Will AMD ever be able to compete with Intel again?

[SlowSpyder]

Considering the difference in R&D budgets (Intel's R&D for 2013 was about 2x AMD's total revenue for the same year) I'd say the fact that AMD can compete with or beat Intel in any benchmarks is a pretty big feat. They have some smart people working for them, and I bet they could compete on the high end as long as they have a manufacturing process that isn't too far behind what Intel has. But if Intel can keep stretching their process advantage further and further then I don't think AMD has a chance. If AMD can get on a level or almost level manufacturing process, I think they can.

But I think the bigger question is will they even want to compete in the high end desktop or x86 server market? AMD needs to focus on markets where they can make money, desktops might not be the smart area for AMD to use huge chunks of their R&D. Mobile is growing and will probably continue to grow, but then again that is a pretty crowded market right now.

 

[mrmt]

Considering the difference in R&D budgets (Intel's R&D for 2013 was about 2x AMD's total revenue for the same year) I'd say the fact that AMD can compete with or beat Intel in any benchmarks is a pretty big feat.

That's the results of when the R&D gap difference wasn't that big. Now this R&D gap is even wider, why should the performance gap not to become wider?

 

[coercitiv]

That's the results of when the R&D gap difference wasn't that big. Now this R&D gap is even wider, why should the performance gap not to become wider?

diminishing returns

 

[sm625]

Once upon a time.... AMD had almost twice the performance per watt. In order to repeat that history, they need another serious innovation in memory latency. 70, 60, even 50 clock cycles is too much. There has to be a way to cut it down to 25. It is far more difficult to reduce cache latency, but they have to figure it out. Combine those two points with a better implementation of SMT and they can have a winning design. But I dont think they can really win on that front. Maybe compete, but not outright win. They need something else, something out of left field.

The latency on SATA and even PCIe is really bad. If the memory controller emulated a SATA port for the OS, and routed that data out onto a DIMM with several GB per sec bandwidth and 95% reduced latency, it would absolutely skyrocket the user's perception of performance. Even just one or two NAND dies tied directly to the cpu using the lowest latency interconnect possible would do wonders for real world user experience. That is the kind of low hanging fruit that they need to grab. Rather than adding $40 to the BOM for an SSD cache, you simply solder a $7 128Gbit NAND die onto the motherboard and you have a 16GB cache for basically nothing. The fact that they didnt do this 5 years ago is one of the big reasons why they are where they are. This is one of the biggest advantages phones have over desktops.... low latency low cost per GB storage.

 

[SlowSpyder]

That's the results of when the R&D gap difference wasn't that big. Now this R&D gap is even wider, why should the performance gap not to become wider?

I didn't say that couldn't happen, it very well could. But, maybe AMD has the right people in place to do more with less compared to before. Maybe not. But it isn't some kind of set in stone hard rule that if Intel increases their R&D by 5% and AMD decreases their R&D by 5% (or whatever the real numbers are) that automatically has to translate into that much bigger performance gap in their products.

 

[mrmt]

diminishing returns

Diminishing returns means that if AMD budget kept a constant rate compared to Intel R&D, the gap would shrink. But the gap is *widening*, because not only Intel budget is growing, AMD budget is shrinking. Plus diminishing returns in a project should be viewd within the context of aa single technology only, not to the entire technology project.

AMD has no answer for Intel's latest QPI implementation, their SATA implementation is already subpar, they are lagging behind in power consumption, they lagged PCIe 3 implementation despite being a leader in graphics, they have no security implementation of their own and they have 0 modem/radio to talk about. So while we may have very low returns in, let's say, adding a new integer unit, there are plenty of technologies that haven't reached maturity and will improve the product if they get R&D funds. This is where AMD lack of R&D will bite.

 

[Arachnotronic]

It will be very difficult for AMD's 2016+ K12/x86 equivalent to go up against Skylake, let alone Cannonlake. Intel's manufacturing lead is formidable, its cost structure is much better (which allows it to more profitably put in more features/performance), and its physical design teams are simply larger/more capable.

Look at the damage that Bay Trail-M is doing to AMD's computing solutions group...the chip that so many have mocked is winning almost all of the attractive low-cost notebook sockets. And, no, there's no "contra-revenue" here -- these are extremely profitable for Intel.

Once Intel moves to Braswell with Gen 8 GPU + 14nm + a focus on lowering system BoM, it really will be more or less lights out for AMD across the spectrum of PCs. And while AMD will hype K12 until it is blue in the face all throughout 2015 (AMD is doing the NVIDIA project denver thing..."high performance ARM core"), it ultimately won't save them.

AMD's best bet is to more or less put all of its energies into becoming an NVIDIA competitor. Semi-custom + dGPU should serve the business well, but any hopes that AMD will -- in an environment where Intel, Apple, Qualcomm, etc. all have soaked up the best engineers -- regain former glory is just misguided at best.

