AMD zen pre-release thread!

[Kuiva maa]

Cat cores are dead, there is no market for them left. It will be zen (construction cores successor) and the ARM only from next year onwards.

 

[Tuna-Fish]

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.

Inside the actual core, FMA3 and FMA4 both have 3 operands and one output, the expense of supporting them is exactly the same. Register renaming means that sharing the register name of the output with one input does not mean anything about how data moves in the CPU itself.

The only thing that FMA4 costs is a few gates more in the decoder. The expense of supporting them in an architecture that already supports FMA3 is minimal.

 

[AtenRa]

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!!

Nice trying but you are forgetting that the FX was released in 2011 (Bulldozer) and 2012 (Piledriver).

 

[frozentundra123456]

Nice trying but you are forgetting that the FX was released in 2011 (Bulldozer) and 2012 (Piledriver).

Well then lets look at those new, faster AMD processors. Oh, right, there aren't any.

 

[AtenRa]

Well then lets look at those new, faster AMD processors. Oh, right, there aren't any.

You are still forgetting about FX9370 and FX9590, also FX8320E is faster and at lower TDP.

 

[ShintaiDK]

You are still forgetting about FX9370 and FX9590, also FX8320E is faster and at lower TDP.

The first 2 are 220W ultra niche CPUs for the most dedicated fans. The 8320E isnt faster. Its slower, a lot slower.
http://www.anandtech.com/show/8864/amd-fx-8320e-cpu-review-the-other-95w-vishera/4
http://www.anandtech.com/show/8864/amd-fx-8320e-cpu-review-the-other-95w-vishera/5

Its actually so slow it makes the 8150 look fast.

But its already way derailed from the topic.

 

[AtenRa]

The first 2 are 220W ultra niche CPUs for the most dedicated fans. The 8320E isnt faster. Its slower, a lot slower.
http://www.anandtech.com/show/8864/amd-fx-8320e-cpu-review-the-other-95w-vishera/4
http://www.anandtech.com/show/8864/amd-fx-8320e-cpu-review-the-other-95w-vishera/5

Its actually so slow it makes the 8150 look fast.

But its already way derailed from the topic.

Where in those links did you see the FX8320E being slower than Core i7 920/860 ???

 

[Ramses]

Not a lot of info but you can pull them both with the bench tool.

http://www.anandtech.com/bench/product/1402?vs=47

 

[AtenRa]

Not a lot of info but you can pull them both with the bench tool.

http://www.anandtech.com/bench/product/1402?vs=47

heh, FX8320E is faster in Single Thread in Cinebench 10

 

[Enigmoid]

heh, FX8320E is faster in Single Thread in Cinebench 10

Yep. Though both chips have a ton of overclocking headroom.

 

[NTMBK]

Cat cores are dead, there is no market for them left. It will be zen (construction cores successor) and the ARM only from next year onwards.

Bay Trail and Kabini based laptops are very popular where I live. Plenty of life in that market segment.

 

[Kuiva maa]

Bay Trail and Kabini based laptops are very popular where I live. Plenty of life in that market segment.

The particular segment is being suppressed from top (increasingly efficient big cores) and bottom (various tablet forms). Intel has the luxury to offer a solution to shoehorn in there exactly because their little core is offered in mobile android based devices,they aren't preparing it for this niche.

Come zen a big part of the laptop power envelope in which today AMD fields jaguar/puma will be replaced with zen. If zen is indeed a puma successor that means AMD will only be competitive in perhaps half of today's cheap laptop segment (the bottom one) and nowhere else ,since they do not have a tablet business and a little core wouldn't even be able to compete with ULV big cores, let alone desktop chips. If zen is a big core, AMD will be having a product potentially competitive from the top part of the cheap laptop segment all the way to x86 servers with everything in between. Now tell me,which scenario makes more sense. To me it is obvious that AMD is preparing a big core. It might be a huge potato,the worst chip ever for all I know or care,the final nail in the coffin of their desktop aspirations. I just don't see how on earth that will be a puma successor especially when Lisa Su specifically said they are about to renew their years old server portfolio. Why even bother field a puma successor vs big fat Xeons, there are faster ways of wasting money (that they don't really have).
And let's not even go to custom etc solutions, big part of it will probably be served by their ARM core, we already have rumors.

http://www.kitguru.net/components/c...om-server-microprocessor-for-facebook-source/

This might be a pipe dream, but it a sign of intention nonetheless. Everything points to a big core, even next gen consoles would be better served by a big core, Sony and MS will not suddenly aim for 15W TDP solutions, console power envelope is somewhat static,therefore open to 14nm or whatnot big cores. If however, AMD against all reason prepares a little core, shareholders might want to get rid of their stock, like ASAP.

