AMD Q313 Results

[MisterMac]

Cheapness will prevail , it has always worked this way
in the semi industry.

http://www.electronicsweekly.com/news/business/arm-datacentre-processor-from-amd-2013-10/

...and your article has what relevance to your words?

perf\w will be terrible - and has to be alot better to disrupt the softwarestacks atm.

 

[krumme]

...and your article has what relevance to your words?

perf\w will be terrible - and has to be alot better to disrupt the softwarestacks atm.

Intel won the x86 server market using the excact same strategy.
It all depends on the a57 and 16nm efficiency.

I think it will take many more years for arm to be there. But thats what we thought in nienties too about Intel

 

[mrmt]

If we look at the results Intel is doing okey on the traditional market. The downturn in the market have stopped. But Amd is tanking big time here. Staying in x86 and amd is dead because Intel is pressed to get all market here. The change for the new strategy is urgent. And that means get away from x86 and a monopoly situation

Intel might be pressed here but I don't think get out x86 is an answer. AMD is shielded by the x86 monopoly, the market is divided by Intel doing all the profits and AMD getting enough money to stay alive. K8 changed this but things are getting back to that point since Conroe.

By getting out of x86, AMD loses this shield. They are going for the ARM market, a market which is essentially open. Almost everybody can buy a license, and there are a lot of bigger-than-AMD players out there, both in the high end market and in the bottom end market.

There are some caveats here. The foundry is TSMC for everybody, but then we have to think about the WSA, and the vanilla ARM core makes a nice baseline for performance projections, but why would Apple and others invest in custom ARM cores if there were huge benefits in doing so?

FWIW it will be hard for AMD to get totally outclassed in the ARM world as they were in x86, even with their limitations, but OTOH they face a much tougher competitive environment in the ARM world, and let's not forget that Intel will take on ARM too. It will be funny if their ARM offerings get pounded by Intel x86 offers.

 

[Tsavo]

If the CPU division loses a bit more, they will be dwarfed by the graphics solution. Then they can rename to ATI lol!

Allied Transistors, Inc has a ring to it.

 

[Pilum]

Intel won the x86 server market using the excact same strategy.
It all depends on the a57 and 16nm efficiency.

I think it will take many more years for arm to be there. But thats what we thought in nienties too about Intel

If Intel had the same business strategy as the classic RISC vendors had in the 90s (concentrate solely on high-profit big iron), ARM could have a valid attack vector. But Intel already adjusted its server strategy to fend off attacks in the low-end segment; the very first Silvermont Atoms weren't mobile, but server C-series. I expect this focus to remain the same for Airmont and Goldmont.

ARM will have advantages for adding custom blocks, but it remains to be seen how much of an advantage that turns out to be. ARM will certainly find some uses, but getting into the server mainstream market will be an uphill battle. Intel has a very strong and seamless lineup from tiny 2C2T microservers up to (with upcoming IVB-EX) 8P 120C240T big iron systems. There aren't any gaping holes in the product lineup which new vendors could easily break into; they'll have to compete with the existing products from the get-go - that's a completely different situation from the one Intel exploited in the 90s.

Also, it's not one vendor trying to break into Intels turf, but several, who will compete against one another as well as Intel. That's actually one problem the RISC vendors suffered from: they fought isolated battles against Intel, instead of unifying their struggle. I don't think that the ARM vendors will fare better; while they may use the same ISA, most of the CPU development will be decentralized. That's not a winning proposition in an industry which pretty much defines "economies of scale".

But we'll see. There are lots of opportunites for Intel to screw up and give the ARM guys a chance to break into the market. And the developments will be interesting to watch, regardless of ARMs success of lack thereof - competition is always good.

 

[krumme]

If Intel had the same business strategy as the classic RISC vendors had in the 90s (concentrate solely on high-profit big iron), ARM could have a valid attack vector. But Intel already adjusted its server strategy to fend off attacks in the low-end segment; the very first Silvermont Atoms weren't mobile, but server C-series. I expect this focus to remain the same for Airmont and Goldmont.

ARM will have advantages for adding custom blocks, but it remains to be seen how much of an advantage that turns out to be. ARM will certainly find some uses, but getting into the server mainstream market will be an uphill battle. Intel has a very strong and seamless lineup from tiny 2C2T microservers up to (with upcoming IVB-EX) 8P 120C240T big iron systems. There aren't any gaping holes in the product lineup which new vendors could easily break into; they'll have to compete with the existing products from the get-go - that's a completely different situation from the one Intel exploited in the 90s.

Also, it's not one vendor trying to break into Intels turf, but several, who will compete against one another as well as Intel. That's actually one problem the RISC vendors suffered from: they fought isolated battles against Intel, instead of unifying their struggle. I don't think that the ARM vendors will fare better; while they may use the same ISA, most of the CPU development will be decentralized. That's not a winning proposition in an industry which pretty much defines "economies of scale".

But we'll see. There are lots of opportunites for Intel to screw up and give the ARM guys a chance to break into the market. And the developments will be interesting to watch, regardless of ARMs success of lack thereof - competition is always good.

Good post. The only thing i dont quite quite agree on is the economies of scale, because it depends on where this economies of scale is in future situation, where arm migh have a chance. If economies of scale eg. means sharing cost with toasters and microwaves the situation will change. Its all a lot of ifs. Intels server portfolio looks the most secure in their entire product stacks, and very strong imho. Its got plenty of legs if needed. Intel is just cruising slowly right now.

 

[Unoid]

So you base investing decisions based upon current performance?

You realize the market is priced based upon future performance?

I stated future performance. meaning I guessed AMD would beat expectations for Q3 and they did, but factoring in marketshare and margins ended up dropping the price.

Instead of sounding condescending, perhaps positive criticism would be nice?

 

[mrmt]

I stated future performance. meaning I guessed AMD would beat expectations for Q3 and they did, but factoring in marketshare and margins ended up dropping the price.

AMD is down because the business that pays the bills, CPU, is shrinking fast, and the new business they got is too small to offset these losses. This is far more important than beat revenue/profit forecast.

 

[krumme]

AMD is down because the business that pays the bills, CPU, is shrinking fast, and the new business they got is too small to offset these losses. This is far more important than beat revenue/profit forecast.

Excactly my interpretation too.

Amd new strategy is working and the "mantling" of the market is a great opportunity that i dont remember amd had before.

But the strategy and mantle is a risk. Its fine if amd could earn money or not lose so many on the traditional market. But they do lose lot here even when times have changed on the market.
And a lot of the cost is fixed here. It takes several years to adopt your strategy and who wants to bet on it while the core business is bleeding? Thats a damn high risk/reward imho.

Now keller says good times is comming. But come on. This time it have to be proved. Only results count.

 

[mrmt]

Excactly my interpretation too.

Amd new strategy is working and the "mantling" of the market is a great opportunity that i dont remember amd had before.

AMD golden opportunity was K8. If they had pursued that route instead of Bulldozer they would be in position to transition to fabless, negotiate the settlement with Intel and a far more solid x86 line.

I may be missing something but I can't think of yet another proprietary APU, and one that is not backed by a software giant is a business opportunity. Quite the opposite, this will be another CUDA.

 

[Ajay]

I may be missing something but I can't think of yet another proprietary APU, and one that is not backed by a software giant is a business opportunity. Quite the opposite, this will be another CUDA.

Not getting buy in from MS was, perhaps, a fatal misstep for Mantle and could relegate it to niche status; in which case it will be more like PhysX than CUDA.

 

[mrmt]

Not getting buy in from MS was, perhaps, a fatal misstep for Mantle and could relegate it to niche status; in which case it will be more like PhysX than CUDA.

I don't think it does. Mantle fundamental problem is that it is AMD only, and even worse, GCN only. If you are not a happy GCN owner, you are out of luck. If you are happy post-GCN owner, you might be out of luck depending on how AMD evolve their GPU architecture. Even if AMD somehow provide backward compatibility, you might depend on the game vendor to get your game running with you post-GCN card.

This is a return of the wild west of the pre-DX world, where every game had to write a code path for specific hardware. Given that gaming will be stuck with anemic AMD hardware for the time being, I don't think the trade off is worth in the PC ecosystem for extra 10-15% in performance, and I'm glad Microsoft is shunning the thing. Maybe the guys running overclocked low end systems will be happy, maybe the guys running 120fps in 120Mhz screens will be happy, but it will be at the expense of added complexity for the entire ecosystem.

 

[krumme]

Consoles constitute like what 90% of the gaming market? Lol.

 

[Ajay]

I don't think it does. Mantle fundamental problem is that it is AMD only, and even worse, GCN only. If you are not a happy GCN owner, you are out of luck. If you are happy post-GCN owner, you might be out of luck depending on how AMD evolve their GPU architecture. Even if AMD somehow provide backward compatibility, you might depend on the game vendor to get your game running with you post-GCN card.

This is a return of the wild west of the pre-DX world, where every game had to write a code path for specific hardware. Given that gaming will be stuck with anemic AMD hardware for the time being, I don't think the trade off is worth in the PC ecosystem for extra 10-15% in performance, and I'm glad Microsoft is shunning the thing. Maybe the guys running overclocked low end systems will be happy, maybe the guys running 120fps in 120Mhz screens will be happy, but it will be at the expense of added complexity for the entire ecosystem.

You are right, but do you really think AMD would have ever gotten buy in from NV, or developed this as an industry X-platform project? Sweeney and Carmack have been asking for something like Mantle for years - they knew, and didn't care, that they'd have to maintain separate code bases for different platforms. Oddly, now that Mantle gives them what they've asked for, I haven't heard much from either, but I haven't looked either. As far as CGN only goes, AMD has to start somewhere. But again, I don't see this as selling a statistically significant number of additional video cards for them, so what is the point?

 

[ShintaiDK]

Consoles constitute like what 90% of the gaming market? Lol.

No. At best consoles might have a slight majority in revenue vs PC. But gaming wise they are beaten long ago.

 

[krumme]

No. At best consoles might have a slight majority in revenue vs PC. But gaming wise they are beaten long ago.

You are right. Its more a less equal - but probably in 5 years mobile is even to the two. Mobile is where the big growth is.
And yes the quality of the pc games is far superior, not only gfx but for entire gaming experience imho. If cod on pc is lame, cod on ps3 is uber lame. FPS without mouse is pathetic.

 

[ShintaiDK]

You are right. Its more a less equal - but probably in 5 years mobile is even to the two. Mobile is where the big growth is.
And yes the quality of the pc games is far superior, not only gfx but for entire gaming experience imho. If cod on pc is lame, cod on ps3 is uber lame. FPS without mouse is pathetic.

 

[Ajay]

Odd graph coming from Intel. Only 2 years cumulative for PCs, but the entire installed base for consoles? Seems like they are selling themselves short

 

[HurleyBird]

Enthusiast notebook?? Wii has twice the software revenue of the 360... in 2013??

 

[ShintaiDK]

Enthusiast notebook?? Wii has twice the software revenue of the 360... in 2013??

Wii is without doubt the best selling console. And people still buy plenty of games for it.