[Techpowerup] AMD "Zen" CPU Prototypes Tested, "Meet all Expectations"

[Arachnotronic]

For me personally, I think it has to be a bit lower than that. Hard to say exactly how much lower, since it depends on the options available at the time. But if priced competitively enough, sure it'll benefit the customers. At least it won't be worth than what we have now with not much competition in the HEDT segment at all.

I don't think that AMD will price its 8 core Zen particularly cheaply and there is a very simple reason why.

AMD is basically on its last legs financially and the whole purpose of going after HEDT is because this is one of the few markets where customers are routinely willing to pay top dollar for performance. Unlike the general PC market it is actually a growth market.

If AMD comes in and gets cute with pricing, say offering an 8 core Zen at $350, and assuming it's actually as good as some of the AMD fans here think, then Intel would rightly and smartly respond by cutting the prices of its HEDT 8C/16T chips to around that price while bringing in some higher core count Xeons to fill in at the higher price points.

This would immediately eliminate the value proposition of the 8 core Zen for most customers (except of course the AMD loyal) and AMD winds up pretty much back where it started -- as the low cost option with relatively little market share.

Unlike Intel which can withstand seeing a dip in its gross profit margins on niche high end CPUs for a little bit as it deals with AMD, AMD will find itself unable to bring in enough revenue/profits in order to sustain itself and the company eventually goes under (given AMD's financials things need to get a lot better over the next year or two for the company to even have a chance of even being around in its current form).

tl;dr -- AMD can't really afford to play the price/value card because it simply doesn't have the financial wherewithal to do so.

 

[looncraz]

Better perf/watt than Skylake as one poster in this thread suggested. At this rate? The expectation ahead of launch will be that Zen will take the absolute performance crown, performance per watt crown, and be much cheaper to produce than any competing Intel chip.

I haven't seen anyone seriously claim that Zen would be close to Skylake. Even the most optimistic scenario puts it barely beyond Haswell, IPC-wise.

The fully objective math, which assumes an exact 40% boost over Excavator, is a dead tie with Haswell. With AMD claiming 40%, NO ONE should expect more. 2% less is twice as likely as 1% more.

I hope for Haswell, but plan for Sandy Bridge IPC.

Compared to Sandy Bridge, though, Zen is wider and more flexible, so it should be capable of beating it. Compared to Haswell, Zen should have lower store performance, identical integer performance, and FPU performance is nearly impossible to determine - it should be the same for legacy code, but worse with AVX.

I'm glad to see AMD focus on this route, they have a bad habit of planning for a future that doesn't materialize. Sometimes that comes out to their advantage, but often enough it only hurts them. Building a CPU to run the very code used in benchmarks better is the route Intel uses, AMD should do the same.

This is far more visible in their GPUs, though every GCN GPU is DirectX 12 compatible, so it finally paid off... sort of.

 

[Arachnotronic]

Apple is using Samsungs 14nm LPE, not LPP.

LPP is suited for more than Mobile and low clocks.

Do you have proof that A9 is 14nm LPE and not LPP?

 

[TechGod123]

I don't think that AMD will price its 8 core Zen particularly cheaply and there is a very simple reason why.

AMD is basically on its last legs financially and the whole purpose of going after HEDT is because this is one of the few markets where customers are routinely willing to pay top dollar for performance. Unlike the general PC market it is actually a growth market.

If AMD comes in and gets cute with pricing, say offering an 8 core Zen at $350, and assuming it's actually as good as some of the AMD fans here think, then Intel would rightly and smartly respond by cutting the prices of its HEDT 8C/16T chips to around that price while bringing in some higher core count Xeons to fill in at the higher price points.

This would immediately eliminate the value proposition of the 8 core Zen for most customers (except of course the AMD loyal) and AMD winds up pretty much back where it started -- as the low cost option with relatively little market share.

Unlike Intel which can withstand seeing a dip in its gross profit margins on niche high end CPUs for a little bit as it deals with AMD, AMD will find itself unable to bring in enough revenue/profits in order to sustain itself and the company eventually goes under (given AMD's financials things need to get a lot better over the next year or two for the company to even have a chance of even being around in its current form).

tl;dr -- AMD can't really afford to play the price/value card because it simply doesn't have the financial wherewithal to do so.

The earliest AMD can even die is Q1 2019. You're terribly pessimistic about Zen and AMD. I wonder why?...

