Is intel underclocking it's CPUs so they can release higher clock models in the future without any difficulty?

[kobymu]

Originally posted by: JEDIYoda

Originally posted by: lookin4dlz
This is a really stupid post, I can't believe the conspiracy theories people come up with.

exactly...lol.....could be an ex amd fanboy..lol

Intel has done this before, blast from the past:

http://www.tomshardware.com/1997/01/02/overclocking_guide/page4.html

Intel is actually releasing different kinds of Pentiums. Some are not capable of multi processing, others have to have a higher supply voltage to run correctly. You have to realize, that Intel is not having a different production line for each clock speed, but is in real now producing more or less only one dye. After the chips have been produced, they are tested, and regarding how they performed in the test, the clock rates are chosen. Some chips would work correctly on a special clock rate only with a higher voltage, which means also getting hotter. Others wouldn't work with multi processing. To find out which kind of CPU you have or are offered, you have to have a look on the bottom side of the chip, where's written something like e.g. SK 106 SSS. These last three letters can tell you what kind of CPU you have, and here's the list . It's obvious, that you should try hard getting a chip, that runs at 3.3 V(S tandard) rather than 3.4 to 3.6 V (V RE), and maybe it makes you feel better to know that you've got a real Pentium, that works with multi processing as well, though you won't be able to use multi processing on a Triton board. If you should be into overclocking, it's almost inevitable to get a SSS-chip, because than you still can increase the voltage on the motherboard to get it running at a higher clock speed (as long as you cool it well!).

Good memories

Same more info on page 10

http://www.tomshardware.com/2003/07/03/getting_up_to_speed/

During the days of the AMD K5 and the Pentium MMX, this was still child's play because the multiplier could be freely selected most of the time. It wasn't seldom that a Pentium 166 MMX could be run at 200 MHz or even 233 MHz, because ultimately, the chips came from the same production line and were labeled according to the demand, and it's the same with chips today.

And if somenoe will call me a fanboy i'm gonna slap him with more links and even even bigger quotes.

 

[superHARD]

Why remake something when you can just relabel it?

RAM, Video cards, CPU's, stuff like...this has been going on for years...

 

[HurleyBird]

kobymu:

Irrelevant and everyone here already knows that.

 

[kobymu]

Originally posted by: HurleyBird
Irrelevant and everyone here already knows that.

My experience here tales me most don?t.

/edit

Irrelevan...

A) Its comes to show past track record.
B) Even though manufacturing technique have changed, its important to remember there are other sides to this, for example one can speculate that, since a modern CPU manufacturing process get constantly optimized, at least until certain stage in the its life, Intel has decided that all resources in that department should focus on better yields only, and not waste these resources on tweaks that will allow higher frequencies.

 

[Hulk]

The current top clocked C2D is 2.93GHz. *Most* people are only overclocking about 500MHz beyond that. I think Intel could release a 3.2GHz part if they needed to but they don't.

One model below their current top of the line part can soundly beat their competitor's fastest offering. What is the point for them to offer faster parts now? They aren't going to be able to sell a faster part for more than $1000. History has shown that about as much as you can sell a CPU in any type of meaningful quantities.

And simply adding another part at 3.2GHz would mean reducing the price of all parts below it. Again not necessary since no one is breathing down their neck.

If I were Intel I'd do exactly what they did. Release a range of chips with better performance and lower prices than their competitor at every price point, keep thermals in check, and sell as many parts as possible. And all the while they can bin the best parts and have a little something held in reserve for AMD if they come up with something. There is simply no reason for them to show their hand at this point in time.

They will have some price cuts in Q2 just to give the market a little boost but nothing major will happen until AMD puts some heat on them.

That is my analysis.

 

[HurleyBird]

Originally posted by: Hulk
And all the while they can bin the best parts and have a little something held in reserve for AMD if they come up with something. There is simply no reason for them to show their hand at this point in time.

Do you really think that if the current C2D's were 400MHz faster or slower that it would affect what AMD comes out with? Not really. AMD is going to come out with the best chips that they can when they launch Barcelona.

I also agree that Intel could release a 3.2GHz part, but for a different reason. All the best chips are going into quad core.

 

[Roy2001]

The rumor is Penry can hit 3.7Ghz easily on air, and perhaps within TDP, in labs. I am not sure TDP is for laptop (35W) or desktop (65W/80W).

