With the current rate of Intel CPU performance increases, could AMD be catching up?

[Haserath]

Yep. Jim Keller is a superhuman that defies the laws of the universe.

/s

Yep, because one guy can design a cpu in a year.

You know he is a manager and not a designer, right?

http://en.wikipedia.org/wiki/Sergei_Korolev

Only many years later was he publicly acknowledged as the lead man behind Soviet success in space.

Competed and most likely would have beat the more resourceful US if he had not died.

 

[Idontcare]

http://en.wikipedia.org/wiki/Sergei_Korolev


Competed and most likely would have beat the more resourceful US if he had not died.

Weak. JFK got us to the moon despite being dead for the final 5 yrs of the race.

When you are good, you are that good.

Korolev wasn't that good, hence neither USSR nor Russia have succeeded in getting a man out of low-earth orbit to date, let alone landed on the moon.

Seriously though it is an immense disservice and slap in the face to the thousands, and it is thousands, of talented and hardworking engineers and project managers that all must work together to successfully complete a project as complex as landing a man on the moon or building a multi-billions xtor IC.

I refuse to believe that either Korolev or Keller would have such egos as to consider themselves any more critical to their respective project's success than hundreds of other equally brilliant colleagues on the same project. Anyone who would say otherwise IMO is really just full of hubris.

 

[mrmt]

Seriously though it is an immense disservice and slap in the face to the thousands, and it is thousands, of talented and hardworking engineers and project managers that all must work together to successfully complete a project as complex as landing a man on the moon or building a multi-billions xtor IC.

Keller may be good, but it's nonsense to imagine he will spend time doing engineering work. He is a manager, his time is expensive, he has duties to his functions and if he doesn't perform them. He needs to ensure that there won't be another Bulldozer receiving money in the R&D pipeline, and that the right projects are provided with enough resources. He is also responsible for the trade offs at the concept level.

If Keller gets down to the engineering level to do engineer's work, not only he won't solve AMD's engineering problem, he will also leave AMD without sound management, something the company is sorely lacking.

 

[Piroko]

In R&D, there is no such thing as getting something you didn't pay, but you sometimes don't get what you paid. So unless you are assuming that amd is really inefficient in r&d, there is no chance for them to churn out chips in more complex nodes using 20% less money.

Honest answer? Not a single fortune500 company can be called efficient in R&D spending. Some companies are only slightly wasteful, others are spending several times more than what they'll actually be able to put into products.
In general larger companies take more risks in R&D, which will bloat that budget considerably. I would be surprised if Intel spends more than 1/10th of their R&D on their own big x86 core. Lots of it is sunk into projects like Larrabee, TPM, McAffee, Thunderbolt, onboard highspeed interconnects, SSD controllers, UEFI, Wlan chips, you get the point. Additionally they pay an arm and a leg for a headstart with 450mm wafer machinery, something which isn't even sure to get any Roi.

As to AMD, we have no knowledge in how the R&D is distributed within the company and thus it's completely futile in basing arguments on R&D alone. We can make educated guesses at what gets the short stick (the canadians afaik) and what might be short term consequences. Long term implications? Forget it, I've given enough reasons why talk about them is more often than not bound to fail.

A small business can spend virtually nothing in R&D and "grow quicker" than a large corporation. It's a meaningless metric when used out of context as it has been so often in this thread.

Volkswagen is the third largest car company in the world, trailing Toyota by less than 8% in units sold. Revenue is expected to be in the same ballpark. It's also the eleventh largest R&D spender in the world (basis 2011, Toyota #1).

If anything, your post could be applied to AMD, but certainly not to VW.

 

[Idontcare]

Keller may be good, but it's nonsense to imagine he will spend time doing engineering work. He is a manager, his time is expensive, he has duties to his functions and if he doesn't perform them. He needs to ensure that there won't be another Bulldozer receiving money in the R&D pipeline, and that the right projects are provided with enough resources. He is also responsible for the trade offs at the concept level.