 

[monstercameron]

The particular segment is being suppressed from top (increasingly efficient big cores) and bottom (various tablet forms). Intel has the luxury to offer a solution to shoehorn in there exactly because their little core is offered in mobile android based devices,they aren't preparing it for this niche.

Come zen a big part of the laptop power envelope in which today AMD fields jaguar/puma will be replaced with zen. If zen is indeed a puma successor that means AMD will only be competitive in perhaps half of today's cheap laptop segment (the bottom one) and nowhere else ,since they do not have a tablet business and a little core wouldn't even be able to compete with ULV big cores, let alone desktop chips. If zen is a big core, AMD will be having a product potentially competitive from the top part of the cheap laptop segment all the way to x86 servers with everything in between. Now tell me,which scenario makes more sense. To me it is obvious that AMD is preparing a big core. It might be a huge potato,the worst chip ever for all I know or care,the final nail in the coffin of their desktop aspirations. I just don't see how on earth that will be a puma successor especially when Lisa Su specifically said they are about to renew their years old server portfolio. Why even bother field a puma successor vs big fat Xeons, there are faster ways of wasting money (that they don't really have).
And let's not even go to custom etc solutions, big part of it will probably be served by their ARM core, we already have rumors.

http://www.kitguru.net/components/c...om-server-microprocessor-for-facebook-source/

This might be a pipe dream, but it a sign of intention nonetheless. Everything points to a big core, even next gen consoles would be better served by a big core, Sony and MS will not suddenly aim for 15W TDP solutions, console power envelope is somewhat static,therefore open to 14nm or whatnot big cores. If however, AMD against all reason prepares a little core, shareholders might want to get rid of their stock, like ASAP.

I don't know about all of that but from the little info we know I'd assume zen is just k12 with x86 bits tacked on.
Small core
Mobile oriented
High efficiency
Scalable
Properly HSA integrated
High clocks
High integration

 

[Abwx]

I don't know about all of that but from the little info we know I'd assume zen is just k12 with x86 bits tacked on.

Small core

As repeated ad nauseam the small core is Skybridge and is a successor to Puma, there s no point developping a new small core when the current Beema and Mullins have the best efficency in the sub 15W segment, and Carrizo will extend this lead up to 25-30W, so where is the need of yet another small core..?..

Mobile oriented
High efficiency
Highly Scalable
Properly HSA integrated
High clocks
High integration

Zen will be all this plus server dedicated, it s the meaning of the bolded and corrected part in your post, indeed you forgot "high performance".

 

[monstercameron]

As repeated ad nauseam the small core is Skybridge and is a successor to Puma, there s no point developping a new small core when the current Beema and Mullins have the best efficency in the sub 15W segment, and Carrizo will extend this lead up to 25-30W, so where is the need of yet another small core..?..



Zen will be all this plus server dedicated, it s the meaning of the bolded and corrected part in your post, indeed you forgot "high performance".

Yeah but high performance in what metric and relative to what?

 

[Abwx]

Yeah but high performance in what metric and relative to what?

Forcibly relative to their most recent big core design, i.e, Excavator, and at the module level when it comes to equivalent throughput.

What they are obviously targeting is a single core, eventualy dual threaded, that has a throughput close to a EXV module, not to say that they wont re use existing parts that are currently used in Puma.

Indeed the urban legend origin of Zen being a small core may have its source in the fact that it was published, in some sites, that they could eventualy use Puma s branch prediction circuitry for Zen...

One has to realize that Vishera at 80% of its max Integer throughput rival a 4770K that is at close of 100% of its own one, an Excavator iteration of Vishera would had no problem rivaling Intel s best if the latter has no subtancial process advantage, so ditching the BD concept for a more traditional approach require designing a core that can be almost as powerfull as an updated BD module using existing parts.

 

[Vesku]

You are still forgetting about FX9370 and FX9590, also FX8320E is faster and at lower TDP.

And if AMD had the FX-8320E out in Jan 2012 instead of 3 years later they probably could have avoided some layoffs.

I think it's pretty clear from Carrizo details, their management statements, and the little they've openly discussed about SkyBridge (is that still happening?) and Zen+K12 they are focusing in on efficiency. This doesn't necessarily mean tiny cores either, just that it will have clocking behavior more like current Carrizo and Broadwell than SandyBridge. This is a good thing, imo, given it's the general focus of the rest of the industry including Intel.

