Will AMD ever be able to compete with Intel again?

[steve wilson]

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?

 

[NTMBK]

This can only end well...

Perhaps. But the question is legitimate. If people misbehave and treat others poorly, then I will just send them on vacation. This will be the one and only warning to everyone in this thread.
-ViRGE

 

[ShintaiDK]

Just as likely as VIA will.

 

[naukkis]

Sure, Intel ain't innovative any more and it takes just some good ideas go best Intel's offerings. But lately AMD have mostly copied Intel's strategies, like implement always lengthening SIMD word length which have come to limit cpu's clocking potential.

We need sky high integer clock frequencies with high IPC(like power 8) and keep fp unit so simple that it will not reduce clocking potential and move SIMD operations to different units with HSA. If AMD could implement that they will have something to counter against Intel.

 

[OCGuy]

The answer is fairly obvious, but this dead horse of a thread should receive the same treatment as the Anti-Intel trolls'.

 

[AtenRa]

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

They sure did and I dont expect that to change soon,

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

If AMD would have access to the same or equivalent manufacturing process at the same time as Intel, they could.

 

[SAAA]

To compete they need to adress both the lower performance per core but more importantly the amount of cores. If somehow AMD can catch the first one I'm not sure of the second: Intel can easily go the moar cores route and start to sell 6-8 cores in mainstream if it's pressed by competition. Sure it's easier than doube the IPC again.
Also it's not like they will mine too much the profit of 12-15 cores chips, remember that in the Pentium days the server and consumer chips were almost the same but now high end has 4 times more cores...

 

[Qwertilot]

Not sure if non competition is entirely inevitable due to the process gap.

I mean what AMD were more less matching Sandy Bridge but selling them notably cheaper than Intel are/were selling Ivy/Haswell? That'd have to qualify as real competition for desktops at least. Or Qcom et al in mobiles.

Maybe more likely to be really interesting if stacked ram finally sorts out IGP to be a really big factor? I guess that'll be low/medium end though

 

[DigDog]

Intel could become complacent again; AMD could recruit a new technician who turns out to be a genius. AMD could do well on the business side and get many profitable contracts while Intel could do the opposite.

in theory. in practice, not for the foreseeable future.

AMD can still grab many business deals that Intel might not be interested in. And there is always the chance that new markets / technological demands might arise, giving them somewhat of a level playing field, where design matters more than "huge stashes of cash and gargantuan industrial prowess" .

 

[mrmt]

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?

AMD was only competitive when they acquired IP and technical teams from DEC and had bleeding edge fab technology from Motorola and IBM. Today they have no source for bleeding edge fab technology, they do not have the money to buy IP or poach a more capable team from others players and their R&D budget is smaller than Nvidia's, let alone Intel's. So no, AMD won't become competitive with Intel. That ship sailed when AMD decided to chase Bulldozer for 6 years.

 

[BSim500]

To compete they need to adress both the lower performance per core but more importantly the amount of cores. If somehow AMD can catch the first one I'm not sure of the second: Intel can easily go the moar cores route and start to sell 6-8 cores in mainstream if it's pressed by competition. Sure it's easier than doube the IPC again.

^ This. Intel's cores at 3.0-3.3Ghz @ 0.8-0.9v can pull as little as 13-16w per core, so a 6-8 core 90-130w CPU will pull only a little more than a Sandy Bridge quad-core even on 22nm still (and still less than an FX-8350). Hell, they've already done that with some Xeon's (ie, E5-2690 v2 = 10x cores (20-threads) + 25MB cache = 130w). It's much harder for AMD to massively boost IPC by 70% or so, which is what counts in most games. Personally, I hope they do get back into it but it's pretty expensive to pull off a new architecture, which is AMD's biggest problem (along with being a process behind).

 

[jpiniero]

Of course not. But they don't really need to with majority of the market heading to smartphones, tablets, ultrabooks, etc.

They will fail there too.

