Something to consider in your Intel vs AMD discussions

[zsdersw]

Much is often made of the fact that Intel is such a large company and spends so very much more than AMD (a relatively small company) on research and development, and that even with such relatively paltry expenditures on R&D AMD is able to generally remain competitive. This is often used to shore up support for AMD while trying to deflate the egos of Intel fanatics.

The fundamental truth that is missing is that no amount (low or high) of R&D expense can overcome the basic material and manufacturing limitations of the product/technology that's being improved. The manufacturing methods/processes employed by Intel and GloFo have very specific capabilities, associated costs, and constraints that set the parameters within which microprocessors must be designed to fit. Intel and AMD are also basing their improvements upon specific previous designs and the demands of the markets in which future processors will be sold.

Company size and R&D expenditures are not realistic ways to set performance expectations.

 

[jiffylube1024]

Much is often made of the fact that Intel is such a large company and spends so very much more than AMD (a relatively small company) on research and development, and that even with such relatively paltry expenditures on R&D AMD is able to generally remain competitive. This is often used to shore up support for AMD while trying to deflate the egos of Intel fanatics.

The fundamental truth that is missing is that no amount (low or high) of R&D expense can overcome the basic material and manufacturing limitations of the product/technology that's being improved. The manufacturing methods/processes employed by Intel and GloFo have very specific capabilities, associated costs, and constraints that set the parameters within which microprocessors must be designed to fit. Intel and AMD are also basing their improvements upon specific previous designs and the demands of the markets in which future processors will be sold.

Company size and R&D expenditures are not realistic ways to set performance expectations.

There's one point that this chain of reasoning doesn't address - while AMD and Intel may be constrained by their previous products and design philosophies, what Intel's market dominance allows them to do is stay one manufacturing node ahead of AMD, while AMD is now almost completely at the whims of third parties such as Global Foundries and TSMC.

So, while Intel's Core i series may be fundamentally superior to AMD's Phenom II architecture, Intel has the added advantage of being able to use the 32nm manufacturing node while AMD is stuck on 45nm for awhile.

This was even more of an issue last node - Intel jumped to 45nm early with Socket 775, while AMD was stuck on 65nm for a very long time.

 

[aigomorla]

Much is often made of the fact that Intel is such a large company and spends so very much more than AMD (a relatively small company) on research and development, and that even with such relatively paltry expenditures on R&D AMD is able to generally remain competitive. This is often used to shore up support for AMD while trying to deflate the egos of Intel fanatics.

The fundamental truth that is missing is that no amount (low or high) of R&D expense can overcome the basic material and manufacturing limitations of the product/technology that's being improved. The manufacturing methods/processes employed by Intel and GloFo have very specific capabilities, associated costs, and constraints that set the parameters within which microprocessors must be designed to fit. Intel and AMD are also basing their improvements upon specific previous designs and the demands of the markets in which future processors will be sold.

Company size and R&D expenditures are not realistic ways to set performance expectations.

Uhhh Welcome to Corporate America?

I have very little fan base when im looking for the most solid as well as best performing platform available.

And its not intel's fault that they are holding onto the crown.

And the Fanism your talking about isnt without numbers.
Its those numbers which generated Fanism.

And if your pulling an unfair moment on intel, can i remind you what boat they sat on during the P4 days and what happened to the guys who were tooting intel's horn then?

Although i wont draw a clear line on who i fully support, i will go with who has the crown.

Right now thats intel every way you look at it.
And if people need to get unhappy at AMD to make AMD change, then so be it.
If they disappear, it means something went wrong, and like Cyrix, there fate is just that.

But i promise you, if AMD does fall, someone will probably buy it, and try to transform the company once again, to try to take the crown.
But as things go right now, a change is needed, and its not going to happen with us just trying to be FAIR in the whole situation.

Your paying for it, why not get whats best available with your own hard cash?

 

[mnewsham]

Uhhh Welcome to Corporate America?

