Samsung Nintendo AMD semicustom ARM-x86 Showdown Scenario

[Gikaseixas]

The claims that Zen is going to be a design focused on low power makes no sense. AMD already has the cat cores for that, and they're quite successful if you count the console design wins. Of course Zen will have to take efficiency into account, but there would be no point in releasing something that isn't considerably better than the construction equipment cores in terms of performance. It just wouldn't be competitive.

Wishful thinking of Intel fanboys, nothing else. Nobody know's where AMD's projections are for Zen in terms of performance so people cannot present personal opinion as fact but still you see that everyday here. FUD is strong here.

I don't expect AMD to beat Intel but I hope they get close to Intel so we can get more choices for our builds, etc.

 

[mrmt]

Wishful thinking of Intel fanboys, nothing else. Nobody know's where AMD's projections are for Zen in terms of performance so people cannot present personal opinion as fact but still you see that everyday here. FUD is strong here.

What's the evidence you have pointing out for a high performance core, and how AMD could have a semi-custom business at all with this core?

 

[Gikaseixas]

What's the evidence you have pointing out for a high performance core, and how AMD could have a semi-custom business at all with this core?

and what evidence you have pointing out for a low performance core? None at all is what it is.

They recruited a few well known names to work both on the graphics and cpu sides so everything is open to failure or success. Point is, you guys will play Nostradamus again and predict future products prowess's based on past failures. Bulldozer failed, FX 5800 failed, HD 2900 failed, Pentium 4 failed but that didn't dictate future failures.

A very well known prediction is "no APU in the Xbone and PS4" guess what?

AMD confirmed that it is developing an all-new high-performance x86 micro-architecture code-named “Zen”... Now let's wait, give them the benefit of the doubt. All this hatred needs to end, every thread is the same thing

 

[mrmt]

and what evidence you have pointing out for a low performance core? None at all is what it is.

Here are the evidences pointing to a low performance core:

- Shrinking R&D budget.

- Smaller R&D teams

- Semi-custom business, which is a very cost-sensitive bracket.

- The fact that they are reusing IP from K12, an ARM core

- The fact that their foundry partner had only low power nodes by the time Zen was being conceived.

- The fact that clean sheet designs are pretty much hard to make. Incremental changes are the way the industry have been progressing in the last 15 years.

But it seems that you have discarded them all, but let's see for what.

AMD confirmed that it is developing an all-new high-performance x86 micro-architecture code-named “Zen”... Now let's wait, give them the benefit of the doubt. All this hatred needs to end, every thread is the same thing

It seems that all you have to offer is an statement of AMD managers, you know, the guys that have been busing screwing up the company in the last 10 years. Who has wishful thinking here?

 

[monstercameron]

Here are the evidences pointing to a low performance core:

- Shrinking R&D budget.

- Smaller R&D teams

- Semi-custom business, which is a very cost-sensitive bracket.

- The fact that they are reusing IP from K12, an ARM core

- The fact that their foundry partner had only low power nodes by the time Zen was being conceived.

- The fact that clean sheet designs are pretty much hard to make. Incremental changes are the way the industry have been progressing in the last 15 years.

But it seems that you have discarded them all, but let's see for what.



It seems that all you have to offer is an statement of AMD managers, you know, the guys that have been busing screwing up the company in the last 10 years. Who has wishful thinking here?

now yoy are putting us between a rock and a hard place, who to believe? amd's managers or a random internet commenter? decisions, decisions...
also why would the r&d budget map 1 to 1 with the output design?

 

[mrmt]

now yoy are putting us between a rock and a hard place, who to believe? amd's managers or a random internet commenter? decisions, decisions...
also why would the r&d budget map 1 to 1 with the output design?

Be my guest, find something better to believe. AMD management has plenty of credibility anyway, does it?

 

[monstercameron]

Be my guest, find something better to believe. AMD management has plenty of credibility anyway, does it?

More than you or I, that's for sure.

 

[ShintaiDK]

now yoy are putting us between a rock and a hard place, who to believe? amd's managers or a random internet commenter? decisions, decisions...
also why would the r&d budget map 1 to 1 with the output design?

Did AMDs managers specify in relation to what? Or are you just freely doing the guess yourself with a healthy dose of hopes mixed in?

How did Phenom beat Core 2 with 50% turn out? Or Bulldozer, the Intel killer?

R&D budgets never mattered right? All those clumsy successful growing companies spending more on R&D are just doing it all wrong.