So when K8L arrives, we should see faster 65nm C2Ds and 45nm C2Ds later this year or earlier next year. If current C2D cannot compete with K8L, I believe Intel would release Penryn core C2D as extreme addition to spoil AMD's lunch at least.

 

[HurleyBird]

I think that it is possible that Intel focusing on 45nm instead of ramping up Conroe's clock speed.

 

[Hulk]

Originally posted by: HurleyBird

Originally posted by: Hulk
And all the while they can bin the best parts and have a little something held in reserve for AMD if they come up with something. There is simply no reason for them to show their hand at this point in time.

Do you really think that if the current C2D's were 400MHz faster or slower that it would affect what AMD comes out with? Not really. AMD is going to come out with the best chips that they can when they launch Barcelona.

I also agree that Intel could release a 3.2GHz part, but for a different reason. All the best chips are going into quad core.

I didn't say that. Read the quote. Intel has no reason to release faster parts right now because AMD isn't pressuring them.

 

[HurleyBird]

Originally posted by: Hulk

Originally posted by: HurleyBird

Originally posted by: Hulk
And all the while they can bin the best parts and have a little something held in reserve for AMD if they come up with something. There is simply no reason for them to show their hand at this point in time.

Do you really think that if the current C2D's were 400MHz faster or slower that it would affect what AMD comes out with? Not really. AMD is going to come out with the best chips that they can when they launch Barcelona.

I also agree that Intel could release a 3.2GHz part, but for a different reason. All the best chips are going into quad core.

I didn't say that. Read the quote. Intel has no reason to release faster parts right now because AMD isn't pressuring them.

That's absurd, looking at the margin hit Intel took and the market share that it lost in desktop and notebook last quarter, AMD most certainly is pressuring Intel. If Intel could magically increase the frequency of their entire C2D line, they would. It would help them to compete against AMD's lower end offerings, and it wouldn't be 'showing their hand' as AMD certainly has a good inclination how much or little C2D's clocks can ramp at 65nm.

All I'm saying is that Intel isn't purposefully under clocking their entire product range. Doing so would be moronic in the extreme.

 

[Viditor]

Originally posted by: HurleyBird

Originally posted by: Hulk

Originally posted by: HurleyBird

Originally posted by: Hulk
And all the while they can bin the best parts and have a little something held in reserve for AMD if they come up with something. There is simply no reason for them to show their hand at this point in time.

Do you really think that if the current C2D's were 400MHz faster or slower that it would affect what AMD comes out with? Not really. AMD is going to come out with the best chips that they can when they launch Barcelona.

I also agree that Intel could release a 3.2GHz part, but for a different reason. All the best chips are going into quad core.

I didn't say that. Read the quote. Intel has no reason to release faster parts right now because AMD isn't pressuring them.

That's absurd, looking at the margin hit Intel took and the market share that it lost in desktop and notebook last quarter, AMD most certainly is pressuring Intel. If Intel could magically increase the frequency of their entire C2D line, they would. It would help them to compete against AMD's lower end offerings, and it wouldn't be 'showing their hand' as AMD certainly has a good inclination how much or little C2D's clocks can ramp at 65nm.

All I'm saying is that Intel isn't purposefully under clocking their entire product range. Doing so would be moronic in the extreme.

Increasing the clockspeed on C2D wouldn't help Intel's margins in the least. Keep in mind that C2D is only going to be 40% of their available chips by the END of this quarter.
At the current clockspeeds, C2D is clocked at exactly what it needs to be for competition with AMD, and as AMD releases their OWN next gen design, Intel will still have a few "clubs in the bag" to help them be competitive. Remember that Intel have no architectual changes between now and Nehalem, so clockspeed (both from the headroom on current offerings and the die shrink of Penryn next Jan) is their only real recourse to countering the "Stars" chips and Barcelona.
AMD did exactly the same thing for the C2D release, though they had less headroom to play with.

 

[HurleyBird]

Originally posted by: Viditor
Increasing the clockspeed on C2D wouldn't help Intel's margins in the least. Keep in mind that C2D is only going to be 40% of their available chips by the END of this quarter.

Well, its more of a choice between increasing margins (selling some more pricier, higher clocked stuff while keeping prices the same for the rest of the line) or putting pressure on AMD (increase the clock speed around the board, but keep price the same so AMD no longer is competitive with the low end C2Ds), either way, increasing clock speeds would help Intel out now (as well as many moths earlier) and it would be pretty stupid of Intel not to release higher clocking parts if they actually could.