I have very little fan base when im looking for the most solid as well as best performing platform available.

And its not intel's fault that they are holding onto the crown.

And the Fanism your talking about isnt without numbers.
Its those numbers which generated Fanism.

And if your pulling an unfair moment on intel, can i remind you what boat they sat on during the P4 days and what happened to the guys who were tooting intel's horn then?

Although i wont draw a clear line on who i fully support, i will go with who has the crown.

Right now thats intel every way you look at it.
And if people need to get unhappy at AMD to make AMD change, then so be it.
If they disappear, it means something went wrong, and like Cyrix, there fate is just that.

But i promise you, if AMD does fall, someone will probably buy it, and try to transform the company once again, to try to take the crown.
But as things go right now, a change is needed, and its not going to happen with us just trying to be FAIR in the whole situation.

Your paying for it, why not get whats best available with your own hard cash?

Disregard price and Intel is the clear and obvious choice across the board.

 

[bryanW1995]

lots of recent speculation that the real reason dirk left is that the board wants to make the company look better to a potential buyer.

What is AMD really worth? Any chance that it could get cheap enough for NV to take it over? What would we call the new company, Daamd NV?

 

[Martimus]

There is more than one way to skin a cat, and the more ways you test out, the better chance you will find one that is faster/more efficient/cheaper/whatever other attribute you are looking for. Also, finding that method first allows you to patent it; meaning you can force your competitors to pay you for that method, or make them find a completely different way to skin that cat.

So of course R&D funding makes a difference. Thinking otherwise is asinine. However I would expect the gains from additional resources to be nowhere near linear.

 

[SlitheryDee]

I don't care which company is on top. So long as there are two or more of them duking it out I win every time.

I don't see how anything in the OP disproves the benefits of increased R&D expenditure though. Obviously there are limitations in every aspect of processor manufacturing. R&D is what allows them to do the most with what they have. Take these limitations away and both companys could make infinitely powered processors that run on rainbows and fairy dust. If I had a pogo stick tall enough I could also jump to the moon.

Larger company size implies proportionally larger departments compared to a smaller company. If each company uses 10% of their revenue on R&D, then the company with more revenue has more R&D. That means more and better paid people using more and better equipment. I can't think of any other way to go about achieving better results.

Of course a company could choose to devote much less revenue to R&D, which is why you can't put total faith in company size as an indicator of money spent on R&D. That's irrelevant to this topic because the op states "Company size and R&D expenditures are not realistic ways to set performance expectations". If we already know that one company is spending more on R&D regardless of that company's relative size, then our expectations of the performance of it's product very well should be high, or else the company is not getting it's money's worth.

 

[smakme7757]

Your paying for it, why not get whats best available with your own hard cash?

+1

I always abide by this rule. I buy the fastest i can afford (which is usually anything exept an extreme processor). I do not have any favoured company and i just buy the best processor and single GPU ard on the market at the time of buying my new system.

On topic though: If i was a betting man i'd bet on the company who is most heavily inveted in it's R&D. You can't put a price on quality R&D and it's payed off for Intel the last few years!

 

[beginner99]

The fundamental truth that is missing is that no amount (low or high) of R&D expense can overcome the basic material and manufacturing limitations of the product/technology that's being improved.

But the more money have the closer you can get to the perfect solution within these constraints. I would also say the bigger the constraints the harder it is to design a good chip and hence the more important money becomes.

Analogy:

Formula 1 and Aerodynamics. There are now so many rules, that actually making any progress within those is impossible with tons of money to spend on simulations and testing in a wind channel.

 

[BoomerD]

I have NEVER owned an AMD processor...until this year when I replaced my wife's aging Dell P4 Celeron rig with a new Dell...with an Athlon XII.
No complaints about it. It handles everything she needs and more.

When I was first planning out my new build, I looked very hard at the AMD PhenomII processors...but found them wanting...especially when compared to the Intel offerings. (even pre-Sandy Bridge, Intel offered better performance for the types of things I use my computer to do.)

