Zen hasn't taped out yet

Page 9 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
Lol, ABWX claims to know about Intel's manufacturing processes, yet doesn't even know their name.

Priceless.
 

Dresdenboy

Golden Member
Jul 28, 2003
1,730
554
136
citavia.blog.de
If it has higher than 40% uplift over Excavator, why didn´t AMD say so. It´s not like they can afford to make any understatements...
As said, that average number (and based on simulations anyway), doesn't apply equally to any benchmark. Take 2 SPECCPU results of different uarchs with one being 40% higher and have a look at the variance of the sub benchmarks.

So, back to Llano ??? 2x Llano cores were almost the same as a single SandyBridge. But this time "Llano" has 40% increase of IPC and SMT.

Could they make such a wide core with 4x ALUs, 2x AGUs and FPU unit at 4-5mm2 on the 14nm FF ???

And if ZEN core is only 4-5mm2 they could make a 8x Core 16x Threads die at 120-140mm2 or even smaller (depending on the L3 cache size and hyper-transports or equivalent).
Llano vs. Sandy Bridge was about old vs. new feature set and lots of uarch improvements (unified scheduler, uOp cache, SMT, 4-wide decode, etc.).

I measured Apples Twister (A9) on the 14nm Samsung die:
cpu block: ~13 mm²
3MB L2 ~3.5 mm²
1 core (with 2x64kB L1 caches): ~4.3 mm²
See http://www.anandtech.com/show/9686/the-apple-iphone-6s-and-iphone-6s-plus-review/3

And the patch data looks like they used Cat core like multiplier (rectangular, not full 1/cycle throughput at DP). One more indicator of area optimization, like 128b data paths, split FMAC, 2 AGUs.


So do you think AMD could cram enough hardware on 4.5mm^2 to be competitive in terms of performance and efficiency against a 8mm^2 Skylake core on integer performance, meaning that Intel is basically doubling the size of the core just to get better FP performance?
The exact size also depends on the libraries they use. My first estimations were based on scaled down XV components with an high performance library compensation factor.
It could be close to 6 mm^2, but then it starts to get unlikely.

They might simply have sacrificed some FP performance scenarios (AVX 256b), exactly that, what still takes ~1/4 of Intel's 256b AVX cores.

Have a look at this well known comparison by Hans de Vries:


The execution units in a Haswell core (dark rectangle in the upper left corner) are about as big as a complete Jaguar core, even while being one node further.

I've already posted an Intel paper, where they show, that while being slightly active (low IPC), such a big HSW core uses already about 75% of the power it uses at full performance. That means, that activating all that logic comes with a significant fixed cost.
 

looncraz

Senior member
Sep 12, 2011
722
1,651
136
This plan sounds like a fast death sentence for AMD ,in other words, BD no 2. Why go with many small cores plus SMT when you can have similar MT performance but better ST performance If you go with a moderate amount of wider cores+SMT?

A physically small core, I believe, is the discussion. Zen is definitely a wide core, wider even than Haswell.
 

mahoshojo

Junior Member
Jul 24, 2015
18
0
36
What a ridiculous argument. 8 core Broadwell-E will probably slot nicely into the $600 price point, so it's not an "either or" between your two scenarios.

I get that you really want AMD to succeed, but you shouldn't assume that the people determining which SKUs to offer at Intel at complete morons.

How much does it cost for 14nm 100mm^2, 150mm^2, 200mm^2 cpu chips?
Intel had been enjoying the fat margin on those high end parts.
People buy them because of no competition.
 

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
How much does it cost for 14nm 100mm^2, 150mm^2, 200mm^2 cpu chips?
Intel had been enjoying the fat margin on those high end parts.
People buy them because of no competition.

Is Intel price gouging? What makes you say that anyway? Intel sold the i7 920 at $280 when released. The i7 6700K was $320. Given that was 7 years ago, that's pretty darn close. It's AMD that has been lowing the prices, due to lack of value, that makes you think Intel is price gouging. And don't bring up the retailers who are upping the price due to high demand, that was not Intel's price point.
 