 

[Fjodor2001]

Apple was close to bankruptcy, but is today one of the top tech companies in the world. Opposite to that Nokia was once the totally dominating force in the mobile phone business, but has been now become more or less irrelevant in just a few years. So of course, anything can happen. And it's happened many times before.

Big companies fail and are replaced by new ones. Former giants get reduced to almost nothing. And companies that used to be almost nothing grow to become giants again. The IT business is a highly dynamic and ever changing landscape.

 

[witeken]

Competing with an ever growing manufacturing disadvantage is almost impossible.

 

[mrmt]

Intel could become complacent again;

K7 and K8 were excellent products but they were only as successful as they were because of Intel's mistakes, not because Intel became complacent. It was the strategic mistake of dumping P3 for Netburst *and* insisting on it for 6 years that allowed AMD to blossom with K7 and K8. If the world didn't take a turn towards mobile, which forced Intel to work P3 as a low power architecture, outside the netburst team influence, Intel would be in dire trouble once they saw they wouldn't be able to go past 4Ghz.

AMD wouldn't need Intel to become complacent again, AMD would need Intel to make another strategic mistake like Netburst. And even so, I doubt that AMD would be able to capitalize on it given the poor state of their balance sheet, that honor would go for Qualcomm, Mediatek, Samsung and others.

 

[witeken]

Apple was close to bankruptcy, but is today one of the top tech companies in the world. Opposite to that Nokia was once the totally dominating force in the mobile phone business, but has been now become more or less irrelevant in just a few years. So of course, anything can happen. And it's happened many times before.

Big companies fail and are replaced by new ones. Former giants get reduced to almost nothing. And companies that used to be almost nothing grow to become giants again. The IT business is a highly dynamic and ever changing landscape.

Yes, but the CPU landscape is surprisingly straightforward: the continuation of Moore's law and making CPUs simply faster. That's something different than Nokia which became obsolete because it didn't follow the trend of touch screens and Android. Processors don't change, but the products they are placed in do. Intel is slowly getting more obsolete with mobile becoming ever more important, but they still have the desktop and server, those won't suddenly go away, and now Intel's also quickly catching up there.

 

[AtenRa]

AMD has a chance if they will use 14nm FinFets in 2016. Intel will still be at 14nm both in Mainstream/High-End desktop and Servers.
We will have to wait and see what AMD targeting at for their new x86 mArchitecture.

 

[CHADBOGA]

Ever is a very time, but we will see how AMD do with their next gen architecture and if that doesn't get close, then I would be inclined to rule out the possibility of them regaining the glory years of 2004/5.

 

[CHADBOGA]

AMD has a chance if they will use 14nm FinFets in 2016. Intel will still be at 14nm both in Mainstream/High-End desktop and Servers.
We will have to wait and see what AMD targeting at for their new x86 mArchitecture.

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

 

[witeken]

AMD has a chance if they will use 14nm FinFets in 2016. Intel will still be at 14nm both in Mainstream/High-End desktop and Servers.
We will have to wait and see what AMD targeting at for their new x86 mArchitecture.

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

Intel will likely launch 10nm in H1 2016.

 

[parvadomus]

AMD is shrinking the gap in mobile, at least Kaveri looks much more competitive than Richland.

 

[steve wilson]

So basically unless AMD get bought out by someone huge (say Microsoft/Google) then they are only going to be competitive at the mid to low end of the market?

I really hope they get bought out.

 

[PliotronX]

I hope they leave multipliers and virtualization unlocked if they do :whiste:

 

[parvadomus]

So basically unless AMD get bought out by someone huge (say Microsoft/Google) then they are only going to be competitive at the mid to low end of the market?

I really hope they get bought out.

Do you remember the K5 days? AMD was so behind everybody was saying it was dead. Then then they got the K6 and K7 (which was on par with Intel P3, or better). And K8 totally destroyed intel offerings.

 

[Fjodor2001]

the CPU landscape is surprisingly straightforward:

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.