 

[mrmt]

I didn't say that couldn't happen, it very well could. But, maybe AMD has the right people in place to do more with less compared to before. Maybe not. But it isn't some kind of set in stone hard rule that if Intel increases their R&D by 5% and AMD decreases their R&D by 5% (or whatever the real numbers are) that automatically has to translate into that much bigger performance gap in their products.

So you expect AMD to do more than Intel, Qualcomm, Samsung and others with 30% less R&D budget than they had in the past just by having the "right people in the right place"?

I'll tell what I'm reading here: You have no basis to believe that and you know this is a miracle. You just *hope* they pull that miracle because you want AMD to succeed.

 

[sandorski]

There isn't much of a Market for that any more. Certainly not a growing Market anyway. The age of the ever more powerful CPU has passed, now it's all about Power Savings and Efficiency.

 

[TeknoBug]

If you looked at the real roadmap (not the fake black background one), AM3+ has no future past 2015 and AMD has been churning out APU's lately so I don't think they're interested in doing more mid-high end CPU's anymore. The FM socket Athlons has some potential though, I'm eager to see if they're going to release the Steamroller version, the Athlon X4 760K isn't half bad (when OC'd to 4.5GHz). The refreshed Haswells is a good indication that Intel isn't too worried.

People talk about possibly an AM4 and DDR4, AMD hasn't even peeped a word about those kinds of things.

 

[piasabird]

Intel will be just like AMD soon if they don't start making what the Customer wants to buy. I just got a new Tablet and it does not have an Intel or an AMD processor in it. The desktop sector has shrunk a lot since the i-phone and the i-pad came out. I think a lot of this was made possible with phone technology and SSD speed and Apple having the guts to come out with new products. As you see processors like the QUALCOMM Snapdragon come into play in the tablet sector competition has increased. The weak will not survive in the fast moving technology sector.

Intel had a virtual monopoly in the desktop sector. I bet you will see tablet designs start to creep into the desktop sector. For lightweight computers that might be ideal. Only gamers really need a full-size desktop.

Monopolies are slow to respond to demands of the customers. Why else has it taken so long for an unlocked Pentium and low powered Celerons and the high-end graphics to come to the desktop? If Intel does not deliver these they will lose even more customers.

 

[AtenRa]

It will be interesting to see what the time gap is between the Foundries' FinFet efforts and Intel's 10nm process.

I dont expect 10nm products earlier than Q4 2016-Q1 2017. And if you actually want me to predict the future, i would say no 10nm products before Q2 2017.
Also since we are talking Desktop here, at least Im, first 14nm Intel Desktop will not release before mid 2015. So 14nm Desktop will carry on for 2016 as well. That leaves a nice opportunity for AMD in 2016 to come back.

AMD could use 14nm FinFets at GloFo in 2016. The only variable as of now is IF they are willing to use an expensive process at that time and for what SKUs.

 

[AtenRa]

It are 20nm FinFETs, not 14nm FinFETs, the process is called 14/16nm.

GloFos 20nm is Planar not FinFet. Samsung 14nm is FinFet.

 

[SlowSpyder]

So you expect AMD to do more than Intel, Qualcomm, Samsung and others with 30% less R&D budget than they had in the past just by having the "right people in the right place"?

I'll tell what I'm reading here: You have no basis to believe that and you know this is a miracle. You just *hope* they pull that miracle because you want AMD to succeed.

No, that's not at all what I said. It seems like if I say anything other than AMD has zero chance of ever competing on the high end again you're going to disagree with me. I think you have to turn a blind eye to tech history to assume that once a company is in a dominant position they are unable to be surpassed by a competitor.

Reread my posts, I never said I see AMD pulling past Intel, I don't know why you reply to me like that is my position. But I don't think it is impossible either, however unlikely (unless, as I also clearly posted, Intel can continue to stretch their manufacturing advantage further and further).

I don't *hope* anything. You made a post stating that gap in the R&D budgets are even bigger today than in 2013, I replied to that. I didn't base my reply on hope, but on logic. Or are you saying that if company A increased it's R&D over competitor, company B, by 7% that automatically means company A's new products will be exactly 7% faster? If that isn't your position than I don't see how you can disagree with what I'm said in reply to you.

I think you should reread my posts and take them at face value, you seem to be adding things to them I never said or implied.

 

[mrmt]

I don't *hope* anything. You made a post stating that gap in the R&D budgets are even bigger today than in 2013, I replied to that. I didn't base my reply on hope, but on logic. Or are you saying that if company A increased it's R&D over competitor, company B, by 7% that automatically means company A's new products will be exactly 7% faster? If that isn't your position than I don't see how you can disagree with what I'm said in reply to you.

No, it doesn't. It could mean 0% increase but it could mean 100% increase if company A focused this 7% in a revolutionary development that company B cannot afford. In R&D you don't get what you didn't pay for, and AMD planners won't try to develop anything they cannot afford to. A shrinking budget for AMD is a warranty that whatever AMD is trying to develop, will be a much narrower scope than previous products.

 

[guskline]

I'm no expert, but it seems to me that Intel have had AMD beat at the top end for a long time now.

Will AMD ever be able to compete with Intel at the mid to high and high end again?