Originally posted by: Viditor
At the current clockspeeds, C2D is clocked at exactly what it needs to be for competition with AMD, and as AMD releases their OWN next gen design, Intel will still have a few "clubs in the bag" to help them be competitive.

AMD will release their next gen chips so that the highest end chips will struggle to find a 200MHz over clock in stock conditions. They can do this because they use SOI (and Intel can't do this because they use bulk silicon --if Intel used SOI you would see the same behavior from them). As I said earlier, it wouldn't matter if C2D's were 400 MHz slower or 400MHz faster, Barcelona will still launch at the same speed and performance. Also, you aren't thinking very hard if you don't realize that AMD knows what current C2Ds can over clock to.

Originally posted by: Viditor
AMD did exactly the same thing for the C2D release, though they had less headroom to play with.

That's a pretty crummy (and false) example.

 

[VirtualLarry]

Originally posted by: Skotty
I believe the C2Ds are currently clocked lower than they could be. They are the first of a new generation that far exceeds the previous, but I think Intel likes to have more incremental improvements. While Intel spends the next however many months or years working on the next leap of architecture, their current line of processors can continue to "advance" by slowly clocking them up to what they are really capable of. I'm sure the reality is a little more complicated than that, but in general, that's how I think it works.

Exactly. It reduces overall R&D expenditures while maintaining a stable profit margin. If Intel released their new chips as fast as they could possibly go, then just possibly, their competitors would do the same, and then Intel would be faced with having to introduce a new generation of chips, which might not be ready for introduction yet. It would greatly increase the overall R&D neccessary, while not substantially increasing their profits.

 

[HurleyBird]

Originally posted by: VirtualLarry
Exactly. It reduces overall R&D expenditures while maintaining a stable profit margin.

What?

Originally posted by: VirtualLarry
If Intel released their new chips as fast as they could possibly go, then just possibly, their competitors would do the same, and then Intel would be faced with having to introduce a new generation of chips, which might not be ready for introduction yet.

They only have one real competitor -- and that competitor is going to release the fastest chips it possibly can anyway, thanks to its SOI process.

Originally posted by: VirtualLarry
It would greatly increase the overall R&D neccessary, while not substantially increasing their profits.

The debate in this thread isn't whether Intel can put in the R&D to make faster parts, it's whether Intel is intentionally under clocking their CPUs or not. If Intel *is* 'under clocking' their CPUs it would take zero R&D to release faster versions.

 

[Viditor]

Originally posted by: HurleyBird

Originally posted by: Viditor
Increasing the clockspeed on C2D wouldn't help Intel's margins in the least. Keep in mind that C2D is only going to be 40% of their available chips by the END of this quarter.

Well, its more of a choice between increasing margins (selling some more pricier, higher clocked stuff while keeping prices the same for the rest of the line) or putting pressure on AMD (increase the clock speed around the board, but keep price the same so AMD no longer is competitive with the low end C2Ds), either way, increasing clock speeds would help Intel out now (as well as many moths earlier) and it would be pretty stupid of Intel not to release higher clocking parts if they actually could.

The problem with your hypothesis is marketing...releasing higher end products tends to marginalize the lower end, and it's the lower end that is (by far) the biggest seller.
Keep in mind that Intel is selling every single C2D they can make. If they made a new higher speed binned chip, it would lower the cost of existing (highest volume) chips downward (they can't keep them the same because of their pricing structure).
This is the danger of reducing prices to gain marketshare...if AMD hadn't been kicking the snot out of Intel for 2+ years, I can guarantee you the C2D wouldn't have launched at such a low price. The problem is that once you lower prices for a line of chips, it's impossible to raise them again until you release a new line.

Originally posted by: Viditor
At the current clockspeeds, C2D is clocked at exactly what it needs to be for competition with AMD, and as AMD releases their OWN next gen design, Intel will still have a few "clubs in the bag" to help them be competitive.

AMD will release their next gen chips so that the highest end chips will struggle to find a 200MHz over clock in stock conditions. They can do this because they use SOI (and Intel can't do this because they use bulk silicon --if Intel used SOI you would see the same behavior from them). As I said earlier, it wouldn't matter if C2D's were 400 MHz slower or 400MHz faster, Barcelona will still launch at the same speed and performance. Also, you aren't thinking very hard if you don't realize that AMD knows what current C2Ds can over clock to.