If Keller gets down to the engineering level to do engineer's work, not only he won't solve AMD's engineering problem, he will also leave AMD without sound management, something the company is sorely lacking.

Keller is good. The problem as I see it is that I'm guessing people here (not everyone of course) simply have no personal experience in witnessing just how large of a challenge it is for a large scale project to design a modern multi-billion xtor product so as to have an appreciation for how much of a group effort it is.

Via has brilliant people too, why aren't they rivaling AMD? Its easy, two reasons, (1) not enough brilliant people (hundreds can not do the work of thousands), and (2) not enough R&D money (hundreds of millions cannot fund the work than needs billions).

This is continuously sliding scale, and all that determines your position on the scale is your headcount and funding. That is what makes Boeing able to build the 787 and Airbus build the A380 but no others can build to similar scale. Same with AMD and Intel at 45nm and 32nm.

But 20nm and lower, it just doesn't work out for a business that is AMD's scale anymore than it worked out for a business that is Vias scale to put out a 28nm of the same complexity and die-size.

I do like the notion of a white-knight funding savior coming in and pumping a few billion into AMD. That would change up the game considerably, at least delay the inevitable for another node or two.

But to make the argument that one man alone makes the difference, sure it is possible, be him viewed as messiah or prophet or mere myth Jesus made a huge difference on mankind too, so maybe Keller is of comparable level.

But we cannot confuse what is possible with that which is probable.

 

[Fjodor2001]

But to make the argument that one man alone makes the difference, sure it is possible, be him viewed as messiah or prophet or mere myth Jesus made a huge difference on mankind too, so maybe Keller is of comparable level.

Perhaps Steve Jobs is a better comparison than Jesus in this case. ()

 

[frozentundra123456]

Keller is good. The problem as I see it is that I'm guessing people here (not everyone of course) simply have no personal experience in witnessing just how large of a challenge it is for a large scale project to design a modern multi-billion xtor product so as to have an appreciation for how much of a group effort it is.

Via has brilliant people too, why aren't they rivaling AMD? Its easy, two reasons, (1) not enough brilliant people (hundreds can not do the work of thousands), and (2) not enough R&D money (hundreds of millions cannot fund the work than needs billions).

This is continuously sliding scale, and all that determines your position on the scale is your headcount and funding. That is what makes Boeing able to build the 787 and Airbus build the A380 but no others can build to similar scale. Same with AMD and Intel at 45nm and 32nm.

But 20nm and lower, it just doesn't work out for a business that is AMD's scale anymore than it worked out for a business that is Vias scale to put out a 28nm of the same complexity and die-size.

I do like the notion of a white-knight funding savior coming in and pumping a few billion into AMD. That would change up the game considerably, at least delay the inevitable for another node or two.

But to make the argument that one man alone makes the difference, sure it is possible, be him viewed as messiah or prophet or mere myth Jesus made a huge difference on mankind too, so maybe Keller is of comparable level.

But we cannot confuse what is possible with that which is probable.

Granted AMD is behind in process nodes, but OTOH, there seems to be diminishing returns, really serious diminishing returns, from node shrinks. Look at Ivy and projected Haswell: really very little IPC and basically no clockspeed increase from Sandy Bridge. Granted there have been other improvements, but even the power savings have been less than spectacular. The only thing that has improved remarkably is the IGP, which is getting into AMD territory of being far more than the average user needs but still far from matching a low end discrete card of the current generation (ie, HD7750 level).

 

[ShintaiDK]

Granted AMD is behind in process nodes, but OTOH, there seems to be diminishing returns, really serious diminishing returns, from node shrinks. Look at Ivy and projected Haswell: really very little IPC and basically no clockspeed increase from Sandy Bridge. Granted there have been other improvements, but even the power savings have been less than spectacular. The only thing that has improved remarkably is the IGP, which is getting into AMD territory of being far more than the average user needs but still far from matching a low end discrete card of the current generation (ie, HD7750 level).