 

[mrmt]

More than you or I, that's for sure.

Do you know that most of the points I made come from AMD and partners statements, but not marketing mumbo-jumbo like Keller's event, but communication that goes to investors and the SEC (e.g., management is criminally liable for these statements, unlike Keller's), don't you?

 

[MiddleOfTheRoad]

How did Phenom beat Core 2 with 50% turn out? Or Bulldozer, the Intel killer?

R&D budgets never mattered right? All those clumsy successful growing companies spending more on R&D are just doing it all wrong.

In concept form, Bulldozer should have been an Intel killer. AMD's management did have reason to be that optimistic about potential performance. It was the execution that failed them.... Designing an 8 core to compete against Intel's quad was a pretty good idea.

But those automated tools and a lousy fab partner really buried it in mediocrity.
http://www.guru3d.com/news-story/former-amd-engineer-talks-about-bulldozer.html

 

[swilli89]

Dude, where do you get your pot from?

Dude, where do you get your 1930's "reefer madness" view on weed from?

 

[mrmt]

In concept form, Bulldozer should have been an Intel killer. AMD's management did have reason to be that optimistic about potential performance. It was the execution that failed them.... Designing an 8 core to compete against Intel's quad was a pretty good idea.

Throwing more hardware than your competitor to solve the same problem? Yes, that's a *bright* idea.

 

[Gikaseixas]

Here are the evidences pointing to a low performance core:

- Shrinking R&D budget.

- Smaller R&D teams

- Semi-custom business, which is a very cost-sensitive bracket.

- The fact that they are reusing IP from K12, an ARM core

- The fact that their foundry partner had only low power nodes by the time Zen was being conceived.

- The fact that clean sheet designs are pretty much hard to make. Incremental changes are the way the industry have been progressing in the last 15 years.

But it seems that you have discarded them all, but let's see for what.



It seems that all you have to offer is an statement of AMD managers, you know, the guys that have been busing screwing up the company in the last 10 years. Who has wishful thinking here?

Dude, i rather believe people in the know then someone behind a desk whom wet dreams are AMD's terrible death.

Having smaller budget doesn't necessarily means "certain failure" and a clean sheet design is the best idea when Bulldozer is all you have as basis don't you agree?

 

[Gikaseixas]

Throwing more hardware than your competitor to solve the same problem? Yes, that's a *bright* idea.

And yet you stated that a clean sheet design is a terrible idea because the industry trend bla bla
Make up your mind

 

[gdansk]

I honestly think Apple is a more likely semi-custom customer than Samsung. Samsung already exited the laptop business, so there's not much reason to get a x86 semicustom.

Really? That's sad. I really like their not yet shipping NP930X2K.

 

[mrmt]

Having smaller budget doesn't necessarily means "certain failure"...

The only victory that matters is the commercial victory. That said, if Zen ends up pissing off all AMD fanboys on earth but is good enough to rake sales on the embedded market then it won't be a failure at all. This is where I think AMD is heading.

and a clean sheet design is the best idea when Bulldozer is all you have as basis don't you agree?

AMD does have solid IP from where to base its chips, K10 is a very nice foundation, the cat cores are also interesting. They don't have to stay with the Bulldozer family at all.

And yet you stated that a clean sheet design is a terrible idea because the industry trend bla bla
Make up your mind

Wow, I just said that Bulldozer was about throwing more hardware at the same problem Intel was trying to solve, meaning it is by definition less efficient than Intel chips, and that's a very bad idea.

 

[The Alias]

So however many years ago when bulldozer was chosen to be designed they forsook taking their existing architecture , giving it a die shrink, and applying whatever optimizations they could to add perf/IPC. In favor of an entirely new architecture. Fast forward a few years when they admitted internally that was a bad idea and returned to the drawing board, they had the same option. Now regardless of whether or not amd is planning to replace the cat cores or the bulldozer cores, they're most definitely going to outperform the previous generations core, because there are no radically forward looking quirks involved in the designs and you absolutely have to assume they're going to take what they learned from the previous smt cores, either the cat cores or k10, and apply it to zen. The only question is how much performance and efficiency will be gained. I'd have to think that it'd be pretty big considering a new architecture that is indeed smt like the previous core

 

[geoxile]

The only victory that matters is the commercial victory. That said, if Zen ends up pissing off all AMD fanboys on earth but is good enough to rake sales on the embedded market then it won't be a failure at all. This is where I think AMD is heading.