There are several points here...
1. AMD will release at whatever speed it takes to surpass C2D in both performance and power/performance...and not much more (for the same reason Intel doesn't).
2. Barcelona/Stars chips will be vastly different to current Opteron/X2 chips (it's a whole new architecture), and that includes their overclockability. SOI is only a very small part of the story...they have a whole new power section, (hopefully) a new DSL process, and are second generation 65nm (remember that C2D is also second generation 65nm).
3. Raising clockspeed also raises the TDP

Originally posted by: Viditor
AMD did exactly the same thing for the C2D release, though they had less headroom to play with.

That's a pretty crummy (and false) example.

How is it false (or crummy)? Both Intel and AMD have done this since 1999...

 

[Hulk]

Originally posted by: HurleyBird

Originally posted by: Hulk

Originally posted by: HurleyBird

Originally posted by: Hulk
And all the while they can bin the best parts and have a little something held in reserve for AMD if they come up with something. There is simply no reason for them to show their hand at this point in time.

Do you really think that if the current C2D's were 400MHz faster or slower that it would affect what AMD comes out with? Not really. AMD is going to come out with the best chips that they can when they launch Barcelona.

I also agree that Intel could release a 3.2GHz part, but for a different reason. All the best chips are going into quad core.

I didn't say that. Read the quote. Intel has no reason to release faster parts right now because AMD isn't pressuring them.

That's absurd, looking at the margin hit Intel took and the market share that it lost in desktop and notebook last quarter, AMD most certainly is pressuring Intel. If Intel could magically increase the frequency of their entire C2D line, they would. It would help them to compete against AMD's lower end offerings, and it wouldn't be 'showing their hand' as AMD certainly has a good inclination how much or little C2D's clocks can ramp at 65nm.

All I'm saying is that Intel isn't purposefully under clocking their entire product range. Doing so would be moronic in the extreme.

Despite your implying that I am "moronic and absurd" I will repond to your post in the name of having a lively discussion. But I will not insult you personally. I respect your opinion. I just have a different one.

Do you follow the stock market? The reality of the current situation of a company like Intel or AMD generally lags the market by a few quarters. It takes a while for sales to turn into reported profits. Intel only gained the upper hand around Q2/Q3 of 2006. We're only 2 quarters ahead of that now.

Have you NOT noticed all of AMD's chip prices plummeting in the last few months? Do you think they just decided to "have a sale?" No, Intel is hurting them and Intel knew it would only take a range of 2.93 to 1.83GHz to do it. If they needed it to be 3.2GHz to 2.13GHz I'm sure they could have managed that since very few 3.2's in comparision to the lower end chips need be available.

I have been watching the computer tech scene for about 20 years, since college, and when AMD or Intel can respond they do. AMD can't respond to Intel's current line of C2D's and Intel knows it. I will say it again. There is no need for them to release higher clocked chips now. And they could if they needed to. Remember the race to 1GHz? Remember the failed attempt by Intel of the 1.13GHz PIII? That's what it looks like when a process technology tops out. They aren't pushing C2D's at all right now because they don't need to.

And AMD does NOT know what Intel's yields are. Yields are closely guarded secrets. AMD does not how well C2D's overlcock. And that is a general indication of yeilds but they don't know how many parts are binned at each speed grade. The fact that just about every C2D overlcocks to nearly 3.5GHz shows they have headroom. And I'm sure AMD is not comforted by that fact.

Finally please keep in mind that there is a price limit on what buyers will pay for mass market chips. History has shown that to be around $1000. Why would Intel release a faster chip in place of the current top of the line 2.93GHz part and simple price all the chips below it at a lower price? That would be stupid my friend.

No, they will hold off a while and release faster part after there is either some pressure from AMD or sales begin to slow. If they release faster parts they know lots of people who like to have the fastest will upgrade. It's really just simple marketing.

And yes, if Amd could somehow release a chip a tiny bit faster than the 2.93GHz C2D part Intel would "magically" release a slightly faster part. Happens all the time. Corvette has 350HP one model year, then 400 the next, etc.. in reponse to another competitor. Yes, they "magically" pull out another 50HP when they need to. Of course there is a limit, but the current C2D process is not at that limit. Do you not think yields of the C2D are better now than 6 months ago? Of course they are but there is no market force driving them to release faster parts.