Performance/watt between SB and IB is around 35%. Not to mention the IGP.

Conroe to Penryn didnt bring much either as an example. Its simply a matter of the parameters. 3Ghz 65nm 65W to 3.33Ghz 45nm 65W. And then 7% IPC increase? Thats 19% performance/watt increase.

 

[mrmt]

As to AMD, we have no knowledge in how the R&D is distributed within the company and thus it's completely futile in basing arguments on R&D alone. We can make educated guesses at what gets the short stick (the canadians afaik) and what might be short term consequences. Long term implications? Forget it, I've given enough reasons why talk about them is more often than not bound to fail.

You are simply expecting them to stay on the bleeding edge on *four* different fronts when their R&D budget, that sustained only three fronts, shrunk 20% instead of grow 30% as reasonably expected. And this after AMD CEO state clearly that they won't go bleeding edge anymore. No matter how tap dance around this issue bringing Toyota and VW, this is AMD reality now and this is what you are refusing to address.

So, let's make of this a starting point for you:

Tell us how can AMD stay on the bleeding edge on *four* fronts with 20% less R&D budget.

 

[Idontcare]

Perhaps Steve Jobs is a better comparison than Jesus in this case. ()

Well that brings up another point which is if Keller really was all that some people want to believe him to be then there is no way Apple would have let him go.

If he really has this irreplaceable mountain of experience and innate capability that propels him to the top of the ranks of uber-engineers then Apple would have come in with their billions and billions in the bank and made Keller a ridiculously lucrative offer that a man of his intelligence simply could not refuse.

And yet they let him go. Apparently despite however supremely awesome he is, the ROI that comes from paying him his compensation package at Apple wasn't worth increasing the compensation package to keep him there.

 

[bononos]

......

I do like the notion of a white-knight funding savior coming in and pumping a few billion into AMD. That would change up the game considerably, at least delay the inevitable for another node or two.

But to make the argument that one man alone makes the difference, sure it is possible, be him viewed as messiah or prophet or mere myth Jesus made a huge difference on mankind too, so maybe Keller is of comparable level.

But we cannot confuse what is possible with that which is probable.

Would it be possible for the US govt to prop up AMD for one more stab at the mainstream cpu market?

The one man army CEO as messiah attitude is popular (or more popular) in America given the celebrity CEO culture. Top corp execs are feted and praised all the time as if they were the ones who routinely made it happen. vs the thousands/hundreds of useless asswipe engineers.

 

[mrmt]

Would it be possible for the US govt to prop up AMD for one more stab at the mainstream cpu market?

AMD business model is dead, they don't have scale, they need to change or die. No government can change that,

 

[Homeles]

Would it be possible for the US govt to prop up AMD for one more stab at the mainstream cpu market?

Ha! Our congress can't do anything.

 

[Idontcare]

Would it be possible for the US govt to prop up AMD for one more stab at the mainstream cpu market?

The one man army CEO as messiah attitude is popular (or more popular) in America given the celebrity CEO culture. Top corp execs are feted and praised all the time as if they were the ones who routinely made it happen. vs the thousands/hundreds of useless asswipe engineers.

You make an excellent point on the latter. I think you nailed it. The belief that one man can make so much difference is tied up part and parcel with the concept that a CEO is worth millions and millions by the same reasonings.

As for the US govt doing something to assist AMD...if the auto-bailout and banking fiasco is any indication I would say no because AMD isn't deep into the pockets of congress as the auto unions and bankers were/are.

They don't just hand out tax-payer money freely to just anyone, you have to have an established, socially-acceptable, means of kicking back oodles of cash to the congressman (jobs that can be taxed which are located in there constituency is a biggie) if you want them to hand you other people's money on a platter.

AMD doesn't have that kind of pull. They did for a while when they had fabs in Texas, but that was over 10yrs ago. Congress now would be more apt to bailout Samsung (fabs in USA) than they would be to bail out AMD IMO.

