Without AMD (essentially an Intel Monopoly) What would CPUs cost?

[Dadofamunky]

To answer the original question: more than they do now!

 

[cbn]

Did you check the difference in margins of the two companies?

Intel - 67% vs AMD - 45%

How much of that profit margin is from AMD's GPU business?

 

[busydude]

How much of that profit margin is from AMD's GPU business?

I am not sure about that, revenues from GPU business were $440 m which is 26.67% of the total sales and with CPU sales contributing to about 3/4th of the total revenues($1.21 bn) its not easy to speculate on the margins.

 

[cbn]

the situation is improving for the better and we can hope they offer good products in coming months(fingers crossed).

I am hoping AMD could eventually get a design win with Apple. Bobcat or Bulldozer cores with "Optimus like" power management to the Fusion GPGPU would really be nice.

Maybe this will happen as OPEN CL and fusion matures?

As far as Windows and Intel go, it sure seems like the upcoming atom chips (Oak Trail and Cedar trail) could make some nice bare minimum systems/netbooks. According to some reading I have done they will even be able to decode 1080p video natively.

 

[Scali]

I am hoping AMD could eventually get a design win with Apple. Bobcat or Bulldozer cores with "Optimus like" power management to the Fusion GPGPU would really be nice.

Maybe this will happen as OPEN CL and fusion matures?

AMD hasn't made a very good impression so far.
OpenCL was an Apple initiatiave, as you might know.
So far, OpenCL support on MacOS has been very poor from AMD, much more mature with nVidia.
OpenCL support on Windows or linux is still non-existent for end-users, as you're probably aware...

If AMD wants Fusion to be a success, and OpenCL to be a driving force, they need to step up in the driver department.

 

[classy]

There are so many variables in pricing this is essentially impossible to answer. AMD has been in the game so long I don't know if anyone could honestly make an assessment of what cpus would cost if AMD wasn't around.

 

[Scali]

There are so many variables in pricing this is essentially impossible to answer.

I think it's not the pricing itself that's variable.
Especially Intel seems to have a pretty rigid set of price points. Often they keep the exact same prices, they just bump all CPU models down a notch, and insert a new one at the top.
Or they replace a whole series at once.
It's always the same exponential pricing curve, with usually the exact same price points.

I don't think it would be very wise to change this scheme, the shape of the curve and the endpoint at about $1000 seems to fit what the consumers are willing to pay for a CPU very well.
So I think the only thing that would change is what CPUs Intel would position where, and when. They may want to drop some of the ultra-cheap options, if there is nobody to compete with anyway.
And they may wait a bit longer with bumping down CPUs to lower price points, and introducing replacements.

Although I'm not too sure how much we would notice of that... For example, take the introduction of Nehalem. Intel didn't have to introduce Nehalem, as they were competing with AMD well enough with C2Q. With the introduction of Nehalem, they only succeeded in devaluating their own C2Q line. Especially the i7 920, which had considerably better price/performance than C2Q. They could have easily positioned it above C2Q in price.
And Gulftown... they didn't have to introduce that one at all. The performance margin was large enough with just 4 cores. There was no direct competition with the i7 range.

So why did they do it? Probably because they figured that this was a better way to make money.
The lower pricings on the i7 puts them lower on the curve, which effecively means higher sales volumes. And the new faster i7s give the people who already owned the previous model another incentive to upgrade.

There's a lot of economic theories on profit maximizing, and just raising prices through the roof isn't one of them

 

[cbn]

AMD hasn't made a very good impression so far.
OpenCL was an Apple initiatiave, as you might know.
So far, OpenCL support on MacOS has been very poor from AMD, much more mature with nVidia.
OpenCL support on Windows or linux is still non-existent for end-users, as you're probably aware...

If AMD wants Fusion to be a success, and OpenCL to be a driving force, they need to step up in the driver department.

http://www.khronos.org/news/archives/

Well according to the the Khronos's own July 22nd News release entitled "If liked assembler, you'll love Open CL" it doesn't sound like Open CL is easy to work with.

In any event It would be great to see Apple with ATI or Nvidia do something really interesting. Maybe Educational and Graphical simulations outside of the role CUDA normally fills?

 

[ModestGamer]

http://www.khronos.org/news/archives/

Well according to the the Khronos's own July 22nd News release entitled "If liked assembler, you'll love Open CL" it doesn't sound like Open CL is easy to work with.

In any event It would be great to see Apple with ATI or Nvidia do something really interesting. Maybe Educational and Graphical simulations outside of the role CUDA normally fills?

assembler makes sense to me. I always found higher level languages convoluted and confusing. But I'm fucking wierd.