 

[Arachnotronic]

The earliest AMD can even die is Q1 2019. You're terribly pessimistic about Zen and AMD. I wonder why?...

Oh, I don't know, because AMD the company is in a classic death spiral and AMD's products have been losing market share for years now. Yeah, nothing at all founded in my pessimism for one of the worst run companies in advanced semiconductor design today. Nothing at all

This company, IMHO, is rotten to the core with executives that are, in my opinion, just trying to stuff their pockets with as much cash as possible before the bitter end comes. Instead of laying off all of those employees who have probably been overworked given all of the previous layoffs recently to save a couple of million bucks, maybe the folks at the top might want to cut back on the lavish bonuses that they're paying themselves? Maybe just a thought?

 

[TechGod123]

Do you have proof that A9 is 14nm LPE and not LPP?

This is ridiculous now.

"What's unusual in this case, however, is that Samsung is using its 14nm LPE (Low Power Early) FinFET process, which it also uses for the Exynos 7420 SoC, while TSMC is using its own 16nm FinFET process, resulting in two different versions of the A9 SoC with different die sizes." - http://www.tomshardware.com/news/iphone-6s-a9-samsung-vs-tsmc,30306.html

 

[Sweepr]

Oh, I don't know, because AMD the company is in a classic death spiral and AMD's products have been losing market share for years now. Yeah, nothing at all founded in my pessimism for one of the worst run companies in advanced semiconductor design today. Nothing at all

Shhh, they are ready to call you a fanboy now that they know you don't buy the hype.

 

[looncraz]

I am expecting Haswell level performance per clock, but lower peak frequency and far fewer power management features. It may prove to be a competent low cost alternative to Intel products in desktops but I don't think it will be impressive in notebooks and other perf/watt sensitive use cases.

AMD already has most of Intel's power management features, I'd expect parity in that regard. As far as frequency, I expect lower only because it is AMD's first shot at Zen on a new process. In theory, there is nothing preventing a 4.5GHz or higher Zen CPU on 14nm LPP.

However, your assessment about mobile is the exact opposite of what I expect. 14nm LPP has shockingly low leakage and AMD has done amazingly well just with Cairo. At 40% higher IPC, a new process, and drop the voltage and clocks to the 14nm LPP's sweet spot, and it is conceivable that Zen-based mobile solutions could have better perf/watt than anything Intel offers.

 

[Abwx]

Do you have proof that A9 is 14nm LPE and not LPP?

(l'Apple A9 utilisant le 14nm LPE – Low Power Early)

http://www.hardware.fr/news/14401/amd-valide-14nm-lpp-glofo.html

 

[TechGod123]

Oh, I don't know, because AMD the company is in a classic death spiral and AMD's products have been losing market share for years now. Yeah, nothing at all founded in my pessimism for one of the worst run companies in advanced semiconductor design today. Nothing at all

Yet you refuse to look at how AMD is staged to make a comeback. All I have been reading in your posts is how Zen will fail adn how there's no way it will be competitive. Then you couldn't believe that A9 used LPE as opposed to LPP because otherwise it would make your ridiculous posts look even worse. I might be done with this thread soon if you keep this up.

(What am I kidding, your posts are quite entertaining so I might as well just keep replying)

 

[looncraz]

If AMD prices its 8 core Zens at $899 and it delivers 8-core Haswell-E performance, will this bring the benefit to the consumers that you hope for?

If AMD prices their 8 core CPUs equally to Intel's similarly performing CPUs, we still have the advantage of using the cheaper AMD platform on which to install those CPUs :thumbsup:

There are quite a few advantages to a unified socket for consumer CPUs.

The breadth of performance available on the same motherboard could be quite impressive. Start with the lowest Excavator-based APU then work your way all of the way up to the 8C/16T Zen-based CPU.

Same RAM, same motherboard... just change out the CPU. THAT's a nice upgrade path!

 

[TechGod123]

Shhh, they are ready to call you a fanboy now that they know you don't buy the hype.

What the hell? Its painfully OBVIOUS he's extremely biased. This isn't about hype. Some of us have REASONABLE expectations about it and we believe it won't be a flop like some Intel shills are making it out to be.

 

[Arachnotronic]

This is ridiculous now.

