Wednesday, December 5th 2018

AMD 3rd Generation Ryzen Probable SKUs, Specs, Pricing Leaked?

by
btarunr
 Discuss (81 Comments)
One of our readers tipped us off with a very plausible looking image that drops a motherlode of information about what AMD's 2nd generation Ryzen (aka Ryzen 3000 series) processor lineup could look like. This includes a vast selection of SKUs, their CPU and iGPU core configurations, clock-speeds, and OEM channel pricing. The list speaks of a reentry for 7th generation A-series "Excavator" as Duron X4 series, followed by Duron 300GE-series based on a highly cut down "Raven Ridge," Athlon 300GE 2-core/4-thread based on an implausible "Zen+ 12 nm" APU die, followed by quad-core Ryzen 3 3000 series processors with and without iGPUs, making up the company's entry-level product lineup.

The core counts seem to jump from 4-core straight to 8-core, with no 6-core in between, for the Ryzen 5 series. This is also where AMD's new IP, the 7 nm "Zen 2" architecture, begins. There appears to be a large APU die (or a 3-chip MCM) with an 8-core CPU and 20-CU iGPU, which makes up certain Ryzen 5 SKUs. These chips are either 8-core/8-thread or 8-core/16-thread. The Ryzen 7 series is made up of 12-core/24-thread processors that are devoid of iGPU. The new Ryzen 9 series extension caps off the lineup with 16-core/32-thread SKUs. And these are just socket AM4.
3rd generation Ryzen Threadripper processors will be client-segment derivatives of the EPYC "Rome" MCMs, which combine up to 64 cores across 8-core 7 nm CPU chiplets with an I/O die handling a monolithic 8-channel memory interface and PCIe. Threadripper SKUs begin at 24-core/48-thread, and go on to include 32-core/64-thread, 48-core/96-thread, and 64-core/128-thread, across X and WX SKUs.

The OEM-channel pricing for all these SKUs seem to linearly succeed the current product stack, with the addition of newer SKUs that have no predecessors taking up gaps in the price-points.

At this point, this picture is either a motherlode of information or some fanboy's wet-dream, and TechPowerUp takes no responsibility for its accuracy.

Related News

Add your own comment

81 Comments on AMD 3rd Generation Ryzen Probable SKUs, Specs, Pricing Leaked?

#26
Fabio Bologna
I would just add my thought to this whole deal...

Just think of this scenario:
Video editing company that works in a small environment and is tired of having computers everywhere to edit potentially 4k/8k footage.
With something like this 3990WX (64c/128t) they could have 8VMs running EACH 8c/16t v-CPUs and 32GB of RAM... Assuming a similar amount of PCIe lanes to the previous gen, each GPU would have 8x connectivity... a nice rtx2070 would do the trick here...
Just imagine this... I know Linus already did this BUT with a dual xeon motherboard! this is all in one chip!
Just remanage it a bit and you could also have a spare for a freeNAS instance locally and SMB shares for each VM... all in what could be a 3u chassis in a rack somewhere in another room...

This is a potential usage I would think...
Posted on Reply
#27
Frick
Fishfaced Nincompoop
EarthDogA reputable joint
Ain't such a thing these days by the old fashioned definition of "reputable".

Me I just hope those Durons are real and they do support overclocking. They would absolutely destroy the price/performance graphs.
Posted on Reply
#29
cdawall
where the hell are my stars
FrickAin't such a thing these days by the old fashioned definition of "reputable".

