Ryzen 2 slide in KitGuruTech video

[revanchrist]

https://www.reddit.com/r/AyyMD/comments/79aiwh/ryzen_2/

Ryzen 2 in Feb 2018? --> -1 Credibility Point. According to AMD's very own roadmap, Ryzen+ in 2018, Ryzen 2 in 2019.
Telling people three months ahead that you can preorder a CPU in January 2018? --> -1 Credibility Point.
Conveniently blurry and low resolution image? Definitely not photochopped. --> -1 Credibility Point.
Too good to be true performance? --> -1 Credibility Point.
Image originated from AyyMD/Wccftech/4Chan? --> -1 Credibility Point.

 

[Ancalagon44]

I think it is fake mostly because the Ryzen coming out next year is not Ryzen 2. At least, it is not what AMD has called Ryzen 2.

Ryzen on 12nm will be a little faster with a little higher clockspeed but anyone who is expecting 20% extra clockspeeds and 50% extra cores at the same time... well... I just don't think it will happen.

It makes no sense for the consumer market anyway. Remember AMD has to produce products people will buy. AMD knows that consumers do not need 12 core CPUs.

Besides - one low resolution blurry shot? Every other good leak we have had has been clearly visible. This is 100% fake.

 

[CatMerc]

It's Ryzen 2, not Zen 2.

Zen+ would still be called Ryzen 2, since it's the second generation Ryzen. Raven Ridge is Ryzen 2 even though it's not even on 12nm.

 

[Arzachel]

It's Ryzen 2, not Zen 2.

Zen+ would still be called Ryzen 2, since it's the second generation Ryzen. Raven Ridge is Ryzen 2 even though it's not even on 12nm.

I'll try to untangle this mess of a post.

Zen is the name of the architecture. The next iteration of the Zen architecture was called Zen+ in the earliest roadmaps but later renamed Zen 2. Zen+ isn't a thing anymore, but the tech community has appropriated this name for the 12nm die shrink of Summit Ridge - Pinnacle Ridge.

Ryzen is a marketing name and follows a very similar usage as Intel's Core: Ryzen 7 1800X vs Core i7 7700K for example. Raven Ridge uses the same nomenclature: Ryzen 5 2500U/Ryzen 7 2700U. Ryzen 2 isn't a thing. Ryzen 3 is, but the 3 doesn't imply what you're saying it does.

To summatize, Zen+ doesn't exist and neither does Ryzen 2.

Edit: yikes, I come off as even more of a pedantic asshole than usual, sorry. I kind of blame AMD on this whole mess.

 

[CatMerc]

I'll try to untangle this mess of a post.

Zen is the name of the architecture. The next iteration of the Zen architecture was called Zen+ in the earliest roadmaps but later renamed Zen 2. Zen+ isn't a thing anymore, but the tech community has appropriated this name for the 12nm die shrink of Summit Ridge - Pinnacle Ridge.

Ryzen is a marketing name and follows a very similar usage as Intel's Core: Ryzen 7 1800X vs Core i7 7700K for example. Raven Ridge uses the same nomenclature: Ryzen 5 2500U/Ryzen 7 2700U. Ryzen 2 isn't a thing. Ryzen 3 is, but the 3 doesn't imply what you're saying it does.

To summatize, Zen+ doesn't exist and neither does Ryzen 2.

Edit: yikes, I come off as even more of a pedantic asshole than usual, sorry. I kind of blame AMD on this whole mess.

I disagree about Ryzen 2 not existing. I see it more as Intel calling 8000 series CPU's their "8th Generation Core". AMD simply calls it Ryzen 2 rather than Ryzen 2nd Gen or anything of the sort.

As for Zen+, I guess I was just working under the assumption that everyone is aware of the difference that the community more or less came up with from Zen 2. Pinnacle Ridge is Summit Ridge Zen+, which is Zen on a better node and some IP updates, such as Precision Boost 2 which was originally intended to be in Ryzen 1 but didn't make it in time.

 

[Arzachel]

I disagree about Ryzen 2 not existing. I see it more as Intel calling 8000 series CPU's their "8th Generation Core". AMD simply calls it Ryzen 2 rather than Ryzen 2nd Gen or anything of the sort.

As for Zen+, I guess I was just working under the assumption that everyone is aware of the difference that the community more or less came up with from Zen 2. Pinnacle Ridge is Summit Ridge Zen+, which is Zen on a better node and some IP updates, such as Precision Boost 2 which was originally intended to be in Ryzen 1 but didn't make it in time.

I've only seen Ryzen 2 used in clickbait articles.

Calling Pinnacle Ridge Zen+ implicitly hints at architectural improvements and I think AMD putting up the same core with iGPU and a Devil's Canyon-esque die shrink as the second generation just makes it more confusing for anyone not obsesivly following CPU releases.

 

[gOJDO_n]

I don't believe that there would be any changes in the core microarchitecutre. But I belive that its pipeline logic will be fine tunned for higher clocks. I expect it to get improved macro-architecture with a much faster IF, PCI-e 4.0 and the precision boost 2 which is present in the Raven Ridge APUs.