"What's unusual in this case, however, is that Samsung is using its 14nm LPE (Low Power Early) FinFET process, which it also uses for the Exynos 7420 SoC, while TSMC is using its own 16nm FinFET process, resulting in two different versions of the A9 SoC with different die sizes." - http://www.tomshardware.com/news/iphone-6s-a9-samsung-vs-tsmc,30306.html

Is there any actual proof of this or is this just an assumption on the writer's part? It seems strange that Apple would use TSMC 16FF+ for the A9 while using "only" 14nm LPE for the Samsung-built A9s.

 

[Arachnotronic]

Yet you refuse to look at how AMD is staged to make a comeback. All I have been reading in your posts is how Zen will fail adn how there's no way it will be competitive. Then you couldn't believe that A9 used LPE as opposed to LPP because otherwise it would make your ridiculous posts look even worse. I might be done with this thread soon if you keep this up.

(What am I kidding, your posts are quite entertaining so I might as well just keep replying)

Believe it or not, I actually don't care whether you continue to discuss this topic with me. I am complying with the TOS of this message board and am giving my opinion which I think is of value to at least some of the people reading this thread. If you don't like what I have to say that's fine; my life doesn't exactly revolve around trying to win the approval of people on this message board.

 

[TechGod123]

If AMD prices their 8 core CPUs equally to Intel's similarly performing CPUs, we still have the advantage of using the cheaper AMD platform on which to install those CPUs :thumbsup:

There are quite a few advantages to a unified socket for consumer CPUs.

The breadth of performance available on the same motherboard could be quite impressive. Start with the lowest Excavator-based APU then work your way all of the way up to the 8C/16T Zen-based CPU.

Same RAM, same motherboard... just change out the CPU. THAT's a nice upgrade path!

This so much. The unified socket will be glorious. Pick up a decent mobo and you'll have great upgrade paths available. Way better than paying more for an X99 mobo or one that is designed for OCing or one that has these extra features that the other ones don't have because of different chipsets etc...This immediately makes Zen better for me if it matches Haswell.

That, and Zen+ will most likely use the AM4 socket as well which means an even bigger upgrade path.

 

[TechGod123]

Is there any actual proof of this or is this just an assumption on the writer's part? It seems strange that Apple would use TSMC 16FF+ for the A9 while using "only" 14nm LPE for the Samsung-built A9s.

Did you not hear of the big chipgate controversy?

"One of the immediate ramifications of dual sourcing is that the die sizes of the A9s are different. The A9 produced by Samsung on their 14nm FinFET Process is the smaller of the two, at 96mm2. Meanwhile the A9 produced on TSMC’s 16nm FinFET process is 104.5mm2, making it about 9% larger." - Source

 

[monstercameron]

Wow. Intel has literally been spending billion$ on trying to eke out all of the performance/watt it can from each architecture and you think AMD is going to go from the lousy Bulldozer lineage to a new architecture built on a relatively shoestring budget on a foundry process optimized for mobile products and wind up with better perf/watt in key PC/server use cases? OK.

You responses used to be more analytical, what happened?

 

[Arachnotronic]

AMD already has most of Intel's power management features, I'd expect parity in that regard. As far as frequency, I expect lower only because it is AMD's first shot at Zen on a new process. In theory, there is nothing preventing a 4.5GHz or higher Zen CPU on 14nm LPP.

However, your assessment about mobile is the exact opposite of what I expect. 14nm LPP has shockingly low leakage and AMD has done amazingly well just with Cairo. At 40% higher IPC, a new process, and drop the voltage and clocks to the 14nm LPP's sweet spot, and it is conceivable that Zen-based mobile solutions could have better perf/watt than anything Intel offers.

I respect your viewpoint even though I don't agree with it; I appreciate the civility in your response.

We'll see in a year how your predictions hold up to what ultimately makes it to the shelves.

 

[TechGod123]

You responses used to be more analytical, what happened?

...

 

[Ferzerp]

What the hell? Its painfully OBVIOUS he's extremely biased. This isn't about hype. Some of us have REASONABLE expectations about it and we believe it won't be a flop like some Intel shills are making it out to be.

Using the recent past performance of a company to form expectations of what type of product they will release makes one a "shill"?

How do you figure?

Which is magical thinking? Expecting a company's to perform similarly to how they have been for the last 6 years or so, or expecting some massive improvement despite fewer resources.

 

[AtenRa]

And if you want to compare 14nm LPP against 28nm HPP

 

[Abwx]

Is there any actual proof of this or is this just an assumption on the writer's part? It seems strange that Apple would use TSMC 16FF+ for the A9 while using "only" 14nm LPE for the Samsung-built A9s.

It means that Samsung LPE is competitive with TSMC s 16FF+...

 

[Abwx]

And if you want to compare 14nm LPP against 28nm HPP

More numbers...



GLOBAL FOUNDRIES FINFETS VS 28NMs

PPA RESULT : LVT


GF PROCESS 28 SLP 28 HPP 14 LPP



SPEED (SS.0P90V.WORST 125C/-40C) (SS.0P765V.WORST 125C/-40C) (SS.0P72V.WORST 125C/-40C)


FMAX GHZ 0.97 1.17 2.41

RELATIVE SPEED 1 1.2 2.48





POWER (FF.1P10V.125C) (FF.0P935V.125C) (FF.0P88V.125C)


TOTAL DYNAMIC 158 210 310
POWER (mW)


RELATIVE DYN 1 1.3 1.9
POWER

TOTAL LEAKAGE 70 119 18.6
POWER (mW)

RELATIVE LEAK 1 1.7 0.27







PPA RESULTS : RVT


GF PROCESS 28 SLP 28 HPP 14 LPP



SPEED (SS.0P90V.WORST 125C/-40C) (SS.0P765V.WORST 125C/-40C) (SS.0P72V.WORST 125C/-40C)


FMAX GHZ 0.79 0.94 1.81

RELATIVE SPEED 1 1.2 2.30





POWER (FF.1P10V.125C) (FF.0P935V.125C) (FF.0P88V.125C)


TOTAL DYNAMIC 100 139 264
POWER (mW)

RELATIVE DYN 1 1.39 2.64
POWER

TOTAL LEAKAGE 16 28 3.9
POWER (mW)

RELATIVE LEAK 1 1.75 0.24

 

[Arachnotronic]

You responses used to be more analytical, what happened?

I like to think that my posts are fairly analytical, but obviously as a judge of my own posts I am not exactly impartial

The bottom line is that the expectations around Zen and Zen-based products seem to be inflating with each and every day.

All we know for sure about Zen is that AMD has said that perf/clock goes up by around 40% relative to Excavator and that AMD expects Zen to make the company much more competitive than it has been over the last several years in high performance processors.

Anything beyond that is basically (informed?) speculation.

The original point that I was trying to make in the text that you quoted is that high IPC/high frequency x86 CPUs are really, really hard to design and build. The reason we see 5-10% IPC improvement per generation from Intel is because wringing out those gains is actually super tough and R&D intensive, especially since Intel needs to make these IPC boosts while keeping frequencies high and keeping power efficiency in check.

If it is proving tough for Intel to scale these new performance heights, I have to maintain a fairly high degree of skepticism about AMD's ability to deliver a design in basically one shot that effectively erases the perf/clock gap that has long existed between Intel's and AMD's top designs while at the same time delivering high clocks, and great power efficiency.

We will see if AMD ultimately delivers but from every bit of information available to me in the public domain, I can't exactly be enthusiastic that Zen will be what some of these posters hope. That's all.

 

[monstercameron]

I like to think that my posts are fairly analytical.

The bottom line is that the expectations around Zen and Zen-based products seem to be inflating with each and every day.

All we know for sure about Zen is that AMD has said that perf/clock goes up by around 40% relative to Excavator and that AMD expects Zen to make the company much more competitive than it has been over the last several years in high performance processors.

Anything beyond that is basically (informed?) speculation.

The original point that I was trying to make in the text that you quoted is that high IPC/high frequency x86 CPUs are really, really hard to design and build. The reason we see 5-10% IPC improvement per generation from Intel is because wringing out those gains is actually super tough and R&D intensive, especially since Intel needs to make these IPC boosts while keeping frequencies high and keeping power efficiency in check.

If it is proving tough for Intel to scale these new performance heights, I have to maintain a fairly high degree of skepticism about AMD's ability to deliver a design in basically one shot that effectively erases the perf/clock gap that has long existed between Intel's and AMD's top designs while at the same time delivering high clocks, and great power efficiency.

We will see if AMD ultimately delivers but from every bit of information available to me in the public domain, I can't exactly be enthusiastic that Zen will be what some of these posters hope. That's all.

OK I understand, I just started to go through that thread again, just found that comment a bit out of character.