AMD zen pre-release thread!

Page 3 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
K12 and Zen are the 2 CPUs for 2016+ (Sharing the same platform).

And the targetted segments are listed to the right.

The 2015 x86 core is Puma+ in case anyone would be in doubt:


This also tells what Zen will be a successor to.
 
Last edited:

Abwx

Lifer
Apr 2, 2011
11,783
4,691
136
K12 and Zen are the 2 CPUs for 2016+ (Sharing the same platform).

And the targetted segments are listed to the right.

The 2015 x86 core is Puma+ in case anyone would be in doubt:


This also tells what Zen will be a successor to.

Only thing that is doubtless is that you re in a total confusion, Zen an K12 have nothing to do with Sybridge and this slide...

Dr. Su said she expects the first example of SkyBridge products to target embedded systems and "some client markets," presumably low-power devices like tablets or small laptops. These products will use either the "Puma+" CPU cores that AMD just shipped in its Mullins APU or a low-power-optimized version of the Cortex-A57 core licensed from ARM




AMD revealed that it is working on not one, but two brand-new, built-from-scratch CPU architectures. It has licensed the ARMv8 ISA and, in Dr. Su's words, "we are already well on our way to developing custom ARM cores." At the same time, AMD is building a brand-new, x86-compatible CPU core that will serve to replace Bulldozer and its lineage.

Since we're talking about a replacement for Bulldozer, this is an entirely different class of beast from the "Puma+" and Cortex-A57 cores in the first SkyBridge parts. AMD's execs noted that these are high-frequency, high-performance CPU cores that will span the range from laptops to desktops to servers—and not just "microservers.

http://techreport.com/review/26418/amd-reveals-k12-new-arm-and-x86-cores-are-coming
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Only thing that is doubtless is that you re in a total confusion, Zen an K12 have nothing to do with Sybridge and this slide...

Zen and K12 is the successors as I wrote. However Puma+ is what Zen lifts from. And for K12 its A57 that it succeeds.
 

Abwx

Lifer
Apr 2, 2011
11,783
4,691
136
However Puma+ is what Zen lifts from. And for K12 its A57 that it succeeds.

Not at all, they are not using Puma cores as basis for Zen or A57s for K12 respectively, the low power cores will have their own evolutions.

Zen, and hence K12, will use whatever parts from any CPUs they currently have but the designs are clearly high performance cores and they will target an Excavator module throughput but in a single core, whith eventualy a two threads per core capability.
 

JDG1980

Golden Member
Jul 18, 2013
1,663
570
136
No transactional memory or AVX2 ?!

Where was this said? According to the Phoronix article, the dropped instruction sets are as follows: TBM, FMA4, XOP, and LWP. All of these are AMD-specific instructions, not included on Intel processors.

It's possible that AMD may skip adding AVX2 and/or TSX instructions to Zen, but I'm not aware of any reliable reports specifically claiming this. Incidentally, keep in mind that TSX doesn't work on current Intel processors either; the silicon is there, but it's bugged and thus disabled in microcode. Only the next generation of Broadwell will have TSX, and it's a safe bet that only a small number of applications (most of them server-side) will take advantage of this. Whether AMD includes TSX or not will depend on how much they want to compete with Intel in the server market.

It is a little early for a pre-release thread, though I find news of them dumping FMA4 and XOP to be a little disheartening. Just a little.

Why? Both of these instruction sets are AMD-only; they were never included on any Intel CPU. Therefore, pretty much nothing is going to rely upon them. Furthermore, as they are complicated operations which have 4 operands apiece (as opposed to three for FMA3), there's a good chance they contributed to the die size bloat of Bulldozer and its derivatives. From an efficiency standpoint, cutting them is the right decision.
 

ThatBuzzkiller

Golden Member
Nov 14, 2014
1,120
260
136
Where was this said? According to the Phoronix article, the dropped instruction sets are as follows: TBM, FMA4, XOP, and LWP. All of these are AMD-specific instructions, not included on Intel processors.

It's possible that AMD may skip adding AVX2 and/or TSX instructions to Zen, but I'm not aware of any reliable reports specifically claiming this. Incidentally, keep in mind that TSX doesn't work on current Intel processors either; the silicon is there, but it's bugged and thus disabled in microcode. Only the next generation of Broadwell will have TSX, and it's a safe bet that only a small number of applications (most of them server-side) will take advantage of this. Whether AMD includes TSX or not will depend on how much they want to compete with Intel in the server market.

Transactional memory has tons of applications so getting it ASAP is important and it's the key to multithreaded performance scaling ...
 

itsmydamnation

Diamond Member
Feb 6, 2011
3,023
3,786
136
Who cares what Bethesda does ...

exactly it is because they are completely average.... which was my point.