 

[Abwx]

I would like to point out 12FDX specs which are no way related to 12LP but could serve to give us some insight on the probable choice of 84nm CPP

https://m.eet.com/content/images/eetimes/1 7 12 14 copared x 800_1505972923.jpg

12FDX is using planar transistors and as such cant be used as reference for 12LP estimations, geometries of the devices are completely differents.

Also a 84nm CPP on planar is not the same thing electricaly speaking as the same sized contact with finfets, despite the size being the same this will result in different transconductance and input capacitance.

FDX has lower conductance than Finfets, but the presence of an electric insulation between gate and substrate reduce notably the gate capacitance, less energy is required to charge those parasistic elements, wich more than compensate for the lower conductance (in respect of 12LP for instance), overall speed can be as good or even better than said finfets.

 

[CatMerc]

I've only seen Ryzen 2 used in clickbait articles.

Calling Pinnacle Ridge Zen+ implicitly hints at architectural improvements and I think AMD putting up the same core with iGPU and a Devil's Canyon-esque die shrink as the second generation just makes it more confusing for anyone not obsesivly following CPU releases.

At the very minimum we can expect changes already implemented in Raven Ridge. There can be any number of other tweaks in the pipeline that wouldn't require much work but would make sense for Pinnacle Ridge.

I'd like to point out that as much as AMD wanted to call Piledriver a new core after Bulldozer in order to keep up appearances after the Bulldozer turd, it was effectively a new stepping of Bulldozer, both using the Orichi floorplan, only 8150 being OC-B2 and 8350 being OC-C0

 

[Abwx]

I'd like to point out that as much as AMD wanted to call Piledriver a new core after Bulldozer in order to keep up appearances after the Bulldozer turd, it was effectively a new stepping of Bulldozer, both using the Orichi floorplan, only 8150 being OC-B2 and 8350 being OC-C0

It wasnt a stepping, or else SB to IB was a simple shrinked refresh...

The improvements :

http://www.hardware.fr/articles/880-1/amd-fx-8350-retour-amd.html

And the resulting perf/clock, bigger than what Intel got from the mentioned evolution :

http://www.hardware.fr/articles/880-6/bulldozer-vs-piledriver-4-ghz.html

ces gains sont très intéressants, comparables à ce qu'Intel offre lors d'un "Tock" (comme ce fut le cas entre Lynnfield et Sandy Bridge) et plus que ce que nous avons eu lors du "Tick" Ivy Bridge (3-4%).

We can only hope that PR will be the same kind of "stepping"....

 

[Topweasel]

I don't believe that there would be any changes in the core microarchitecutre. But I belive that its pipeline logic will be fine tunned for higher clocks. I expect it to get improved macro-architecture with a much faster IF, PCI-e 4.0 and the precision boost 2 which is present in the Raven Ridge APUs.

Yes, No, No, and yes.

 

[Smoblikat]

no way it stays on AM4 if it's official... AM4+ for sure

can't wait to not be able to find a X400 series board!!

If AMD's track record is anything to go by then AM4+ should still be physically compatible with AM4, and most boards would just need a BIOS update to support the new procs. Wasnt everything from AM2 -> AM3+ physically compatible?

 

[blublub]

I think naming is simple:

Ryzen = Zen
Ryzen 2xxx = Zen+ = 12nm improved process and some architecture tweaks that are in Raven Ridge and didn't make it on it Ryzen in 2017
Ryzen 3 = Zen 2 probably on 7nm

(tapatalk spam removed - please change your tapatalk settings to remove the "sig", thanks. Moderator VirtualLarry)

 

[EXCellR8]

If AMD's track record is anything to go by then AM4+ should still be physically compatible with AM4, and most boards would just need a BIOS update to support the new procs. Wasnt everything from AM2 -> AM3+ physically compatible?

IIRC AM3 (not AM3+) processors could physically fit in the AM2+ socket but I'd have to look into it. perhaps someone else knows more about that I think memory bus type came into play at some point too I can't remember

 

[Shmee]

Most likely fake me thinks, but time will tell.

 

[bauerbrazil2014]

https://videocardz.com/74260/amds-james-prior-talks-ryzen-2-and-vega-11

AMD (Ry)zen 2 will use AM4 socket
James Prior reassured that AM4 socket is here to stay (till 2020). The work on Zen 2 has already begun when fundamental parts of Zen 1 were already known. The important thing here is to distinguish Zen 2 from Zen 1 tick-tock process. The upcoming Ryzen 2000 series are likely to use refined Zen+ architecture. A die shrink and architecture optimizations are to be expected. So the Ryzen 2, or more precisely Zen 2 might actually arrive with Ryzen 3000 series, while Ryzen 2000 (or Ryzen 1×50) will use refined Zen1/Zen+ 12nm process instead.

If everything goes according to the plan, forward compatibility for Zen+ and Zen2 will be available with a simple BIOS flash on existing AM4 motherboards.