Mar 10, 2006
11,715
2,012
126
Is Intel price gouging? What makes you say that anyway? Intel sold the i7 920 at $280 when released. The i7 6700K was $320. Given that was 7 years ago, that's pretty darn close. It's AMD that has been lowing the prices, due to lack of value, that makes you think Intel is price gouging. And don't bring up the retailers who are upping the price due to high demand, that was not Intel's price point.

How quickly people forget that AMD tried to launch the piece of junk FX 9590 for Intel Extreme Edition prices...
 

looncraz

Senior member
Sep 12, 2011
722
1,651
136
Is Intel price gouging? What makes you say that anyway? Intel sold the i7 920 at $280 when released. The i7 6700K was $320. Given that was 7 years ago, that's pretty darn close. It's AMD that has been lowing the prices, due to lack of value, that makes you think Intel is price gouging. And don't bring up the retailers who are upping the price due to high demand, that was not Intel's price point.


Nobody said anything about gouging, but CPU margins have been increasing for Intel, rather than remaining steady. Good for profit, not so good for the consumer pricing. However, I'm actually happy they have done this as it allows room for AMD to build a product line.

Intel would have been foolish to drop the prices as their margin increased, they had the luxury of working on a repeating pattern of product launches, stagnant prices, and then price reductions only as new products of theirs arrive. This pricing structure protects them on several fronts, as well as protecting margin in case they have to use a more expensive process next time around.
 
Mar 10, 2006
11,715
2,012
126
Nobody said anything about gouging, but CPU margins have been increasing for Intel, rather than remaining steady.

Intel has seen ASPs rise as people buy a richer mix of products, but I don't think your statement that CPU margins have been rising is really accurate.
 

Jovec

Senior member
Feb 24, 2008
579
2
81
We are years past the point where new nodes mean lower production costs. I doubt Intel margins vary more than +/- 5%. I'd imagine Intel's labor costs are also quite a bit higher than AMDs with respect to things like the number of people on teams, doing QA, etc.
 
Last edited:

AtenRa

Lifer
Feb 2, 2009
14,003
3,361
136
Intel Nehalem Core i7 920 November 2008
Fab process = 45nm
Die size = 296mm2
Transistor Count = 731M
Price = $284

Intel Sandybridge Core i7 2600K January 2011
Fab process = 32nm
Die size = 216mm2
Transistor Count = 995M
Price = $317

Intel Ivybridge Core i7 3770K April 2012
Fab process = 22nm
Die size = 160mm2
Transistor Count = 1400M
Price = $313

Intel Skylake Core i7 6700K August 2015
Fab process = 14nm
Die size = 120mm2
Transistor Count = 2000M ??
Price = $350

According to Moore Law and Intel's claims of cheaper transistors with every new fab process, taking Nehalem as the start up, we should had

45nm Nehalem 731M transistor die for $285

32nm SandyBridge 1400M transistor die for $285
22nm Ivybridge 2800M transistor die at $285
14nm Skylake 4800M transistor die at $285

What Intel got,

Nehalem 45nm = $284 / 731M Transistors = $0,388 per transistor
SandyBridge 32nm = $317 / 995M Transistors = $0,318 per transistor
IvyBridge 22nm = $313 / 1400M Transistors = $0,223 per transistor
Skylake 14nm = $350 / 2000M ?? Transistors = $0,175 per transistor

So at the end there are two things that happen, either Intel got higher margins because of the lower transistor cost and smaller die sizes with each generation.
Or, Moore law is broken financially and Intels margins are the same as before at 45nm and the reason they make smaller and smaller dies with each generation.

I dont say which of the two is true (perhaps both) cause i dont have the data, so ill leave it to you.
 

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
The exact size also depends on the libraries they use. My first estimations were based on scaled down XV components with an high performance library compensation factor.
It could be close to 6 mm^2, but then it starts to get unlikely.

They might simply have sacrificed some FP performance scenarios (AVX 256b), exactly that, what still takes ~1/4 of Intel's 256b AVX cores.

