How much does overclocking affect CPU manufacturers

[Idontcare]

It is an absolute steal, and by the time you have arranged all the shiny gubbins that us overclockers "need" (decent OCing mobo, SSD, overpriced RAM, overrated PSU, aftermarket cooler) they look almost criminally underpriced.

Doh, I knew I was forgetting something. Yeah, cooling. An H100, sell me some tubing and fishtank pump to move some water around and I'll give you $125.

Sell me four noizeblocker fans for another $120 and I'm now $250 into a cooling device for my $315 CPU.

Compare that to a CPU that took 4+ yrs to develop and likely at an expense of around $3B-$4B, and can only be manufactured in a fab that cost $3-$4B to setup, in addition to requiring a manufacturing process that itself took 4yrs to develop at an additional expense of $3-$4B.

Intel spends all that money to make a sophisticated product that no one else can make and yet I will spend nearly as much on some tubing, a fishtank pump, and four fans. :\

 

[iCyborg]

Its his dream rig for Gods sake. Let the guy live a little.

Look you even made him change the 580 to 5450 in his sig.

I believe the plan is to buy 7970 asap, and the poor guy that he is, he sold 580 while they still fetched a good price

 

[lOl_lol_lOl]

I believe the plan is to buy 7970 asap, and the poor guy that he is, he sold 580 while they still fetched a good price

Being the poor student he is, he probably donated the 580 to the homeless as part of an effort to help them survive in society through CUDA programming. :thumbsup:

 

[Concillian]

I bought a 2600K which at the time was about the most expensive top-performing CPU I could throw my money at in terms of single-threaded performance (and multi-threaded provided we aren't talking those handful of emberrasingly parallel situations where SB-E and Gulftown thump it still)...and despite Intel extracting as much "value" from me as a consumer as they did, I still gave ASUS even more money than Intel (my MIVE-Z cost more than my 2600K), and the same was true for OCZ (my 240GB V3 is the most expensive component in my rig).

The price compression in SKU's that are not the absolute top-tier xtreme SB-E stuff is just amazing. The mind boggles at times that here you have a $200-$300 peice of silicon that makes everything go and yet the mobo itself, or the storage device, can fetch an even higher ASP.

Seems a little out of whack if you ask me. The 2600K could have been a $500 cpu and I am pretty sure I would have still bought it and felt it was priced right. That it was only $300 was nice though, I hardly feel taken advantage of by Intel in the grander scheme of things.

I think the entire system you're talking about is a totally different world. In my current rig the i3-530 has the highest total price of any one component aside from the video card (HD5770 that at the time was around $150).

If you combine the costs in motherboard and CPU, to get a single threaded upgrade from SB, I'd need to spend roughly double what I did for my i3-530 + mobo. It's real.

IMO the squeezing has been even larger on the motherboard side of things as the CPU side. You can't get an Hxx chipset like clarkdale and previous generations. A >$300 motherboard is somewhat of a foreign concept to me, given everything truly necessary is available on a <$100 one... except the ability to overclock. It's gone from being able to do some pretty reasonable OCing with a $60-80 board to a ~$120 price of entry or so.

The people buying things like H100s, $20 fans, and such are not the same people who were overclocking PPro 150s to 180 or P90s to 120s. The whole landscape has changed. I'm not whining about it, overclocking of the past was getting more than you paid for, and it definitely would be considered stealing in other industries. It was good while it lasted, but those days are gone. The end was inevitable. That doesn't mean I have to like it, though.

 

[Idontcare]

The people buying things like H100s, $20 fans, and such are not the same people who were overclocking PPro 150s to 180 or P90s to 120s.

Since I have personally done both, and the voices in my head assure me I am but one person in the physical world, I'm confused by what you are intending to communicate to me here.

...and it definitely would be considered stealing in other industries.

True. Just try getting away with buying Windows 7 Home Premium and "overclocking" or "unlocking" it to have the same capability of Windows 7 Ultimate. Not kosher in today's society.

But do the same with the shaders on your HD6950 or your 2500K and its all ethically/morally/legally legit.

I get that the difference is a matter of ownership versus licensing...we own our CPU's whereas we merely license the rights to use a specific feature set of our Operating System.

But its only a matter of time before the CPU guys figure out a way to sell us CPU licenses rather than selling us the CPUs. If I was making 6-figures at Intel as an accountant or sales/marketing that is what I'd be trying to do anyways.

 

[nyker96]

I think OCer is like 1% in the population I doubt people get affected much. But it's those system builders who overclock a low end chip then sell it at a high end price that worries cpu makers most.

 

[jpiniero]

I think OCer is like 1% in the population I doubt people get affected much. But it's those system builders who overclock a low end chip then sell it at a high end price that worries cpu makers most.

Do whiteboxes even exist anymore?

But do the same with the shaders on your HD6950

I imagine the overclockers market has always been a little strange to the cpu manufacturers. On one hand, the overclockers try to save some bucks by buying a cheaper chip and overclocking but at the same time they also upgrade more often than the regular populace. At least now Intel has been able to get the overclockers to buy a mid-high end product (like the 2500K) as opposed to the cheapest thing they make (like the 300a)

I think the ability to unlock shaders/core was intentionally done by AMD to help sales actually.

 

[dagamer34]

Doesn't Windows 7 report the original CPU clock speed in System in the Control Panel? Of course, it's not like any customer actually expects to get different components than what they paid for...

 

[TheDug]

Welcome to the forums TheDug! :thumbsup:

I actually forgot my old username and password and the email it was associated with so just made myself a new one.


Personally no overclocking on the CPU affects me a lot. I'm currently on an E5200 overclocked to 3.4Ghz so as fast as a stock e8400 with a 4850 graphics card. I remember my CPU cost me exactly half the price of an E8400! I can afford an i5/i7 but the thrill of getting a cheap ass part to work faster than rated is gone now .