AMD does have solid IP from where to base its chips, K10 is a very nice foundation, the cat cores are also interesting. They don't have to stay with the Bulldozer family at all.



Wow, I just said that Bulldozer was about throwing more hardware at the same problem Intel was trying to solve, meaning it is by definition less efficient than Intel chips, and that's a very bad idea.

It's possible that between AMD's A57 and Armv8 ISA license they have access to some ARM IPs that they can learn from.

 

[JDG1980]

By semi-custom designs, I think AMD has in mind the next generation of consoles. That's where their biggest current design wins are, after all. For less demanding embedded applications, they're going to have trouble competing with ARM (and Intel is, of course, having the same problem). For in-between stuff like streamers and set-top boxes, they don't need a completely new design; the cat cores should be fine, and there's no reason they cannot receive further incremental design and process improvements. But the current generation of consoles are hurt by low clock rate and IPC, and Sony and Microsoft are going to want something better next time.

What AMD needs in the x86 arena is a high-IPC architecture that can work efficiently anywhere from 15W to 150W. That's what they are currently lacking and I think that's what Zen will aim to provide.

 

[mrmt]

What AMD needs in the x86 arena is a high-IPC architecture that can work efficiently anywhere from 15W to 150W. That's what they are currently lacking and I think that's what Zen will aim to provide.

AMD cannot afford this architecture.

 

[frozentundra123456]

So however many years ago when bulldozer was chosen to be designed they forsook taking their existing architecture , giving it a die shrink, and applying whatever optimizations they could to add perf/IPC. In favor of an entirely new architecture. Fast forward a few years when they admitted internally that was a bad idea and returned to the drawing board, they had the same option. Now regardless of whether or not amd is planning to replace the cat cores or the bulldozer cores, they're most definitely going to outperform the previous generations core, because there are no radically forward looking quirks involved in the designs and you absolutely have to assume they're going to take what they learned from the previous smt cores, either the cat cores or k10, and apply it to zen. The only question is how much performance and efficiency will be gained. I'd have to think that it'd be pretty big considering a new architecture that is indeed smt like the previous core

And I am sure they said the same thing about Bulldozer.

Assuming a big improvement is just as big a stretch as the naysayers who say it has to be terrible because of AMDs lack of resources. Bottom line is nobody really knows, and nobody even knows if it will come out on time.

 

[Asterox]

The shrinking R&D budget, the big semi-custom business, the focus on microservers, the sister ARM core, the focus on low power notebooks, and that's just the most evident factors.

AMD shrinking R & D budget is a thing of the past. This fact it should already be clear right, so why walk around with the eyes wide shut???

http://www.vrworld.com/2015/02/09/amds-real-china-play-strategic-investment/

"Chinese Academy of Sciences and the Institute of Computing Technology, is set to make key investments into AMD’s technologies and product lines.

These investments are set to begin immediately. Loongson’s targeted investments will be funded by a $19 billion war chest (120 billion RMB) that the company has set aside for such ventures."

 

[MiddleOfTheRoad]

Throwing more hardware than your competitor to solve the same problem? Yes, that's a *bright* idea.

Why the hell not? That's exactly what people do for GPU's -- If you want to play games well in 4K at high detail, good luck doing it with a single video card. SLI or Crossfire is pretty much the only way to get there.

 

[Arachnotronic]

AMD shrinking R & D budget is a thing of the past. This fact it should already be clear right, so why walk around with the eyes wide shut???

http://www.vrworld.com/2015/02/09/amds-real-china-play-strategic-investment/

"Chinese Academy of Sciences and the Institute of Computing Technology, is set to make key investments into AMD’s technologies and product lines.

These investments are set to begin immediately. Loongson’s targeted investments will be funded by a $19 billion war chest (120 billion RMB) that the company has set aside for such ventures."

$19 billion, how cute! Intel spent, what? Nearly $12 billion in R&D last year?

And don't think for a second that the Chinese Academy of Sciences will hand AMD its entire $19 billion "war-chest" to build server chips.

 

[MiddleOfTheRoad]

And I am sure they said the same thing about Bulldozer.

Assuming a big improvement is just as big a stretch as the naysayers who say it has to be terrible because of AMDs lack of resources. Bottom line is nobody really knows, and nobody even knows if it will come out on time.

Agreed -- there are many stages where they have to succeed. Even if Jim Keller delivers a monster design, Global Foundaries can still botch it by being late or fabbing it poorly. They have to execute well at every stage this time around.