Would I consider an AMD processor in another build? Yes...IF it offered better performance than what Intel offered.

I don't consider myself a fanboy...I've owned both nVidia based GPU's as well as ATi based...with no real preference. (except that ATI drivers seem to be more problematic)
I'm more of a "best bang for the buck" kind of guy.

Myself, I hope AMD manages to hold on. Killing any semblance of competition for Intel could be bad for the hobby.

 

[nyker96]

that's right, down with AMD, down!! Intel rules the world!

 

[JFAMD]

There's one point that this chain of reasoning doesn't address - while AMD and Intel may be constrained by their previous products and design philosophies, what Intel's market dominance allows them to do is stay one manufacturing node ahead of AMD, while AMD is now almost completely at the whims of third parties such as Global Foundries and TSMC.

You have to be careful about thinking that the next node is going to be the silver bullet.

Each node is progressively more expensive and each node becomes progressively smaller in benefit. There is a point where, before too long, the laws of dinimishing returns come into play. I seem to recall 90nm Opterons beating 65nm Xeons in performance and price and power, so the next node is not always a golden ticket.

Disregard price and Intel is the clear and obvious choice across the board.

That is preposterous. It's like saying if you ignore her head she could be a supermodel. You have to deal with the whole thing, you don't get to pick and choose.

One could also say disregard performance and AMD is the clear and obvious choice across the board. The big difference is that less than 5% of the market is buying the top bin procs, so price has a FAR GREATER impact on purchase than performance.

I have often seen people say "this is the best performance for my budget", but have you ever seen someone choose the performance level then look at the price?

 

[mnewsham]

You have to be careful about thinking that the next node is going to be the silver bullet.

Each node is progressively more expensive and each node becomes progressively smaller in benefit. There is a point where, before too long, the laws of dinimishing returns come into play. I seem to recall 90nm Opterons beating 65nm Xeons in performance and price and power, so the next node is not always a golden ticket.




That is preposterous. It's like saying if you ignore her head she could be a supermodel. You have to deal with the whole thing, you don't get to pick and choose.

One could also say disregard performance and AMD is the clear and obvious choice across the board. The big difference is that less than 5% of the market is buying the top bin procs, so price has a FAR GREATER impact on purchase than performance.

I have often seen people say "this is the best performance for my budget", but have you ever seen someone choose the performance level then look at the price?

My next build is going to have a 4k budget, im going to go with whatever processor performs best regardless of price. Does that make me stupid? maybe, but it's my money and i can afford it.

 

[nenforcer]

Is this a thread promoting the VIA C7 / Via NANO platform?

 

[BoomerD]

Is this a thread promoting the VIA C7 / Via NANO platform?

Via's bringing back the Cyrix? ^_^

 

[86waterpumper]

but it's my money and i can afford it

I don't know that anybody here is impressed. Just because you can afford something does not mean it is wise to do so but hey it's america so feel free. The percentage of people in this day and age of 400 dollar laptops that would spend even over 1k on a desktop computer is very very small indeed. The price per performance ratio DOES have to be taken into consideration, and when you do things still look alright for AMD.
On my 4 core phenom rig, I can run prime95 or something and max all four cores out, and the thing will still run various browers, and stream vids etc. For probably 90 percent of home users we are at a place where different architectures don't matter alot besides benchmarks on quad cores @ 3.5ghz or above. More people should probably buy a ssd than any other upgrade right now. If I were to build another computer right now I'd probably go AMD again. Why? Because there is nothing I will have to do in the next two to three years that it won't do, and it's a awesome amount of processing power for the money.
Most people that are going sandy bridge are paying as much for their motherboard as I did for my mb + cpu, and they are getting nothing for their money whatsoever in alot of cases far less mb features. If that is what they want to do then fine, but it's not the best decision in every case. These days intel chipset motherboards ought to have a nice big apple symbol stamped on them because they are using the same tactics. You will get the highest mhz and performance, but along with it you will receive price gouging, limited feature sets, forced drm etc etc so it's all in what you want isn't it.