Transactional memory has tons of applications so getting it ASAP is important and it's the key to multithreaded performance scaling ...

no it isn't...... people need to write code that's multithreads well, Transactional memory helps with the task, it doesn't do it for you.
 

beginner99

Diamond Member
Jun 2, 2009
5,312
1,750
136
Let's say AMD manages to push out that Zen in 2016 in 16nmFF as a solid 95W 8core/16 thread design with singlethreaded performance of Intels chips a la 4XXXK series.

Yeah I will call that a huge success if Zen matches Haswell in IPC. Considering that Bulldozer and successors have about half the IPC of Haswell makes this however very, very unlikely.

People keep saying that FX sucks...I mean sure...NOW it does. But when their 8 cores were new...Intel had no 8 threaded alternative that was actually remotely equal in affordability.

Wrong. i7-920 + higher models (Nehalem) and i7-860 + higher models (Lynnfield) existed before BD and are "8 Threaded" due to hyper-threading. In fact I still own such a Lynnfield and it still beats any FX in gaming...even though it was made on 45 nm (higher process node) and uses less power. But regardless. You claim was false not to mention that 4 Intel big cores usually are just about as fast as 8 BD cores.
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
Wrong. i7-920 + higher models (Nehalem) and i7-860 + higher models (Lynnfield) existed before BD and are "8 Threaded" due to hyper-threading. In fact I still own such a Lynnfield and it still beats any FX in gaming...even though it was made on 45 nm (higher process node) and uses less power. But regardless. You claim was false not to mention that 4 Intel big cores usually are just about as fast as 8 BD cores.

Not on games with new Game-Engines like Frostbite 3, CryEngine 3+ etc etc. Also games using Mantle.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Not on games with new Game-Engines like Frostbite 3, CryEngine 3+ etc etc. Also games using Mantle.

Frostbite 3 didnt fix scaling. There havent been much benefit for FX CPUs outside the BF franchies. And even then, it requires multiplayer gaming to take effect. And Mantle didnt change anything either there. The Intel cores are still faster.

For Cryengine, it the same thing. Did you see Evolved? 8 AMD cores again compete with 2 Intel cores. Very game specific again with Crysis.
 
Last edited:

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Zen, and hence K12, will use whatever parts from any CPUs they currently have but the designs are clearly high performance cores and they will target an Excavator module throughput but in a single core, whith eventualy a two threads per core capability.

Please show me the documentation for this. AMD doesnt even want to tell in relation to what when they make their statements. So I am inetrested in what you have found.
 

coercitiv

Diamond Member
Jan 24, 2014
7,118
16,488
136
Please show me the documentation for this. AMD doesnt even want to tell in relation to what when they make their statements. So I am inetrested in what you have found.
From the same article Abwx quoted:
AMD's execs noted that these are high-frequency, high-performance CPU cores that will span the range from laptops to desktops to servers—and not just "microservers."
Whether or not they will deliver a truly high performance core is to be seen, but this statement is anything but ambiguous.
 

cytg111

Lifer
Mar 17, 2008
25,226
14,718
136
Please show me the documentation for this. AMD doesnt even want to tell in relation to what when they make their statements. So I am inetrested in what you have found.

You could start with providing sources for all you own claims (including there will be no apu in the consoles) and not your own inferences on vague slides.
Besides didnt you read OP? It specifically says that "Shintai you dont need to post your s here...please dont.".. and thus, as you are here anyway speading your fud we can safely assume you're in full troll mode? Guess ill try the polite version too : Shintai plz go away.
 

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
From the same article Abwx quoted:

Whether or not they will deliver a truly high performance core is to be seen, but this statement is anything but ambiguous.

So against AMD communications to investors stating that among AMD target markets desktop isn't one of them, you are offering a tech report article, which quotes an unnamed AMD executive saying that Zen will be high performance?
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
From the same article Abwx quoted:

Whether or not they will deliver a truly high performance core is to be seen, but this statement is anything but ambiguous.

But not in relation to what. And thats the point. While Abwx already decided what it would be against. Even using a product as compare that already abandoned both server and desktop to begin with.

2 custom CPUs developed, that is both to perform higher than any AMD CPU with over twice the budget that these 2 have to share. Points to evolution of the cat cores. As per AMDs own slides and targets as well.
 
Last edited:

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
Huh? Where on that roadmap do you find Zen? And where does it say it's not a high performance CPU?

On the presentation roadmap, this is the link straight for AMD press release, which you can find here:

http://www.amd.com/en-us/press-releases/Pages/ambidextrous-computing-2014may05.aspx

And here's the link for the slide deck.

http://phx.corporate-ir.net/External.File?item=UGFyZW50SUQ9MjAxMjA5fENoaWxkSUQ9LTF8VHlwZT0z&t=1

Check slide 9:

AMD said:
2016:

Developing 64-bit ARM cores alongside new 64-bit x86 cores]

New 64-bit x86 Core is Zen, correct? Unless AMD is with another secret project out there we don't know of.