 

[MaxPayne63]

Why would anyone expect AMD to do any better with taxpayer money than they have with their investors' money?

 

[Piroko]

You are simply expecting them to stay on the bleeding edge on *four* different fronts when their R&D budget, that sustained only three fronts, shrunk 20% instead of grow 30% as reasonably expected. And this after AMD CEO state clearly that they won't go bleeding edge anymore. No matter how tap dance around this issue bringing Toyota and VW, this is AMD reality now and this is what you are refusing to address.

So, let's make of this a starting point for you:

Tell us how can AMD stay on the bleeding edge on *four* fronts with 20% less R&D budget.

First front GPUs: You imply that AMD has to stay at the bleeding edge because Nvidia would be on the bleeding edge. But Nvidia has been pulling away from said edge with ther big GPUs, trailing the actual bleeding edge products (Arm based) by almost a year now. And even then they ramped up slower than in the past. Shouldn't be hard to realize savings by adapting to Nvidias pacing - and I'm quite sure that this is what's being done, considering what happened at the last layoff round.

Second front low power APUs: Yep, they need to stay close to it with these products. But considering that it will be a TSMC fabbed chip afaik and TSMC itself is said to be quite a bit better at fabbing than glofo currently... I don't think this is a problematic front at the moment.

Third front high power APUs and Desktop/Server chips: Throwing them together because a lot of R&D is shared between them. The most problematic field and at the same time the most lucrative one. Don't have a solution here or else I would've sent an application and would be earning millions by now :awe:
They might fuse C32 with AM3 or drop G34, that could potentially save them money without taking it away from core development.

Fourth front console chips: Pretty much done, free resources to use elsewhere or remove.

Fifth front Node development/tool development: What's left of the ties to Glf will probably be cut. The first one to go was Soi, I wouldn't be surprised if there are more non-obvious money drains from AMD to Glf hidden in the R&D budget.

I'm not enough of an optimist, I wouldn't buy AMD shares myself. But I'm not a pessimist either, compared to them I wouldn't touch Facebook or Apple shares with a ten foot pole currently.

 

[Torn Mind]

You make an excellent point on the latter. I think you nailed it. The belief that one man can make so much difference is tied up part and parcel with the concept that a CEO is worth millions and millions by the same reasonings.

I think people tend to forget that once a guy moves "up the chain", what he is responsible for becomes less "deep" and more "wide". Take a coordinator taking the NFL head coaching job. Yes, some NFL coaches do call the plays or get involved in the gameplanning of the area they were coordinator in, but many times, they don't involve themselves in the details as much and delegate to a new specialist because they simply do not have the time to do a subordinate's job anymore.

 

[mrmt]

I'm not enough of an optimist, I wouldn't buy AMD shares myself. But I'm not a pessimist either, compared to them I wouldn't touch Facebook or Apple shares with a ten foot pole currently.

AMD fronts:

1) GPU Core
2) ARM core
3) Low power x86 core
4) High power x86 core

I don't think they are really investing on 4) because 2) and 3) will take a lot of resources in the next quarters.

They need 2) to be good otherwise their ARM effort will be a waste and 3) needs to keep up with Atom, ARM A15 and ARM64, if not their share on the market is toast too.

You are simply arguing that they can focus on those four with a budget 20% smaller than they had two years ago, and that they will succeed in those four fronts. And you don't call yourself an optimist.

 

[Piroko]

And your whole argument is based on a single variable which, while correlating with the subject, is in itself packed with plenty variables and also only part of plenty variables.

I've given examples that a higher R&D does not automatically force success or that a lower R&D does not spill doom. I've given examples of companies that lowered their R&D and still got more successful. I've given examples how and why and where the R&D can fluctuate without direct consequences to their main product line(s).

R&D itself isn't a good thing, unless it gets you Roi. See Rambus as a company which thought that everything can be fixed with more R&D. Or Nokia. Or Sony. Or Audi back in the 90s. Stop being focused on R&D so much. It's not a good indicator.