Please, before you respond with ill advised adjectives like moronic and absurd reflect upon this a bit. You are not comprehending the market forces at work here.

 

[HurleyBird]

Originally posted by: Hulk

Despite your implying that I am "moronic and absurd" I will repond to your post in the name of having a lively discussion. But I will not insult you personally. I respect your opinion. I just have a different one.

I never claimed you were moronic. Read the quote. I said it would be moronic of Intel to be underclocking their entire C2D range this entire time.

Originally posted by: Hulk
Do you follow the stock market? The reality of the current situation of a company like Intel or AMD generally lags the market by a few quarters. It takes a while for sales to turn into reported profits. Intel only gained the upper hand around Q2/Q3 of 2006. We're only 2 quarters ahead of that now.

Have you NOT noticed all of AMD's chip prices plummeting in the last few months? Do you think they just decided to "have a sale?" No, Intel is hurting them and Intel knew it would only take a range of 2.93 to 1.83GHz to do it. If they needed it to be 3.2GHz to 2.13GHz I'm sure they could have managed that since very few 3.2's in comparision to the lower end chips need be available.

I never said that AMD isn't also hurting. In fact, I haven't really commented on AMD's finances since they aren't relevant to this thread. However, in regards to Intel they are still hurting and losing market share and margins. C2Ds currently blow away AMDs offerings except on the lower end. If Intel was under clocking the entire range they would be acting stupid, as they could also blow away AMDs low end if the entire range was clocked a few hundred MHz faster. This is the last time I repeat this.

Originally posted by: Hulk
And AMD does NOT know what Intel's yields are. Yields are closely guarded secrets. AMD does not how well C2D's overlcock. And that is a general indication of yeilds but they don't know how many parts are binned at each speed grade. The fact that just about every C2D overlcocks to nearly 3.5GHz shows they have headroom. And I'm sure AMD is not comforted by that fact.

Never claimed AMD knew yields, just that they have a better picture of what speeds C2D can ramp to than you do. With the 3.5GHz headroom claim, we've been over this. Bulk silicon needs more headroom than SOI. After 1.13GHz P3 Intel likes to play it safe. They might be able to release faster C2Ds, but safety goes down when they do.

Originally posted by: Hulk
Finally please keep in mind that there is a price limit on what buyers will pay for mass market chips. History has shown that to be around $1000. Why would Intel release a faster chip in place of the current top of the line 2.93GHz part and simple price all the chips below it at a lower price? That would be stupid my friend.

Top end is always $1000, but Intel if Intel could launch a faster C2D they could lower the price of the 2.93GHz but keep all other prices stable, which would increase ASPs due to more high end chips being sold. Or they could stop making the lowest end C2D and bump every other processor down in price, maintaining the current price structure. That would put more pressure on AMD, especially on the processors that are able to compete price/performance wise with the C2Ds.

Originally posted by: Hulk
And yes, if Amd could somehow release a chip a tiny bit faster than the 2.93GHz C2D part Intel would "magically" release a slightly faster part. Happens all the time. Corvette has 350HP one model year, then 400 the next, etc.. in reponse to another competitor. Yes, they "magically" pull out another 50HP when they need to. Of course there is a limit, but the current C2D process is not at that limit. Do you not think yields of the C2D are better now than 6 months ago? Of course they are but there is no market force driving them to release faster parts.

If they feel the pressure, I don't doubt that they could if they felt it was safe to do so. However, after 1.13GHz P3, Intel won't release parts that they don't feel are safe. That's the reason we didn't see a 4GHz P4, even though the absence of one was a huge loss of face for Intel.

Originally posted by: Viditor
AMD did exactly the same thing for the C2D release, though they had less headroom to play with.

That's a pretty crummy (and false) example.

How is it false (or crummy)? Both Intel and AMD have done this since 1999...[/quote]

It's crummy because a 200MHz speed bump just doesn't count.

 

[Viditor]

Originally posted by: HurleyBird

I never said that AMD isn't also hurting. In fact, I haven't really commented on AMD's finances since they aren't relevant to this thread. However, in regards to Intel they are still hurting and losing market share and margins. C2Ds currently blow away AMDs offerings except on the lower end. If Intel was under clocking the entire range they would be acting stupid, as they could also blow away AMDs low end if the entire range was clocked a few hundred MHz faster. This is the last time I repeat this.