 

[jiffylube1024]

You have to be careful about thinking that the next node is going to be the silver bullet.

Each node is progressively more expensive and each node becomes progressively smaller in benefit. There is a point where, before too long, the laws of dinimishing returns come into play. I seem to recall 90nm Opterons beating 65nm Xeons in performance and price and power, so the next node is not always a golden ticket.

The next node is the silver bullet! Yes smaller nodes have problems - AMD with their Athlon XP Rev A, Intel with 90nm P4, but once those problems are solved you get a much superior chip to the last node.

The 65nm Xeons that were getting stomped were P4-based chips. Compare a 65nm Core 2 Duo or C2D-based Xeon and it will absolutely murder the 90nm Opteron.

As for this progressively smaller benefit -- hogwash! Compare the 90nm P4 to the 65nm Core2Duo to the 45nm Core 2 Duo to the 45nm Core i5/7 to the 32nm Core i5/7 and you get a chain of better and better processors.

Yes, nodes become more and more difficult to "master" (as seen with the major current leakage in the 90nm P4 Presscott revision compared to 130nm Northwood, but once problems are solved, you get more CPU per mm^2 and thus can pack in more cores or more sophisticated core and more features -- all for a lower price!

Tell AMD and Intel and Nvidia to just stick with a node and build better chips dammit! They can't - smaller nodes provide not only smaller chips per transistor but also cooler running chips. Without agressive chip shrinkage (no puns intended) we wouldn't have 1000 MHz chips in cell phones and CPU's the size of Intel's Atom that can run Windows acceptably and 6x cores in a single CPU die.

 

[mnewsham]

I don't know that anybody here is impressed. Just because you can afford something does not mean it is wise to do so but hey it's america so feel free. The percentage of people in this day and age of 400 dollar laptops that would spend even over 1k on a desktop computer is very very small indeed. The price per performance ratio DOES have to be taken into consideration, and when you do things still look alright for AMD.
On my 4 core phenom rig, I can run prime95 or something and max all four cores out, and the thing will still run various browers, and stream vids etc. For probably 90 percent of home users we are at a place where different architectures don't matter alot besides benchmarks on quad cores @ 3.5ghz or above. More people should probably buy a ssd than any other upgrade right now. If I were to build another computer right now I'd probably go AMD again. Why? Because there is nothing I will have to do in the next two to three years that it won't do, and it's a awesome amount of processing power for the money.
Most people that are going sandy bridge are paying as much for their motherboard as I did for my mb + cpu, and they are getting nothing for their money whatsoever in alot of cases far less mb features. If that is what they want to do then fine, but it's not the best decision in every case. These days intel chipset motherboards ought to have a nice big apple symbol stamped on them because they are using the same tactics. You will get the highest mhz and performance, but along with it you will receive price gouging, limited feature sets, forced drm etc etc so it's all in what you want isn't it.

That doesn't change the fact that if money is no object that Intel wins in performance, most users wont be buying the most expensive CPU out there but that doesn't matter. All I'm saying is that Intel has the best processors available. I agree that for most people price IS a factor. For me and many others who want the best performance... not so much. I agree that Intel is not the best choice for everyone at every price point but if you want the best performance regardless of price, Intel is your only option.

 

[Edgy]

Intel's annual R&D budget exceeds AMD's annual revenue TOTAL - if this fact doesn't lead to a conclusion that it makes a huge competitive difference, then I don't think any argument will.

It's not ALL about fab process node either. Compare Intel's 45nm products to AMD's 45nm. And then AMD's 65nm to Intel's 65nm during Conroe and subsequent products. I think it's safe to say most would consider Intel's to be superior technologically.

Then look how long it's taken AMD to get Fusion & Bulldozer out (err... getting or will get them out)... Also consider Intel's so called tick/tock development strategy...