With the impending release of Haswell E, Devils Canyon etc by Intel, the obvious answer is no. However, When CEO Rory took control his statements and actions appear to support this concept. It appears AMD's emphasis is changing to mobile ( not a bad thing). They still appear to have a persona of "moar cores" but CEO Rory seems to be trimming the company to be more competitive.

 

[SlowSpyder]

No, it doesn't. It could mean 0% increase but it could mean 100% increase if company A focused this 7% in a revolutionary development that company B cannot afford. In R&D you don't get what you didn't pay for, and AMD planners won't try to develop anything they cannot afford to. A shrinking budget for AMD is a warranty that whatever AMD is trying to develop, will be a much narrower scope than previous products.

So you are agreeing with what I said when I posted this in reply to you:

But, maybe AMD has the right people in place to do more with less compared to before. Maybe not.

That a change in R&D budgets doesn't automatically mean a bigger performance gap. It could be R&D to make the exact same parts cheaper and more profit margin. It could be R&D Intel adds to enter different markets with their CPU's that AMD doesn't compete in. It could be R&D that Intel uses to find a successor to silicon for future products. And of course the R&D could certainly translate into a bigger performance difference.

The topic of the thread is "Will AMD ever be able to compete with Intel again". My posts, to summarize, are saying basically I don't think it is absolutely impossible unless Intel continues to stretch their manufacturing process lead, but I'm not sure AMD would even want to try (in regards to higher end desktop / x86 server) given the current direction of personal computing trends. But I do believe it could happen, at least on a level where AMD is a viable alternative above the low end, if that is where AMD really wanted to focus their resources. If you're not disagreeing with that, I'm not sure why your posts seem so argumentative when replying to me.

 

[Mand]

The real question is:

"Compete in what?"

High-end desktop CPUs, meant to be used with a discrete GPU? I really doubt it. Small-device, SoC, or budget machines? Sure.

 

[AtenRa]

The real question is:

"Compete in what?"

High-end desktop CPUs, meant to be used with a discrete GPU? I really doubt it. Small-device, SoC, or budget machines? Sure.

AMD will most certainly use the new x86 mArchitecture for its server parts. Then a desktop product could be possible.

 

[bunnyfubbles]

sure, AMD still has a significant advantage in terms of GPU, if they can solve the memory bandwidth problem, they can produce something that I don't know that intel could compete with. (granted, this advantage might not last forever, as Intel is continually upping their efforts to improve their iGPU, and with the materials advantage, they can cut corners to help close the gap)

The problem is that I don't know how appealing that product would be to more than just slightly above casual gamers who can't afford (or just don't want to spend) more than midrange.

Obviously there would be a ton of users on tech sites like this one that would be thrilled with such a product, but it still would have the weak CPU and weak performance/watt and thus not be quite so appealing to all the people who just don't need a beefy GPU

 

[crashtech]

It's not just the process lead that makes Intel's high-end parts so dominant, as evidence see how well Sandy Bridge parts still compete with the newest offerings. If AMD can successfully introduce a new µarch that is substantially superior to Bulldozer and its successors, they could become competitive again even with a process disadvantage.

 

[TheThirdMan]

The real question is:

"Compete in what?"

High-end desktop CPUs, meant to be used with a discrete GPU? I really doubt it. Small-device, SoC, or budget machines? Sure.

This is my thinking. Even if they managed to design a new architecture to compete with Intel Core, by the time it gets to market, the demand for high powered desktop CPUs will be even less than it is now. PCs with full powered CPUs can't be long for this world except for PC gamers, professional creatives and servers, which must make up a tiny part of the market.

 

[AtenRa]

This is my thinking. Even if they managed to design a new architecture to compete with Intel Core, by the time it gets to market, the demand for high powered desktop CPUs will be even less than it is now. PCs with full powered CPUs can't be long for this world except for PC gamers, professional creatives and servers, which must make up a tiny part of the market.

Those three make a HUGE market, total server revenue for 2013 was 50-55B. Also, Big Data and Cloud servers are on the rise and they will continue to rise in volume and revenue.
Gaming Hardware market was close to 18B in 2013 and its increasing.

 

[DrMrLordX]

AMD has a considerable advantage over Intel right now when running HSA-enabled apps on Kaveri chips. Trouble is, there aren't enough of those apps (yet) to cast Kaveri or any future AMD APUs in a sufficiently-positive light. That advantage will disappear eventually, but if AMD is going to make any hay right now or in the very near future, HSA is their silver bullet (and, to a lesser extent, so is OpenCL 2.0).

 

[witeken]

Everything seems straightforward until something disruptive comes along that changes the landscape.

The development of mobile phones also seemed straightforward, and so did the development of horse carriages until the cars came along. The history is full of such events. The problem is that until the disruptive event has actually happened, it is very hard to predict. So everything seems straight forward until it happens, and then all changes so fast you don't even know what hit you until years later.

Yes, but I don't think that disruptive event will come from AMD. They do about the same as Intel, just with a lower budget and without the fabs.