Huh? AMD is gaining HUGE amounts of marketshare...in fact they are currently at the highest marketshare in their history.
It's understandable that marketing of chips is heavily skewed on an enthusiast board...
You have to realize that their are 3 basic determinations for sales (if all things like branding and advertising are equal).

1. Performance - This is only important to a minority of customers (and not even the highest end customers). This is the only area where Intel might increase their margins by adding higher clocked chips to their lineup. Pure performance is second to

2. Price - For the majority of customers (especially business), even an Athlon XP or old Celeron would be just fine... These people do small spreadsheets, e-mail, and surfing the web...and that's all! They would buy a Sempron over a C2D E6600 if the Sempron was $50 less!
And remember that we're talking about system cost, not chip cost. Try shopping Dell, HP, etc... and check the relative cost of systems. The cheapest one represents over half the systems sold. If Intel added higher clocked SKUs to the line, the bottom end would necessarily need to drop down a bit. If they dropped it from an ASP of $89 to $84, they would lose 4.5% off the margin of their largest selling line. Similarly but not the same is

3. TCO - (Total Cost of Ownership) This is where the highest end customers play. What's important to them is power/performance, RAS, upgradeability, and service. This is by far the highest margin area. If Intel sold higher clocks here, then performance would go up but power/performance would drop. It would also increase TCO because increased heat means increased cooling. That is why the highest end server chips tend to be very moderately clocked...

One last (but probably the most important) point is that adding SKUs is expensive! The more different parts you have, the harder it is to choose what's going to get made (i.e. how many of each bin). If you choose wrong, you end up with excess inventory in that SKU, and that is VERY expensive!
It takes about 3 months to go from raw wafer to chip, so Intel has to make a lot of complex determinations about the future every day. It's even harder for them than it is for AMD because AMD uses a system called APM, which allows them to change even an individual chip in the middle of wafer production. The bottom line is: The more SKUs you have, the more expensive your production costs are.

The last point is that it's a general rule of thumb that you use slightly different manufacturing for higher bins. Process A gets you higher yields but lower bins, process B does the opposite.
The smartest choice is to be as close to process A as you can while still dominating performance...keeps the cost down.

Originally posted by: Viditor
AMD did exactly the same thing for the C2D release, though they had less headroom to play with.

That's a pretty crummy (and false) example.

How is it false (or crummy)? Both Intel and AMD have done this since 1999...

It's crummy because a 200MHz speed bump just doesn't count.[/quote]

It's an 8.5% increase (followed by another one in Dec), but as I said...they had less headroom.

 

[BigDH01]

I think Intel has some headroom right now. However, there is no need for them to release a faster chip. Why release a faster chip when they are already defeating AMD?

Think about it like this... If Intel released the fastest possible C2D that they can produce on this process right now, every enthusiast that could afford it would buy it. I'm also assuming this price is ~$1000. Ok, so Intel would currently be crushing AMD at this price range. So the enthusiast buys the Extreme now and waits for K8L to decide which chip is faster (this enthusiast always has to have the fastest). When K8L is released, either this enthusiast has or doesn't have the fastest chip depending on the speed of the K8L. If K8L is faster, this person spends money on the AMD chip and switches. If the C2D Extreme is still the fastest, then the person doesn't need to upgrade. You see, Intel only made 1k on this deal.

Now, let's say Intel releases a slower clocked C2D Extreme first, like they have. The enthusiast will buy this chip because it is still the fastest on the market. Intel waits and knows that they can release a faster chip to compete with K8L. So when the K8L is released, Intel eats up the rest of their headroom and releases the fastest C2D Extreme that they can. If it is faster than the K8L, then the enthusiast will purchase it (again ~1k). So now Intel has made 2k from the enthusiast.

Let's take another visual example. Let's say you pick out 4 chips for your desktop enthusiast lineup. These chips come out of the C2D line and the C2Ds range and, based on yield, you know you can release chips ranging all the way up to 3.6GHz. However, you know only 3 GHz is needed to completely dominate the competition. It makes complete sense to release 4 chips ranging from 3GHz down to whatever is needed to compete. This way the 3GHz chip can be your king and earn your high margin. If you draw this out on a number line, it makes sense to only shift these 4 chips up in clockspeed as slow as possible to maximize the margin on each individual speed grade and maximize profit. So instead of 3.6GHz being the only high margin earner, every grade from 3GHz-3.6GHz will, at one point, be a high margin earner. In this way, you get the most possible value out of your product while still pleasing the consumer.