 

[Idontcare]

I actually forgot my old username and password and the email it was associated with so just made myself a new one.

You might be surprised by the number of banned members who suffer the same "memory loss" issues. :|

 

[TheDug]

You might be surprised by the number of banned members who suffer the same "memory loss" issues. :|

Me banned? No way! Don't they ban by noting IP addresses?

 

[HumblePie]

Way back in 1995/1996 when I worked on the original Pentium, there were really no limits on overclocking at all. Multipliers were so unlocked that it would just do whatever you told it to - there were no fuses, so we relied on the BIOS to set the value correctly. So if you bought a Pentium 90, then the chip just relied on the system builder to set the ratio correctly.

At that time, we at Intel had a pretty substantial problem with "remarking" in which various organized criminals would change the markings on the chip to make a 90MHz CPU look like a 120MHz CPU. Then when the customers returned their CPU's back to Intel because they were defective, we would find out that someone had actually just changed the laser etching. They'd sand down the cover - or just pop it off and replace it - and then re-etch it. This was actually a really seriously expensive problem - a criminal could buy a 90MHz CPU and then change some writing on it and resell it for much more money and when it had problems, Intel was left to foot the bill.

So I remember when we "locked" the first multipliers back then. I was in the meeting with the engineers and the marketing team and I sat in the corner and didn't say much, but I can say that I was there and there was absolutely no talk about the enthusiasts overclocking... at that time in ~1996, the first clock multiplier locks were all about limiting losses due to organized crime.

Clearly over time since 1996 things have changed, and my experience 15 years is likely barely relevant to the modern industry except as a (hopefully) interesting anecdote. I actually have no clue about the industry impact nowadays. But I do know that ~15 years ago, there was no discussion at all about overclocking impacting revenues and the first multiplier locks - which I helped work on personally - were definitely not implemented with overclocking enthusiasts in mind.

*Not a spokesperson for Intel Corp. *

Great story. I have a few family members who worked for Intel for a long time. Basically from what I understand was Intel never became aware of OCing for enthusiasts as a market until they released the Celeron line to combat AMD's K6 2 line. At the time a Pentium 2 @ 450 Mhz was going for $800 off price grabber. The 400 Mhz version was $600. AMD had released the K6 2 line, which while slower, was far less pricey. A 500 Mhz, overclockable to 600 Mhz, K6-2 released at the same time as the Pentium 2 line was going for $80-$100. Sure the actual performance was about 3/4ths of the Pentium line, but for the price versus performance ratio it couldn't be beat. Intel noticed a small dip in sales and a rise in sales for the K6 2. They though it was because of the price point only. Which was a big factor, but didn't realize a big reason at the time was because of the enthusiasts that would rather pay $80 for a chip and OC it up 5/6ths the performance of a $600 chip.

So Intel a few short months later introduced the original Celeron. Since they didn't have any design plans in place for the Celeron at the time the Celeron literally was "bad" pulls of Pentium 2's but with a motherboard bus speed rate "limited" to 66 Mhz versus the 100Mhz. They then priced them out to be competitive with AMD's K6-2 line at $80.

When consumers, like myself, realized that the Celerons were actually $600 and $800 P2's that were stuck at 66Mhz instead of 100Mhz, it wasn't long before we were buying the motherboards from Abit and others that allowed manual manipulation of the motherboard bus clock. So buying that original Celeron 300 Mhz and cranking the FSB from 66 to 100 yielded a 450 P2. Or if you got really lucky like me and tried one of the Celeron 333's you MIGHT get a 500 Mhz chip. Faster than any chip being sold on the market. This is what I did and I did get lucky. From what I understand Intel saw sales for the Celeron go through the rough and finally figured out why. So they went around trying to lock down the Celeron and neuter it further while coming out with actual "designs" to make a neutered chip to compete with the AMD lines at the time. I think Intel spent too much time doing this and AMD finally got the jump on them with the K7 line. It woke them up somewhat.

Still it's a niche market and they realize they can use it to squeeze a little bit extra out of it. The threat of organized crime and fraud of relabeling CPUs is a thing of the past.

 

[velis]

Here's my take at the OP:
Previously, OCing didn't really have much of an impact on the overall CPU sales. I suppose the figures are below 0.1 percent of market (compare enthusiast forums member counts with annual CPU sales and you'll get a glimpse of the picture).

The introduction of K series CPUs by Intel was a stroke of genious IMO: Not only did they give us overclockers back what we always wanted (easy overclocking - multiplier only), but the also made sure all of us buy the chip that's "reasonably" priced vs previous instances where we bought the cheapest stuff available and nitrogen cooled it into oblivion.

Additionally, since the enthusiast community is all happy about these chips, not only do we buy them, but also people who have no idea what K stands for nor what the difference between K and non K means for them. So I suppose Intel at this moment knows exactly how many enthusiast level chips they sell, but I should be so brave to guess that not even 10&#37; of these chips get overclocked. As one of my friends with a 2600K said: "there's just no need to". So they are actually getting a great margin for non K chips that are sold as K variants just for the hype in the community.

 

[TheDug]

All people have is speculations. Thanks for your thoughts everyone, but does anyone have facts and figures?

 

[Idontcare]

The threat of organized crime and fraud of relabeling CPUs is a thing of the past.

I can't help but chuckle when I read this and thinking to myself "yeah, because Intel doesn't like the competition!" (in reference to why the government doesn't like the mafia )

Me banned? No way! Don't they ban by noting IP addresses?

All people have is speculations. Thanks for your thoughts everyone, but does anyone have facts and figures?

The data exists but there is zero chance it will ever make it to the public domain. That kind of data is priceless for competitive reasons, no business would willingly part with it.