So according to you basically in terms of raw performance it's a very long shot, and in terms of efficiency they do have a shot because of the trade offs Intel made in terms of FP performance. Still doesn't put AMD in a good place on the server market, does it?

From what you are saying here they will not be able to get much, if anything on servers but at least will move to being badly beaten on desktop, an improvement from the utterly ruinous situation they are now.

Really, Zen as a product family is looking more and more like Bulldozer for me. Server optimized IP being sold on the consumer market, but at least this time they are not going against fundamental markets trends (efficiency), meaning that it won't be such an unmitigated failure.
 

witeken

Diamond Member
Dec 25, 2013
3,899
193
106
According to Intel, the price per transistor drops 30% or 1.43x per generation.

So these things are wrong with your calculation:

- You blindly assume that every generation is 2x as dense, while 2x, while it is probably closer to 1.9x or 1.8x.
- You don't take into account the increasing wafer cost; something like 30% at 14nm.

Things like yield and R&D costs (which grow 30% per node) also change the economics. For Skylake, the bad 14nm yields obviously impact its cost.
 

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
According to Intel, the price per transistor drops 30% or 1.43x per generation.

This is a dead end discussion. To get how much Intel would be improving each generation you would need *a lot* of market data such as mercury report and *a lot* of time dwelling on financial statements because what matters is Intel cost, not Intel ASP.

Anyone trying to infer how much Intel is getting from each node through ASP is either ignorant or outright dishonest.
 
Apr 30, 2015
131
10
81
ARM are developing physical IP with Global Foundries, at 10nm and 7 nm; I wonder if this would be available to AMD? It would be interesting to see AMD processors on ARM 10nm physical IP; how would it compare with Intel's designs, on their own process?
 

Boze

Senior member
Dec 20, 2004
634
14
91
Well if price does not matter and you want max MT performance, then you should be looking at the Intel Xeon series. Perhaps the 18C/36T E7-8890v3 at "just" $7174.

Seriously, I guess even you have some price sensitivity.

Also, Zen is 2016Q4, so it'll be competing with Broadwell-E, not KabyLake-E or CannonLake-E which will be much later.

So just a hypothetical example:

-Broadwell-E performs 10-20% better than Zen
-Zen costs $600
-Broadwell-E costs $1000

Which one would you pick?

I should only look at the E7-8890 v3 if it outperforms other processors for my mixed usage scenario. If that processor outperforms a Broadwell-E in gaming, I'll buy it. Currently, no one bothers to even benchmark those because they don't, though.

And I would buy the Broadwell-E. 20% more performance is ENORMOUS, and there's nowhere else in the system I could make up that performance deficit for $400. At least no place that I know of... I already have 3 GTX 980 Ti cards, so would adding a 4th make that up for $400? First I'd have to find a $400 one, which isn't likely, unless its used, maybe.

And no, price is not an issue for me, at all. You're talking to someone who ordered a Storinator S45 with 45 WD Red 6 TB drives. Total price was around $27000. And its sole purpose in life is to store television and movies for me. So let me repeat.

Price is not an issue.
 

Fjodor2001

Diamond Member
Feb 6, 2010
3,926
404
126
Ok, I see. Well in that case I don't think 8 core Zen is what you're looking for. And you should maybe even be looking at something with more performance than the Intel HEDT series too.

Having said this, I think there are lots of other more price sensitive buyers out there for whom 10-20% extra performance will not justify spending several hundred extra dollars. So assuming Zen performance is reasonable close to Intel HEDT and priced noticeable lower at the time it is released, I think Zen will find its market.
 
Last edited:

dark zero

Platinum Member
Jun 2, 2015
2,655
138
106
I should only look at the E7-8890 v3 if it outperforms other processors for my mixed usage scenario. If that processor outperforms a Broadwell-E in gaming, I'll buy it. Currently, no one bothers to even benchmark those because they don't, though.

And I would buy the Broadwell-E. 20% more performance is ENORMOUS, and there's nowhere else in the system I could make up that performance deficit for $400. At least no place that I know of... I already have 3 GTX 980 Ti cards, so would adding a 4th make that up for $400? First I'd have to find a $400 one, which isn't likely, unless its used, maybe.