 

[Scali]

Well according to the the Khronos's own July 22nd News release entitled "If liked assembler, you'll love Open CL" it doesn't sound like Open CL is easy to work with.

Cuda is easier to use, that's for sure.
DirectCompute is probably even more terse than OpenCL though.
Personally I think DirectCompute and OpenCL are closer to HLSL/GLSL shader programming than assembler though.
Cuda is very close to regular C/C++, especially since it integrates almost seamlessly into your application.
Cuda code is compiled and linked into your program, and you basically just call your Cuda routine like any other C/C++ function.

With the others you have to load your sourcecode and compile it at runtime, then feed it to the GPU, very similar to how shaders work.

 

[stevech]

well, the OP hasn't seen more than one short/succinct answer to the question, paraphrased as: If AMD was absent, would Intel's monopoly, Moore's law aside, negatively affect PC buyers?

 

[Idontcare]

well, the OP hasn't seen more than one short/succinct answer to the question, paraphrased as: If AMD was absent, would Intel's monopoly, Moore's law aside, negatively affect PC buyers?

I think the OP has had more than one satisfactory answer. The decision makers at Intel want to make money, and for that to happen Intel has to continue to make money, and large quantities of it.

Large quantities of money require large quantities of customers, and customers come from new markets plus upgrades from existing ones.

In both cases the answer to selling more product is not to raise prices on existing product or to forestall the development of new products.

The decision makers at Intel will ensure new products come, just as the decision makers at Microsoft do. As do the multitude of employees of Intel. No one aspires to have their career path end at being a minion of a monopoly, hardly the sort of worklife that provides one with genuine intellectual satisfaction. Those folks will find themselves taking employment elsewhere, even if it means a reduction in financial compensation with the remainder of payment delivered in terms of cerebral stimulation.

 

[cbn]

Cuda is easier to use, that's for sure.
DirectCompute is probably even more terse than OpenCL though.
Personally I think DirectCompute and OpenCL are closer to HLSL/GLSL shader programming than assembler though.
Cuda is very close to regular C/C++, especially since it integrates almost seamlessly into your application.
Cuda code is compiled and linked into your program, and you basically just call your Cuda routine like any other C/C++ function.

With the others you have to load your sourcecode and compile it at runtime, then feed it to the GPU, very similar to how shaders work.

Hi Scali,

I am not a programmer but has CUDA always featured higher level language "qualities" like C++? Or is that a feature that got added in later development kits?

According to Wikipedia, CUDA "provides both a low level API and a higher level API".

 

[cbn]

AMD hasn't made a very good impression so far.
OpenCL was an Apple initiatiave, as you might know.
So far, OpenCL support on MacOS has been very poor from AMD, much more mature with nVidia.
OpenCL support on Windows or linux is still non-existent for end-users, as you're probably aware...

If AMD wants Fusion to be a success, and OpenCL to be a driving force, they need to step up in the driver department.

I know you are referring to AMD in this post, but how is Intel's support for Open CL coming along?

Currently I don't see Intel's name listed under the "implementations" section of this Wikipedia source ---->http://en.wikipedia.org/wiki/OpenCL

Or does Intel have no real interest in OPEN CL and would rather persue "many core" x86 design instead?

 

[aphorism]

Hi Scali,

I am not a programmer but has CUDA always featured higher level language "qualities" like C++? Or is that a feature that got added in later development kits?

According to Wikipedia, CUDA "provides both a low level API and a higher level API".

CUDA has always been based off of good old C. C++ is basically an augmented version of C with classes. i think what you are referring to is PTX which is a pseudo-assembly language. programs written in C for CUDA are compiled into PTX and then the compiler in the driver generates the code that runs natively on the gpu.

 

[Scali]

I am not a programmer but has CUDA always featured higher level language "qualities" like C++? Or is that a feature that got added in later development kits?

C++ is a very new feature, it was introduced with the Cuda update released with Fermi.

According to Wikipedia, CUDA "provides both a low level API and a higher level API".

Yes, the low level API works quite similar to OpenCL: you have to manually compile your code into a runtime object, and then use API calls to run it.
The higher level API is what I mentioned, with the Cuda code almost seamlessly integrating into your regular C/C++ code.

I know you are referring to AMD in this post, but how is Intel's support for Open CL coming along?

I have no idea really.
Intel officially supports the OpenCL standard, they've mentioned it a few times in press releases as well.
Perhaps they were/are planning to release OpenCL support once Larrabee is on the market. I doubt that they have much interest in regular x86 support, as there's not much use. All software already supports x86 natively, and that will always be more efficient than OpenCL code. Enabling OpenCL support for their x86 processors would only open the door to more GPGPU support in applications, which is not in Intel's interest obviously.
And Intel's current GPUs are probably slower at OpenCL than their CPUs, so not much point in supporting those either (although I think their GPUs may be capable of it, technically. They have full DX10 support, and certain operations such as triangle clipping are also done on their shaders, rather than with dedicated hardware).