Me I just hope those Durons are real and they do support overclocking. They would absolutely destroy the price/performance graphs.
Only if they unlock are they the coolest things ever
Posted on Reply
#30
AMX85
Duron? why now? where's Sempron? lol
Posted on Reply
#31
goldman
i find it really hard to believe there will be no ryzen 3 made with zen 2 cores, tbh
Posted on Reply
#32
john_
Whatever it is using bulldozer architecture should be killed. Anyone buying a bulldozer CPU, is a potential lost and never coming back future customer. We can see this now. AMD's Ryzen processors are selling for peanuts and still can't really take over the market. AMD does have a big part of the retail market in Europe, based on mindfactory numbers, but I doubt this is the case in the US market also. The damage that bulldozer done on AMD's image is huge and still not repaired. That bulldozer cancer should be totally removed from the market, not change name.
Posted on Reply
#33
XiGMAKiD
As others say it looks fake, if AMD gonna increase the core count per die 12-core is more plausible and dual socket TR is probably gonna steal some of EPYC sales which is not good while being too niche of a product
Posted on Reply
#34
Darmok N Jalad
cdawallOnly if they unlock are they the coolest things ever
Long live the “pencil trick!”
Posted on Reply
#35
FlanK3r
EarthDogNo.....Because it is fake?

How can they not reach 5 GHz on a 8c/16t part, but can on 32/46c 64/96t parts with its BASE clock higher than most of the other that are boost???????????????

This really seems fake (and click bait). We saw this and chose not to front page it for those reasons.

Oh, links please....

Why post it when its clearly BS?
Why do you need link? If I have direct contacts, its easy way ask them via social media. I have many more good contacts. One example (but I did not solved with this guy) U can se we are friends and he is working in AMD.
Posted on Reply
#36
EarthDog
Because anyone can say such things is why... I appreciate you showing me facebook image of some random AMD guy though. :)
Posted on Reply
#37
Markosz
Sorry, but fake as f....

Look at those Threadrippers, higher core count, but also much much higher clock speed?
If 64 core BASE 5Ghz speed doesn't smell fake, then nothing does.
Posted on Reply
#38
Tomorrow
HD64GSomething important is missing and it's the -amost certain to be made- 6C/12T CPUs. Very huge absense given the great success of the 1st gen's 1600(X) and 2nd gen's 2600(X) Ryzen cpu series.
That will be the new Ryzen 3's. Using either one 6 core chiplet and a dummy chiplet+IO die or using 6 core chiplet and navi die+IO die.
FlanK3rIts fake, total bulshit to improve the views on video:mad:. I have also confirmed by AMD, thats fake info.
Wow AMD confirmed to you specifically that it's a fake? Man you have connections bro.
cdawall800mhz over the current Gen chip with double the core overhead. Yep sure, I will believe it when I see it.
People get too caught up in clockspeeds saying 5Ghz boost is impossible. Actually percentage gains are under 15% for pure clockspeed. The rest comes from IPC improvements. Seems doable considering 7nm advantages at the same power. Also chiplets allow better speed binning.
Posted on Reply
#39
R0H1T
He didn't say anything about TR, the youtuber I mean.
MarkoszSorry, but fake as f....

Look at those Threadrippers, higher core count, but also much much higher clock speed?
If 64 core BASE 5Ghz speed doesn't smell fake, then nothing does.
Posted on Reply
#40
cdawall
where the hell are my stars
TomorrowPeople get too caught up in clockspeeds saying 5Ghz boost is impossible. Actually percentage gains are under 15% for pure clockspeed. The rest comes from IPC improvements. Seems doable considering 7nm advantages at the same power. Also chiplets allow better speed binning.
The chart shows 5ghz base clocks on the 3990wx. Even a 5ghz boost clock is a substantial gain for these, binning this early in the game will be curious. We already saw the binning wasn't the greatest with the current gens.
Posted on Reply
#41
Tomorrow
cdawallThe chart shows 5ghz base clocks on the 3990wx. Even a 5ghz boost clock is a substantial gain for these, binning this early in the game will be curious. We already saw the binning wasn't the greatest with the current gens.
Yeah i agree that 5Ghz base is not doable, i guess it is meant as 5Ghz boost but the placement is wrong but then again the chart has several errors in it like the Duron branding which i think is unlikely to be brought back. Even official AMD slides have errors in them from time to time.
Posted on Reply
#42
EarthDog
Its possible, just seemingly not with 32/38 cores and 64/96t. If the 8c/16t part can't do it (according to that sheet), then surely the larger parts can't. Can you imagine the TDP on those things with 6 chiplets, LOL.
Posted on Reply
#43
cdawall
where the hell are my stars
EarthDogIts possible, just seemingly not with 32/38 cores and 64/96t. If the 8c/16t part can't do it (according to that sheet), then surely the larger parts can't. Can you imagine the TDP on those things with 6 chiplets, LOL.
At 4.2 across 32 cores I pull over 800w at the wall. My chip is a reasonably good bin at that.
Posted on Reply
#44
R0H1T
cdawallAt 4.2 across 32 cores I pull over 800w at the wall. My chip is a reasonably good bin at that.
That's because TR's being clocked way beyond it's ideal efficiency curve, also in part due to 14nm GF being unsuitable for high frequencies. TSMC's 7nm should help a lot more in this regard.

forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
Posted on Reply
#45
Darmok N Jalad
Another variable is architectural changes that may allow the chip to run at a faster clock. It’s not all 7nm magic. Intel can make 5.0 ghz on 14nm+++. It seems like 2xxx Ryzen is architecturally limited to its current max clocks, not fabrication limited. Ryzen 3xxx is both a new process and a revised architecture.
Posted on Reply
#46
efikkan
Darmok N JaladAnother variable is architectural changes that may allow the chip to run at a faster clock. It’s not all 7nm magic. Intel can make 5.0 ghz on 14nm+++. It seems like 2xxx Ryzen is architecturally limited to its current max clocks, not fabrication limited. Ryzen 3xxx is both a new process and a revised architecture.
Don't forget that TSMC's 16/12nm and Samsung/GloFo 14/12nm are really ~20nm if you compare it to the density of Intel 14nm and 22nm. On top of density, Intel's nodes have also higher performance.

So the transition for AMD to 7nm will be a larger jump in density than Intel to 10nm, despite the maximum density of Intel 10nm and TSMC 7nm is fairly comparable. And we still don't know the performance characteristics of the new nodes, which is important for clockspeed and so on, and this could go either way.
Posted on Reply
#47
Metroid
It's not fake, if you think is fake then it means that you were always so disappointed by Intel for milking in or amd for hyping in and then never delivered. You need to understand that Lisa Su means when she wants to and the way I see is that Lisa Su wants to change the hype and then the disappointment cases of many years. Believe in, this time AMD will make history.

Also, amd will deny anything at this point.
Posted on Reply
#48
EarthDog
MetroidIt's not fake, if you think is fake then it means that you were always so disappointed by Intel for milking in or amd for hyping in and then never delivered. You need to understand that Lisa Su means when she wants to and the way I see is that Lisa Su wants to change the hype and then the disappointment cases of many years. Believe in, this time AMD will make history.
Lisa's burner account above. :p

So weird to use the third person when talking about yourself! Ha!
Posted on Reply
#49
Darmok N Jalad
efikkanDon't forget that TSMC's 16/12nm and Samsung/GloFo 14/12nm are really ~20nm if you compare it to the density of Intel 14nm and 22nm. On top of density, Intel's nodes have also higher performance.

So the transition for AMD to 7nm will be a larger jump in density than Intel to 10nm, despite the maximum density of Intel 10nm and TSMC 7nm is fairly comparable. And we still don't know the performance characteristics of the new nodes, which is important for clockspeed and so on, and this could go either way.
Apple appears to have worked some magic on 7nm with the A12X. The performance seems to be really good, and that’s in a passively cooled enclosure. Granted, that SOC isn't aiming for 5.0ghz either. Still, it’s encouraging to see one large vendor having some success with 7nm already.
Posted on Reply
#50
GhostRyder
If true that would be a HUGE improvement overall in all areas. Now we will have to see if it actually happens.
Posted on Reply
Add your own comment
Dec 18th, 2024 06:01 EST change timezone

Latest GPU Drivers

New Forum Posts

Popular Reviews

Controversial News Posts