Alas someone high up at AMD consciously decided they can delay their regularly scheduled R&D development toward new tech & improvements (most likely the K9) during their Athlon 64 dominance era and they were in denial even when first leaked reports of Intel's then upcoming Conroe outperforming Athlons were spreading.

I think we can infer few things from all this:
1. Intel's R&D budget allowed them to develop Conroe much quicker comparatively than AMD could with their Fusion/Bulldozer. But more R&D budget does not necessarily mean better products - it does allow bit more flexibility such as concurrent development of multiple projects and better recovery if any snafu.
2. Currently, Intel not only has more R&D budget advantage but strategically trying to be more streamlined and efficient in converting those R&D directly into improved products toward more visible ROI.
3. It is shocking (and all kudos to their R&D and an opportune period of total idiocy afflicting their competitor) that AMD even had it's brief period of technological superiority with Athlon 64.

AMD's hopefully gonna catch up relatively speaking with Fusion and Bulldozer finally. Looking at their latest roadmaps and commitment, it looks like they too will adopt a strategy similar to Intel's tick/tock cycle.

In my opinion, the largest disadvantage AMD has compared to Intel is not process tech or innovation. I think its marketing and marketing people. AMD obviously does not have the budget to compete with Intel in marketing. They really truly do need innovative and daring people in their marketing to compete. Trouble is, I think AMD marketing is more conservative than any I've ever worked with ... including Intel.

Knowing various people from both Intel and AMD marketing "machines" and their philosophies, I wouldn't hold my breath for AMD's marketing to make any significant changes any time soon.

Then again we're now heading toward more and more SOC solutions and OEM vendor relationships so I am curious as to how AMD competes with Intel in those spaces now that supposedly the OEM level playing field has been set at equal competitive level after all that jazz with lawsuits and settlements...

 

[degibson]

... The fundamental truth that is missing is that no amount (low or high) of R&D expense can overcome the basic material and manufacturing limitations of the product/technology that's being improved...

Tell that to the guys who invented CMOS. (Fairchild Semiconductor) The truth is that the only think that can overcome the basic material and manufacturing limitations of the product/technology is R&D.

 

[veri745]

In my opinion, the largest disadvantage AMD has compared to Intel is not process tech or innovation. I think its marketing and marketing people. AMD obviously does not have the budget to compete with Intel in marketing. They really truly do need innovative and daring people in their marketing to compete. Trouble is, I think AMD marketing is more conservative than any I've ever worked with ... including Intel.

Knowing various people from both Intel and AMD marketing "machines" and their philosophies, I wouldn't hold my breath for AMD's marketing to make any significant changes any time soon.

Then again we're now heading toward more and more SOC solutions and OEM vendor relationships so I am curious as to how AMD competes with Intel in those spaces now that supposedly the OEM level playing field has been set at equal competitive level after all that jazz with lawsuits and settlements...

Good post. I agree that AMD hasn't really attempted to market their products in the US very actively. I think some of that is changing in other developing markets like India and China though.

From what I've heard, most of AMD's marketing dollars actually goes in to those markets where they think they can make the most headway in developing and budget-minded markets where performance/dollar matters most.

 

[sandorski]

Comparing R&D expenditure doesn't mean much really. Intel has blown huge chunks of $R&D through the years on failed and canceled projects. AMD seems to be much better on that front, at least I have never heard of them scrapping projects.

 

[PreferLinux]

I have: K9.

 

[sandorski]

I have: K9.

Canceled after 6 months.

 

[halley]

My next build is going to have a 4k budget, im going to go with whatever processor performs best regardless of price. Does that make me stupid? maybe, but it's my money and i can afford it.

If you go with whatever processor performs best regardless of price AND USEFULNESS OF IT, that makes you look selfish. We live in a world where wasting is no more encouraged.
Does that make you stupid? I don't know because I don't where you live, what you do to earn your living, what your hobby is...