 

[AtenRa]

AMD fronts:

1) GPU Core
2) ARM core
3) Low power x86 core
4) High power x86 core

I don't think they are really investing on 4) because 2) and 3) will take a lot of resources in the next quarters.

They need 2) to be good otherwise their ARM effort will be a waste and 3) needs to keep up with Atom, ARM A15 and ARM64, if not their share on the market is toast too.

You are simply arguing that they can focus on those four with a budget 20% smaller than they had two years ago, and that they will succeed in those four fronts. And you don't call yourself an optimist.

As far as i know they will not design their own ARM core, they will use the "default" ARM design.

http://www.anandtech.com/show/6418/...d-opteron-cpus-for-servers-production-in-2014

Today's announcement is about a processor license, not an ARM architecture license - in other words, AMD will integrate an ARM designed 64-bit core for this new Opteron.

As of now, we know they are focusing on low power cores(Kabini/Temash) and performance/watt for their Server/APU (big-core) designs.

High Performance APU (like Trinity and Kavery) R&D is very low, once they have designed a new GPU architecture (GCN Tahiti) and a Server CPU Big-Core architecture (PileDriver, SteamRoller) they only have to merge the two.
Each technology(GCN, SteamRoller) generates its own R&D thus the APU design needs a fraction of the R&D money and time.

So yes, they can keep all three fronts with the current R&D budget.

 

[Torn Mind]

In the semiconductor industry, fixed costs are high and hence what can work in other industries will not for this particular one. Developing new products involves expensive skilled labor and lots of capital, both of the physical and human varieties.

 

[Ajay]

Ugh! The new version of iOS sucks. Typed up a post an then bang, the edit box just disappears Sorry for the OT, just needed to vent.

 

[mrmt]

R&D itself isn't a good thing, unless it gets you Roi. See Rambus as a company which thought that everything can be fixed with more R&D. Or Nokia. Or Sony. Or Audi back in the 90s. Stop being focused on R&D so much. It's not a good indicator.

You are mixing two things here. Those companies you mentioned are example of bad R&D, companies that generated a lot of research but could not bridge the gap between the lab and the market. Those companies generated products that from an engineering stand point, there was nothing wrong with them. It's just that those products were adapted to the market. Too expensive, not "fashion", not with features the consumers wanted, etc. This has nothing to do with AMD current situation.

There is a minimum you need for the development of a given product. You can't get a clip and a bubble gum and expect to have an explosive charge. You need a given level of resources, and this is what we are discussing here. In IC, you can't just chop 20% of your R&D budget and develop a chip in the following node at the same relative performance level in the same time frame, because it doesn't work, as the costs will go up 30%. Nobody ever achieved that, and this is what you are suggesting that AMD will accomplish.

Find an IC company that slashed 20% of their R&D budget and could get bleeding edge products out of the door in the following node in an acceptable time frame and we can start discussing again, if not, I rest my case.

 

[Piroko]

In the semiconductor industry, fixed costs are high and hence what can work in other industries will not for this particular one. Developing new products involves expensive skilled labor and lots of capital, both of the physical and human varieties.

Huh. Fixed costs are high if you want to build a forklift truck. Just the raw material needed for its counterweight is worth 1/6th of its sales price. Margins are in the area of 5~10% tops.
The semiconductor industry is a gold diggers dream in comparison. Fabless companies have almost no fixed cost. Developing new products takes little time compared to other industries.
The differenciation is that it's a very volatile market. What sells today will be a turd in a year. You have to constantly adapt your product, that's not something business administrators learn in college or like as a situation to be in. A lot of semiconductor companies went belly up not because their products were bad, but because they failed to capitalize on it.

 

[mrmt]

Huh. Fixed costs are high if you want to build a forklift truck. Just the raw material needed for its counterweight is worth 1/6th of its sales price. Margins are in the area of 5~10% tops.

Err... those are variable costs.