And on the same slide AMD states the target market for these two new cores:

AMD said:
Ambidextrous
Computing Leadership
x86 and 64-bit ARM for:

Dense Server
Embedded
Semi-Custom
Ultra-Low
Power Client

So here you have it, straight from AMD's own mouth what are the markets AMD is targeting with Zen and K12 and from that what are the performance order of magnitude we can expect from the chips. Desktops aren't even included on the presentation, let alone high end workstations.

We can chose between that information AMD is feeding to its investors and tech report. I don't see many reasons to pick tech report, especially after their poor coverage of AMD's CEO dismissal.
 

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
Frostbite 3 didnt fix scaling. There havent been much benefit for FX CPUs outside the BF franchies. And even then, it requires multiplayer gaming to take effect. And Mantle didnt change anything either there. The Intel cores are still faster.

For Cryengine, it the same thing. Did you see Evolved? 8 AMD cores again compete with 2 Intel cores. Very game specific again with Crysis.

We are comparing Core i7 920/860 (2.66-2.8GHz) against FX8350. In games using Frostbite 2.5 (BF3) and Frostbite 3 (BF4, HardLine, Dragon Age: Inquisition etc) and CryEngine 3 (Crysis 3, RYSE: Son of Rome etc) and in every Mantle game (Thief, Sniper Elite III etc) the FX8350 is faster.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
We are comparing Core i7 920/860 (2.66-2.8GHz) against FX8350. In games using Frostbite 2.5 (BF3) and Frostbite 3 (BF4, HardLine, Dragon Age: Inquisition etc) and CryEngine 3 (Crysis 3, RYSE: Son of Rome etc) and in every Mantle game (Thief, Sniper Elite III etc) the FX8350 is faster.

Is it winning due to scaling or simply frequency against 3-4 years older CPUs?
 

DrMrLordX

Lifer
Apr 27, 2000
22,534
12,402
136

Okay, so FMA4 is back in and phoronix got the info wrong. Noted.

Where was this said? According to the Phoronix article, the dropped instruction sets are as follows: TBM, FMA4, XOP, and LWP. All of these are AMD-specific instructions, not included on Intel processors.

So? They're still potentially useful to people who own CPUs that support those ISA extensions, and there's compiler support for them as well. I'll grant that XOP is less useful for AMD moving forward with AVX2 support on Excavator-based chips and presumed AVX2 support on Zen, but FMA4 still has some value (which is, apparently, why it's still supported by Zen, see above).

Whether AMD includes TSX or not will depend on how much they want to compete with Intel in the server market.

All the indicators from AMD point to Zen targeting some segment of the server market. The only thing we know so far is 95w TDP for 8c/16t on some variant of Samsung's 14nm node (which is also difficult to gauge, since we haven't seen any 14nm products either, have we?). There are lots of claims going around right now, such as some guy on semiaccurate claiming up to 32c/64t for Zen and K12 in 2016. Between that and vague slides there isn't much we can know for sure, which is why preview threads this far out are nothing but an invitation for speculation and the inevitable snarking and arguing.

Furthermore, as they are complicated operations which have 4 operands apiece (as opposed to three for FMA3), there's a good chance they contributed to the die size bloat of Bulldozer and its derivatives. From an efficiency standpoint, cutting them is the right decision.

Apparently news of FMA4's demise was premature . . .
 
Aug 11, 2008
10,451
642
126
Oh wow, it beats a 3 (soon to be 4) generation old processor from the competition under the most favorable game engine. I want one!!
 

Abwx

Lifer
Apr 2, 2011
11,783
4,691
136
So against AMD communications to investors stating that among AMD target markets desktop isn't one of them, you are offering a tech report article, which quotes an unnamed AMD executive saying that Zen will be high performance?

"Unnamed" AMD executives are Mark Papermaster and Jim Keller, read TR article, this will spare people some cluless statements like this one...
 

Phynaz

Lifer
Mar 13, 2006
10,140
819
126
"Unnamed" AMD executives are Mark Papermaster and Jim Keller, read TR article, this will spare people some cluless statements like this one...

Do I need to point out all the broken promises from AMD executives in regards to CPU performance?
 

mrmt

Diamond Member
Aug 18, 2012
3,974
0
76
Do I need to point out all the broken promises from AMD executives in regards to CPU performance?

This, but I wouldn't point my finger at them, yet. We don't know in what context AMD executives said that the design would be "high performance". Techreport doesn't even bother to atribute the quote to one of the execs, so we don't know who said what, when and in which context. And Techreport isn't exactly in a good phase now, they bluffed badly when Rory was fired. AMD corporate information isn't exactly forthcoming, but it's far better than techreport.
 
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/    |