And no, price is not an issue for me, at all. You're talking to someone who ordered a Storinator S45 with 45 WD Red 6 TB drives. Total price was around $27000. And its sole purpose in life is to store television and movies for me. So let me repeat.

Price is not an issue.
And how about a 10% performance lose against Intel on stock.... but with 10% more overclocking potential?
 

Boze

Senior member
Dec 20, 2004
634
14
91
Ok, I see. Well in that case I don't think 8 core Zen is what you're looking for. And you should maybe even be looking at something with more performance than the Intel HEDT series too.

There's nothing better for gaming right now. There's a such thing as throwing good money after bad, and that's what I'd be doing if I, say, went with a dual processor LGA 2011-3 motherboard.

Two Broadwell-E processors might give me 20 total cores and 40 threads, but that doesn't mean anything if my games can't take advantage of it.

My upgrades are always based on what kind of benefit I'll say for gaming, primarily, then video editing. At some point, you're just saying you have bigger better badder stuff to dick measure.

I'm not interested in that. I built a 270 TB storage array because I legitimately believe that it will last me for the next 5-10 years of Blu-ray ripping, not because I wanted to say that I have 270 raw TB of storage.

Having said all this, I would be ecstatic if benchmarking showed that a dual processor -E series solution actually improved FPS in games, but I'm not hopeful that DirectX 12 is going to be able to make use of even 8 cores, much less 10 to 20.
 

mysticjbyrd

Golden Member
Oct 6, 2015
1,363
3
0
I should only look at the E7-8890 v3 if it outperforms other processors for my mixed usage scenario. If that processor outperforms a Broadwell-E in gaming, I'll buy it. Currently, no one bothers to even benchmark those because they don't, though.

And I would buy the Broadwell-E. 20% more performance is ENORMOUS, and there's nowhere else in the system I could make up that performance deficit for $400. At least no place that I know of... I already have 3 GTX 980 Ti cards, so would adding a 4th make that up for $400? First I'd have to find a $400 one, which isn't likely, unless its used, maybe.

And no, price is not an issue for me, at all. You're talking to someone who ordered a Storinator S45 with 45 WD Red 6 TB drives. Total price was around $27000. And its sole purpose in life is to store television and movies for me. So let me repeat.

Price is not an issue.
And how about a 10% performance lose against Intel on stock.... but with 10% more overclocking potential?


He is such an outlier that his purchasing habits aren't relevant.
 

mysticjbyrd

Golden Member
Oct 6, 2015
1,363
3
0
There's nothing better for gaming right now. There's a such thing as throwing good money after bad, and that's what I'd be doing if I, say, went with a dual processor LGA 2011-3 motherboard.

Two Broadwell-E processors might give me 20 total cores and 40 threads, but that doesn't mean anything if my games can't take advantage of it.

My upgrades are always based on what kind of benefit I'll say for gaming, primarily, then video editing. At some point, you're just saying you have bigger better badder stuff to dick measure.

I'm not interested in that. I built a 270 TB storage array because I legitimately believe that it will last me for the next 5-10 years of Blu-ray ripping, not because I wanted to say that I have 270 raw TB of storage.

Having said all this, I would be ecstatic if benchmarking showed that a dual processor -E series solution actually improved FPS in games, but I'm not hopeful that DirectX 12 is going to be able to make use of even 8 cores, much less 10 to 20.

So wait, you have an assload of money to throw away on hardware, but pirate movies and software?
 

Boze

Senior member
Dec 20, 2004
634
14
91
So wait, you have an assload of money to throw away on hardware, but pirate movies and software?

No, I legitimately buy the Blu-rays, but I hate having to load discs, so I rip them all to .MKV for use with PLEX Media Server.
 

tential

Diamond Member
May 13, 2008
7,355
642
121
So wait, you have an assload of money to throw away on hardware, but pirate movies and software?

Lol, a lot of people fall into this. Just because you have money to throw away doesn't mean you won't save a penny wherever humanly possible.

Have you not seen the many many ways company's/those with wealth cheat the system to save a nickel?
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |