[AT] AMD Reorganizes Business Units - no more high performance x86 cores?

[pTmdfx]

It is just a reorg, and it is likely that it wants to separate the financial numbers of the server, embedded & SC businesses from its client business. Nothing really implies that it will no longer build any forms of big cores, especially in x86. So I'm kind of curious about how people could have this derived from this single piece of news...

 

[mrmt]

It is just a reorg, and it is likely that it wants to separate the financial numbers of the server, embedded & SC businesses from its client business. Nothing really implies that it will no longer build any forms of big cores, especially in x86. So I'm kind of curious about how people could have this derived from this single piece of news...

Not from a single. Q&A, conferences, FAD, everything is pointing out for no big cores and *nothing* to new big cores.

 

[mrmt]

Lisa will be COO and interim over the Enterprise, Embedded and Semi-Custom Business Group. It's Mr. Byrne that will oversee the Computing and Graphics Business Group that you mentioned.

Byrne will report to Lisa, no? The ultimate responsibility of the division lies with her, not with Mr. Byrne.

 

[inf64]

Not from a single. Q&A, conferences, FAD, everything is pointing out for no big cores and *nothing* to new big cores.

You must have missed x86 variant of K12 then. It's not "big core" per se but it is high-performance core. And it will focus more on perf./watt, like Core family.

 

[CHADBOGA]

You must have missed x86 variant of K12 then. It's not "big core" per se but it is high-performance core. And it will focus more on perf./watt, like Core family.

So will it only seek to compete against i3 and below?

 

[cytg111]

So will it only seek to compete against i3 and below?

It will be so hardcore that intel will have to invent an i9++ to compete. Nice fishing though.

 

[mrmt]

You must have missed x86 variant of K12 then. It's not "big core" per se but it is high-performance core. And it will focus more on perf./watt, like Core family.

I didn't miss, I just don't think it will be a competitor to Cannonlake in the same way Bulldozer was for Sandy Bridge. K12 will face a lot of cost constraints, because AMD cannot afford another wafer guzzler like their current big core chips, plus AMD is now with a much smaller R&D budget than they had with at the time they were developing their small core line.

Without another DEC to sell them IP and engineering teams, I wouldn't hold my breath for K12. K12 should be an extremely cost efficient chip, not a top notch performance chip.

 

[sefsefsefsef]

So will it only seek to compete against i3 and below?

i3 uses the same cores as i5, i7, and Xeon (well, except for having Turbo disabled ...). In this thread, people have only talked about the K12 *core*, not products that will use this core in varying numbers, like you seem to be suggesting. This was not a very effective anti-AMD jab.

 

[NTMBK]

I didn't miss, I just don't think it will be a competitor to Cannonlake in the same way Bulldozer was for Sandy Bridge. K12 will face a lot of cost constraints, because AMD cannot afford another wafer guzzler like their current big core chips, plus AMD is now with a much smaller R&D budget than they had with at the time they were developing their small core line.

Without another DEC to sell them IP and engineering teams, I wouldn't hold my breath for K12. K12 should be an extremely cost efficient chip, not a top notch performance chip.

Given that we have little hard info on Sky/Canyonlake and none whatsoever on K12, its a bit premature to declare winners and losers. Wait for more information.

 

[inf64]

i3 uses the same cores as i5, i7, and Xeon (well, except for having Turbo disabled ...). In this thread, people have only talked about the K12 *core*, not products that will use this core in varying numbers, like you seem to be suggesting. This was not a very effective anti-AMD jab.

Exactly, very well put sir .

 

[mrmt]

Given that we have little hard info on Sky/Canyonlake and none whatsoever on K12, its a bit premature to declare winners and losers. Wait for more information.

it's not. we cannot realistically expect AMD to be competitive with that R&D budget of them.

 

[NTMBK]

it's not. we cannot realistically expect AMD to be competitive with that R&D budget of them.

Yes, because R&D spending really made the difference for Itanium and Larrabee... It's an important factor to consider, but far from the only one.

 

[mrmt]

Yes, because R&D spending really made the difference for Itanium and Larrabee... It's an important factor to consider, but far from the only one.

Itanium has more revenue share than AMD server business.

 

[teejee]

Yes, because R&D spending really made the difference for Itanium and Larrabee... It's an important factor to consider, but far from the only one.

I agree, people in this forum often uses size of R&D to prove superiority. That is not a relevant parameter.
Sometimes smaller R&D is even better because it forces new solutions and new partnerships.

 

[Homeles]

I agree, people in this forum often uses size of R&D to prove superiority. That is not a relevant parameter.
Sometimes smaller R&D is even better because it forces new solutions and new partnerships.

This isn't an industry where you can make a startup in your garage. You need money, and obscene amounts of it to get anywhere.

 

[Fjodor2001]

This isn't an industry where you can make a startup in your garage. You need money, and obscene amounts of it to get anywhere.

Then how did ARM succeed?

 

[Homeles]

Then how did ARM succeed?

Because they're from an earlier era. And for what it's worth, they're the exception, not the norm. Silicon Valley is rather akin to the restaurant industry... for every business that succeeds, ten fail.