 

[William Gaatjes]

I think it is fake.
I do not know how the organisation is internally of AMD, but since Zen was completed long ago, it might be that since the release of zen / ryzen earlier this year, another team might have been working on improving where possible without major architectual changes.
Clockspeed improvements i also believe that will be the case. IF changes, not so much but higer memory clockspeeds, yes.
Memory latency has not changed between the recent released apus and ryzen 1800x when comparing at same ddr4 -2400.
The thing that makes me wonder is the official jedec speed for DDR4 when it comers to professional workstation and server cpus.
Those machines are not going to be overclocked and need to be reilable.
I glanced to the ddr4 speed bin and DDR4-2666, DDR4-2933 and DDR4-3200 are described in the document JESD79-4B (released June 2017)..
https://www.jedec.org/standards-documents/results/jesd79-4 ddr4
But i have reasonably faith those ratings are at the official voltage of 1.2V.

 

[DrMrLordX]

I think AMD will be looking to tweak the IPC by a few percent and boosting clocks by 10-15%. I can't see them adding more cores at this point.

Actually with the use of lower density libraries . . .

I’m more pessimistic on clock speed increases and think they should be getting double digit IPC gains. Zen is a new architecture and that usually means more low hanging fruit for the picking.

Not sure about IPC gains, but if they're going to lower-density libraries for Zen+ then we may be in for a pleasant surprise wrt clockspeed. Remember that the switch from 14nm LPP to 12nm LP alone should buy AMD another 400 MHz in max boost clocks. If they switch libraries then we may see another 100-200 MHz? Maybe?

I remember Carrizo had HD libraries and the Carrizo clockspeeds tanked hard vs. Kaveri (though power usage in Carrizo's intended clockspeed range was actually pretty good).

 

[William Gaatjes]

GF12LP is not an optical shrink. Its a library optimization with some transistor level enhancements.

https://www.anandtech.com/show/11854/globalfoundries-adds-12lp-process-tech-amd-first-customer

"The new 12LP relies on the groundwork set by the 14LPP, but uses 7.5T libraries, which is one of the ways that enables GlobalFoundries to shrink die sizes by increasing transistor density. Since the library contains different elements, IC developers have to “recompile” their designs to take advantage of the process. Meanwhile, since the 12LP and the 14LPP are very similar, for GlobalFoundries’ existing customers migration path to the 12LP is pretty straightforward."

My guess is GF 12LP 7.5T provides performance better than GF 14LPP HD (CPP=78nm ,9T) . For maximum performance 12LP will probably bring a combination of higher track height, relaxed CPP and transistor level enhancements. I am thinking CPP=84nm, 10.5T, taller fins and improved drive current to help AMD hit significantly higher clocks with Pinnacle Ridge.

This may be a silly question, but does increasing the density not contribute more to the dark silicon problem ?
I always assumed the other reason AMD goes for multi die(besides the obvious cost and yield advantage) is avoiding the dark silicon problem while increasing total core count per cpu package.
If AMD would go for more cores per die would the dark silicon problem not return again ?
Would keeping the same die size but purely go for speed enhancements(Boost and Extended Frequency Range maintained longer) not be a a good solution to avoid the dark silicon problem ?
Also, when the die continues to get smaller, creating an mcm solution would be more of a problem when thinking of all the microbumps needed when more connectivity is added.

edit:
Forgot to add :
The 12nm version of zen would not be a drop in replacement for the multi die packages like EPIC and threadripper) when not the same die size and position of the micro bumps.
If it would be a drop in replacement die, then AMD could seriously save some money for not having to develop the packages again from scratch.


edit:
typing error:

 

[hojnikb]

IIRC AM3 (not AM3+) processors could physically fit in the AM2+ socket but I'd have to look into it. perhaps someone else knows more about that I think memory bus type came into play at some point too I can't remember

actually its even more flexible than that
https://www.asrock.com/mb/nvidia/n68c-gs fx/

supports pretty much anything from am2 to am3+, provided you have the right memory.

 

[Yotsugi]

I sincerely doubt the core count, but the clocks?
Who knows, PR is client-only, AMD&GloFo may sacrifice density and go for highest performing option possible, as several people here have already noted.

 

[SPBHM]

I really hope AMD can increase the clocks like that and more, it's the main problem with their CPUs right now, 8 cores is more than needed already for $300-400.

 

[french toast]

I'm hoping for some more L3 cache, 15% higher turbo's, increase in fabric speed to decrease inter ccx latency by >20% and new agesa that further decreases memory latency and improve memory compatibility and over clocking.
Maybe a few % IPC increase through low hanging fruit would be nice.

I think that leak is total bull...but would give intel a huge shock

 

[Yotsugi]

some more L3 cache

But Zeppelin already has plenty, especially given that L3 is victim to L2.

increase in fabric speed to decrease inter ccx latency by >20%

Questionable.

 

[NostaSeronx]

If CMT.
8-core to 16-core.
Feasible.
Much wow.
16-core in one gen.
Much appeal.
Only quad-module CCX, no change to CCX.
Wow.