 

[Hulk]

Originally posted by: BigDH01
I think Intel has some headroom right now. However, there is no need for them to release a faster chip. Why release a faster chip when they are already defeating AMD?

Think about it like this... If Intel released the fastest possible C2D that they can produce on this process right now, every enthusiast that could afford it would buy it. I'm also assuming this price is ~$1000. Ok, so Intel would currently be crushing AMD at this price range. So the enthusiast buys the Extreme now and waits for K8L to decide which chip is faster (this enthusiast always has to have the fastest). When K8L is released, either this enthusiast has or doesn't have the fastest chip depending on the speed of the K8L. If K8L is faster, this person spends money on the AMD chip and switches. If the C2D Extreme is still the fastest, then the person doesn't need to upgrade. You see, Intel only made 1k on this deal.

Now, let's say Intel releases a slower clocked C2D Extreme first, like they have. The enthusiast will buy this chip because it is still the fastest on the market. Intel waits and knows that they can release a faster chip to compete with K8L. So when the K8L is released, Intel eats up the rest of their headroom and releases the fastest C2D Extreme that they can. If it is faster than the K8L, then the enthusiast will purchase it (again ~1k). So now Intel has made 2k from the enthusiast.

Let's take another visual example. Let's say you pick out 4 chips for your desktop enthusiast lineup. These chips come out of the C2D line and the C2Ds range and, based on yield, you know you can release chips ranging all the way up to 3.6GHz. However, you know only 3 GHz is needed to completely dominate the competition. It makes complete sense to release 4 chips ranging from 3GHz down to whatever is needed to compete. This way the 3GHz chip can be your king and earn your high margin. If you draw this out on a number line, it makes sense to only shift these 4 chips up in clockspeed as slow as possible to maximize the margin on each individual speed grade and maximize profit. So instead of 3.6GHz being the only high margin earner, every grade from 3GHz-3.6GHz will, at one point, be a high margin earner. In this way, you get the most possible value out of your product while still pleasing the consumer.

Exactly. Performance-wise there is little pressure on Intel from AMD at this point in time. Intel is smart to take a wait and see attitude.

I stand by my initial point that if they wanted to Intel could release a 3.2GHz part right not. So yes they are underclocking their chips. I know my E6400 which is running in my computer at 1.35V (cpuz) at a 50% overclock was definitely underclocked by Intel's rating. There is quite a bit of headroom in the C2D line. And we don't know if Intel is binning the really good parts to be released at 3.2 or even 3.46GHz at some time in the future.

Then again if the 45nm initial yields are really excellent they may decide to have that stepping take off where Conroe ends to keeps the thermals in line.

 

[HurleyBird]

Originally posted by: Viditor

Originally posted by: HurleyBird

I never said that AMD isn't also hurting. In fact, I haven't really commented on AMD's finances since they aren't relevant to this thread. However, in regards to Intel they are still hurting and losing market share and margins. C2Ds currently blow away AMDs offerings except on the lower end. If Intel was under clocking the entire range they would be acting stupid, as they could also blow away AMDs low end if the entire range was clocked a few hundred MHz faster. This is the last time I repeat this.

Huh? AMD is gaining HUGE amounts of marketshare...in fact they are currently at the highest marketshare in their history.

You misunderstood my post, which is really my fault for wording a weak sentence. When I said "However, in regards to Intel they are still hurting and losing market share and margins." the 'they' I was referring to was Intel, not AMD.

In response to some of the other replies. You guys may or may not be right that Intel is sitting on higher clocking parts, but I still think it would be stupid of Intel to sit on those parts because some A64s are still competitive price/performance wise against the lower clocked C2Ds. However, since this discussion is going in circles, I'm going to leave the discussion were it's at.

 

[Viditor]

Originally posted by: HurleyBird

Originally posted by: Viditor

Originally posted by: HurleyBird

I never said that AMD isn't also hurting. In fact, I haven't really commented on AMD's finances since they aren't relevant to this thread. However, in regards to Intel they are still hurting and losing market share and margins. C2Ds currently blow away AMDs offerings except on the lower end. If Intel was under clocking the entire range they would be acting stupid, as they could also blow away AMDs low end if the entire range was clocked a few hundred MHz faster. This is the last time I repeat this.

Huh? AMD is gaining HUGE amounts of marketshare...in fact they are currently at the highest marketshare in their history.