They also exploited a big gaping hole in computing... that low power hole is filled, and we're seeing the industry consolidation that happened in the 90s for x86, now for the smartphone market.

The problem that AMD's having right now is with money. It's actually very clear. Ever heard of the project management triangle? Idontcare's talked about it a few times.

You can either keep costs down, delay your time to market, or cut back on the quality of your product. You can pick two at most. AMD doesn't have a choice but to cut back on costs, and they've chosen to keep their time to market down. However, their quality has suffered. We've got 2, arguably 3, and potentially even four products (if Seronx is correct about Kaveri having disabled GPU and CPU logic) in recent history that are perfect examples of this.

Bulldozer needed another year in the oven before Vishera markedly improved over it. Trinity was a wash compared to Llano in terms of CPU performance (the arguable one I mentioned before), whereas Richland provided a fairly significant performance boost without even changing the silicon. Jaguar products were blown away by Puma. If Kaveri does in fact have Excavator cores and disabled GPU logic like Seronx has suggested, that'd be a fourth product to follow this trend.

AMD simply needs cash. You can't argue other otherwise, unless you're misinformed or intentionally ignoring AMD's precarious situation.

At this rate, if AMD makes it long enough to deliver K12, it's probably going to be another disappointing launch, followed by a later revision to significantly improve on it. Perhaps the console bucks they made last year could fuel a decent product, but they need to keep that cash flowing. AMD's got some wonderful talent, even after loosing a fair bit of it, but you need money to get the kind of manpower to keep up with your competitors.

 

[Sheep221]

Then how did ARM succeed?

ARM has been most of its existence unknown company that made mobile CPUs, they gained more attention only in last 3 years with mass deployment of android phones and tablets. So I assume they had to work very hard and earn shitloads of money over a time to be where they are now.

 

[NTMBK]

I agree, people in this forum often uses size of R&D to prove superiority. That is not a relevant parameter.
Sometimes smaller R&D is even better because it forces new solutions and new partnerships.

I wouldn't go quite that far- it's certainly a relevant parameter. "Lack of resources" is certainly one way that a project can fail.

 

[NTMBK]

Itanium has more revenue share than AMD server business.

Yes, Itanium is a massive success.

But let's look at other examples. What do you think cost the most to develop- Netburst, or Banias? Prescott or Yonah? Bobcat or Bulldozer? Throwing money and staff at a problem in no way guarantees you success. There are dozens of ways that a project can fail, and lack of resources is just one of them.

Please note that I am certainly not saying that AMD are guaranteed to come out on top in this fight. In fact I would say that the odds of it happening are pretty low. But the chance certainly exists, and we won't know until we actually see the results from all the relevant players. (Not just AMD and Intel...) It's not as simple as "Company A has a bigger pile of cash, hence it wins".

 

[teejee]

This isn't an industry where you can make a startup in your garage. You need money, and obscene amounts of it to get anywhere.

Classic strawman, nothing more to say.
Rule #2
http://www.craigbellamy.net/2013/06/24/10-commandments-of-rathional-debate/

 

[el etro]

AMD got too late into the efficiency run... They should focus the power consumption of the processors since first Fusion processor(2010)...

 

[NostaSeronx]

They should focus the power consumption of the processors since first Fusion processor(2010)...

AMD did focus on power consumption since the first fusion processors(Llano/Bobcat).

http://www.hotchips.org/wp-content/...23.17.111.Practical_PGandDV-Kosonocky-AMD.pdf
http://www.hotchips.org/wp-content/...p-CPUs/HC23.19.930-Llano-Fusion-Foley-AMD.pdf

 

[ShintaiDK]

The entry cost is so steep that there is no new companies. And its all due to R&D cost. And more and more established companies are falling from the list, rather than being added.

Lets see the top 10 2013 MPU revenue companies. ARM is not there. But for reference its from 1985, and the Holding company from 1990.

MediaTek 1997.
Broadcom 1991.
Spreadtrum 2001.
nVidia 1993.
AMD 1969.
Intel 1968.
Samsung Electronics (1938) 1969.
Apple 1976.
TI (1930) 1951.
Qualcomm 1985.
Freescale 2004. (But its a spinoff of Motorola 1928.)

The quick one would note that even with the chinese startup. There havent been a single new CPU company worth mentioning the last 13 years. And with 350M$ yearly revenue to reach spot number 10. One would think the bar of entry was low enough.

However, funny enough. Even in top10 of MPU revenue. They are already companies stating that they wont go below 28nm due to IC design costs. In short, their R&D budgets cant afford it.

R&D budget matters everything. And those that cant afford it are becoming ever faster irrelevant. You are not getting anything you didnt pay for in R&D.

 

[DeathReborn]

You must have missed x86 variant of K12 then. It's not "big core" per se but it is high-performance core. And it will focus more on perf./watt, like Core family.

The real question is... Is the K12 x86 core "High Performance" in relation to Intel or in relation to ARM? Based on AMD's recent history they'll say Intel and deliver ARM.

I really want AMD CPU's to succeed but they really need to pick up the pace or run the risk of going under before it launches.