 

[mrmt]

Bay Trail and Kabini based laptops are very popular where I live. Plenty of life in that market segment.

Where I live AMD was pretty much wiped out. I can find a few FX chips here and there, but notebooks are almost 100% Intel.

How's the trend there, AMD is growing or shrinking?

 

[monstercameron]

Where I live AMD was pretty much wiped out. I can find a few FX chips here and there, but notebooks are almost 100% Intel.

How's the trend there, AMD is growing or shrinking?

down here in florida, my local bigbox stores still have a few kabini/beema skus mixed with predominantly intel pentiums and celerons [non-baytrail]

 

[erunion]

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...








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

Your source is the words of a random internet tech journalist.

Let me tell you how this will play out. The people basing their predictions on actual AMD statements and documents will be right(shintai). The people whose predictions run contrary to AMDs statements will be wrong(you).

Your confirmation bias has caused you to believe a random person over official AMD slides.

 

[Shehriazad]

Your source is the words of a random internet tech journalist.

Let me tell you how this will play out. The people basing their predictions on actual AMD statements and documents will be right(shintai). The people whose predictions run contrary to AMDs statements will be wrong(you).

Your confirmation bias has caused you to believe a random person over official AMD slides.

But he's right about one thing in this case...this slide is not related to Zen.

 

[erunion]

But he's right about one thing in this case...this slide is not related to Zen.

That's a red herring.

The statement he quoted is in direct contradiction to the other slide that clearly says Zen is targeted at low power clients. (EDIT: My mistake, I was thinking the slide that mentioned "2016+" mentioned "Zen" by name but it doesn't. So it does still requires correlation with the information that Zen is coming in 2016")

But to address the red herring: The skybridge slide is labeled 2015 so the cores are a57/puma+. But the concept isn't going into the trash after 2015. Ambidextrous strategy will continue in 2016+ when the cores are k12/zen. So the slide does tell us about Zen.

Its likely that AMD used the generic arm/x86 label deliberately to make the connection to the future cores.

We've seen wishful thinking and confirmation bias poison people's thinking on this forum for years.

 

[itsmydamnation]

That's a red herring.

The statement he quoted is in direct contradiction to the other slide that clearly says Zen is targeted at low power clients.

But to address the red herring: The skybridge slide is labeled 2015 so the cores are a57/puma+. But the concept isn't going into the trash after 2015. Ambidextrous strategy will continue in 2016+ when the cores are k12/zen. So the slide does tell us about Zen.

Its likely that AMD used the generic arm/x86 label deliberately to make the connection to the future cores.

We've seen wishful thinking and confirmation bias poison people's thinking on this forum for years.

if you listen to jim kellar, he talks about this form the perspective of not boiling the ocean. Seattle was to get arm to x86 server platform level, skybridge was to figure out how to create the interconnecting fabric allowing for either x86 or arm to drop into the SOC. He then specifically states that the fabric in K12 is significantly upgraded in terms of features and in terms of scope ( specifically mentions high performance servers).

Also when asked about K12 vs A57 jim specifically states that they are both beautiful cores but they target completely different power and performance ranges.


What he doesn't say is if zen and k12 use the same fabric or if zen and k12 target the same performance envelopes.

 

[AtenRa]

And if AMD had the FX-8320E out in Jan 2012 instead of 3 years later they probably could have avoided some layoffs.

Well, that has nothing to do with the fact that even Bulldozer(FX8150) was faster than Core i7 920/860.
http://anandtech.com/bench/product/434?vs=108

I think it's pretty clear from Carrizo details, their management statements, and the little they've openly discussed about SkyBridge (is that still happening?) and Zen+K12 they are focusing in on efficiency. This doesn't necessarily mean tiny cores either, just that it will have clocking behavior more like current Carrizo and Broadwell than SandyBridge. This is a good thing, imo, given it's the general focus of the rest of the industry including Intel.

As i have said so many times, efficiency can be achieved at high performance as well. Efficiency doesnt mean low power and low performance only.
I dont expect 140-150W TDP CPUs but ZEN will not be a low power low cost small core design.

 

[Ramses]

Well, that has nothing to do with the fact that even Bulldozer(FX8150) was faster than Core i7 920/860.
http://anandtech.com/bench/product/434?vs=108

Was $50 cheaper too