So I don't know the status of their OpenCL... perhaps they have developed a CPU runtime, but just don't see a reason to release it... and perhaps they have not invested time in it at all. My guess would be the former though.

AMD's CPU OpenCL runtime works fine on Intel CPUs though, so end-users can still use OpenCL on Intel processors if they want.

 

[Obsoleet]

well, the OP hasn't seen more than one short/succinct answer to the question, paraphrased as: If AMD was absent, would Intel's monopoly, Moore's law aside, negatively affect PC buyers?

I answered it shortly and succinctly. I'm sure someone else has as well.
http://forums.anandtech.com/showpost.php?p=30175053&postcount=132

 

[bullbert]

I answered it shortly and succinctly. I'm sure someone else has as well.
http://forums.anandtech.com/showpost.php?p=30175053&postcount=132

I say you are wrong. The Feds never broke up Microsoft. I doubt they would ever have broken up Intel, either.


FYI: late 1980s, a top bin Intel 386 was $3700 and a top bin Intel 387 was also $3700. Compare 1980's $7400 to today's $990. Thank you AMD (and Cyrix and Motorola and etc).

 

[jvroig]

FYI: late 1980s, a top bin Intel 386 was $3700 and a top bin Intel 387 was also $3700. Compare 1980's $7400 to today's $990. Thank you AMD (and Cyrix and Motorola and etc).

I believe in the need for competition, and while I am sure prices won't be exactly as they are now for all market segments if there were completely no competition, comparing the prices from 1980 to today doesn't take into account other factors other than competition. For one thing, back then, the market wasn't that big (right now, there's a PC in every desk in every office, and in almost every home; back then, no such luck), so the volume of shipments weren't that big, and that affected price as well. As the volume of shipments went up exponentially, the price also adjusted. This isn't a reverse supply/demand curve that you learn in Economics 101. This is simply the size of the market increasing exponentially. If the prices remained prohibitively high, the potential buyers in several market segments would have been left untapped (what economists call "opportunity losses"). Price adjustment and market segmentation that happened (various consumer lines, several levels of server lines) were just ways to capture consumer surplus, and that would have happened anyway even without AMD/Cyrix/Motorola.

Again, I'm not saying competition in this industry is unnecessary. I've already said early in this thread that I believe strongly for the need for balanced competition. All I'm saying now is that quoting prices from 30 years ago is not a justification for the need for competition; those were very different market environments from today.

 

[Scali]

Not to mention that back then the differences between low-end and high-end were very different.
In the late 80s, 8088-based computers were still being sold, they were the low-end, and very affordable, despite a lack of competition from other vendors.
286 and 386SX systems formed the mainstream, 386DX was aimed at the workstation/server market. If you compare that to the prices of high-end Xeons or Opterons(!) today, the prices are suddenly a lot less ridiculous.

 

[bamacre]

Without AMD (essentially an Intel Monopoly) What would CPUs cost? $300?
look at the obscene prices Micro$oft is forcing on us with their essential monopoly, for Win 7, a face-lift for the same ole OS.

Therefore, I have bought only AMD for many years, to do my little part to avoid $300 CPUs.

Without AMD, another cpu maker would enter the market. Intel and AMD are certainly not the only makers of microprocessors.

When AMD held the crown, their most popular cpu (the X2 3800+) was $300 and stayed at that price until Intel released their Core 2 processors.

AMD and Intel care about profit, neither one is in the game because they care. They care about you because they care about profit. Thank capitalism, not AMD.

 

[richierich1212]

Without AMD, another cpu maker would enter the market. Intel and AMD are certainly not the only makers of microprocessors.

Serious? AMD has hardly any marketshare now, and if another chip maker were to enter the market now this is the time to do so. But nope, noone is stepping up.

 

[bamacre]

Serious? AMD has hardly any marketshare now, and if another chip maker were to enter the market now this is the time to do so. But nope, noone is stepping up.

A quick Google search shows AMD has a 10-15% market share, that's pretty significant. You're right, no one else is stepping up, but AMD isn't gone either.

 

[Scali]

A quick Google search shows AMD has a 10-15% market share, that's pretty significant. You're right, no one else is stepping up, but AMD isn't gone either.

AMD is still the second-biggest CPU manufacturer in the world.
It's just that Intel is so much larger still.

 

[Phynaz]

AMD is still the second-biggest X86 CPU manufacturer in the world.
It's just that Intel is so much larger still.

Yes, I sweat the small things.