You misunderstood my post, which is really my fault for wording a weak sentence. When I said "However, in regards to Intel they are still hurting and losing market share and margins." the 'they' I was referring to was Intel, not AMD.

In response to some of the other replies. You guys may or may not be right that Intel is sitting on higher clocking parts, but I still think it would be stupid of Intel to sit on those parts because some A64s are still competitive price/performance wise against the lower clocked C2Ds. However, since this discussion is going in circles, I'm going to leave the discussion were it's at.

Fair enough, mate...
I'll leave too, with one final thought.
Since Intel is only capable (this quarter) of making 40% of their chips C2Ds, that means they need to keep market conditions conducive to Netburst sales still...
If they made C2Ds attractive across all speed grades, then how would they sell Netburst?
If they weren't selling Netburst, they couldn't meet their sales demand...

As further evidence of this, note how late they are bringing out the C2D Celerons. It's not because they couldn't do it before, it's because they won't have the manufacturing capability to replace Netburst until then (note that Netburst is being EOLed around the same time as the C2D Celeron introduction).

 

[coldpower27]

Originally posted by: HurleyBird

Originally posted by: Viditor
Increasing the clockspeed on C2D wouldn't help Intel's margins in the least. Keep in mind that C2D is only going to be 40% of their available chips by the END of this quarter.

Well, its more of a choice between increasing margins (selling some more pricier, higher clocked stuff while keeping prices the same for the rest of the line) or putting pressure on AMD (increase the clock speed around the board, but keep price the same so AMD no longer is competitive with the low end C2Ds), either way, increasing clock speeds would help Intel out now (as well as many moths earlier) and it would be pretty stupid of Intel not to release higher clocking parts if they actually could.

Originally posted by: Viditor
At the current clockspeeds, C2D is clocked at exactly what it needs to be for competition with AMD, and as AMD releases their OWN next gen design, Intel will still have a few "clubs in the bag" to help them be competitive.

AMD will release their next gen chips so that the highest end chips will struggle to find a 200MHz over clock in stock conditions. They can do this because they use SOI (and Intel can't do this because they use bulk silicon --if Intel used SOI you would see the same behavior from them). As I said earlier, it wouldn't matter if C2D's were 400 MHz slower or 400MHz faster, Barcelona will still launch at the same speed and performance. Also, you aren't thinking very hard if you don't realize that AMD knows what current C2Ds can over clock to.

Originally posted by: Viditor
AMD did exactly the same thing for the C2D release, though they had less headroom to play with.

That's a pretty crummy (and false) example.

SOI has absolutely nothing to do with it at all, as to how high AMD will clock it's chips to, Intel itself has sold chips that have pretty poor overclocking headroom and were basically at the absolute limits of the process nodes they were on, and we know Intel doesn't currently use SOI.

Remember SKU's of Pentium 3 1.0GHZ CuMine, as well as Pentium 4 Northwood C 3.4GHZ were at basically the absolute limits of how far the process will safely go.

Increasing clockspeeds now would mean comprimising the remainder of the NetBurst inventory Intel needs to get rid of, there is a reason Core 2 Duo still remains a $163 USD SKU and higher part until April 22nd where it falls to a $113 USD part which is still out of low end territory.

The marketshare Intel has lost both in the notebook and desktop arenas was low end cheap marketshare where Intel isn't competing with Core 2 Duo technology, were talking Athlon 64's, Low End Athlon 64x2, and Low End Turion's and Turion X2's.

While your mentioning the marketshare Intel lost, you haven't talked about the marketshare they gained in the lucrative server arena where margins are considerably higher.

As have been re-iterated by many others there are many good reasons to keep some speed in reserve, until AMD shows their hand with their K10 derivatives.

 

[coldpower27]

Originally posted by: Viditor
There are several points here...
1. AMD will release at whatever speed it takes to surpass C2D in both performance and power/performance...and not much more (for the same reason Intel doesn't).
2. Barcelona/Stars chips will be vastly different to current Opteron/X2 chips (it's a whole new architecture), and that includes their overclockability. SOI is only a very small part of the story...they have a whole new power section, (hopefully) a new DSL process, and are second generation 65nm (remember that C2D is also second generation 65nm).
3. Raising clockspeed also raises the TDP

Assuming AMD can of course, they will probably release them as high as they can comfortably yield them in the price brackets they set. Whether or not they will beat the available Intel processors we will have to see.