[Corasik]

Probably a million different valid answers to this one. Some already covered (binning, cheap oem cases, bad cooling, keep costs of stock coolers down etc).

Add reliability (overvolting can reduce lifespan)

Add stability (how long do people test their overclocks before declaring they are stable.) Is it stable if it primes for an hour, 10 hours, 100 hours.... I personally have had systems which have been prime stable for 24hours, only to fail during a 36hour prime session, and no, it wasnt a temperature issue .

Would you be happy if you know that your bank account was stored on computers with overclocked processors. How about computers used by doctors and scientists for medical research and treatments?

When you buy a processor from AMD or Intel, and run it at stock speed you are maximising the odds that there will be no errored calculations. Of course for a gaming rig, a minor error may not even be enough to cause a game to crash, and you may never discover that the processor made errors from time to time. Or you might just mark up a crash on microsofts "dodgy" O/S.

Truth is that if windows is crashing on a regular basis, its almost always a hardware issue.

Its great for enthusiasts that most processors clock really well, but if your after absolute reliability, and a reasonable assurance that the processor will meet the design specs, then the speed intel and amd select for "stock speed" is a very "safe" place to be.

For the record... I do overclock my CPU's, and I am not responsible for the balance on your bank accounts

 

[franco_22]

thanks guys, I knew it was a daft question. How did I not think of the thousands of office PCs worldwide that surely don't need 4ghz and a gtx 480