AMD X399 !!!!!

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Mar 10, 2006
11,715
2,012
126
Snowy Owl is HPC APU with 16C/32T and 4096 GCN cores.

So the TDP would be actually correct, for this chip.

I think there may have been confusion, X399 has to work with 8 channel memory controller, but there will be variety of chips available for this chipset.

AMD was making a big deal about how Naples doesn't require a discrete PCH.

We are getting trolled, seriously.
 

imported_jjj

Senior member
Feb 14, 2009
660
430
136
Not all that difficult for them to do what Intel does with HEDT and just push some consumer mobos with the server socket.
They are somewhat resources limited so they might not do it this year but they have no reason not to at some point.
Ofc if there is no x399, who cares, anyone can buy a server board and the 16 cores SKU. Nobody buys 16 cores for gaming or OC really so it would be more a matter of branding than anything else.
 

french toast

Senior member
Feb 22, 2017
988
825
136
Not all that difficult for them to do what Intel does with HEDT and just push some consumer mobos with the server socket.
They are somewhat resources limited so they might not do it this year but they have no reason not to at some point.
Ofc if there is no x399, who cares, anyone can buy a server board and the 16 cores SKU. Nobody buys 16 cores for gaming or OC really so it would be more a matter of branding than anything else.
what would the costs be for that?
 

Topweasel

Diamond Member
Oct 19, 2000
5,436
1,654
136
Yeah this is crazy false rumor. AMD wants to come out punching. I get that but they aren't ready for a Pro-sumer platformer yet. The R7 is enough for now. Not quite X299 strong on the connectivity. But cost effective enough to undercut demand for it. But realistically if AMD wants to get into the HEDT they need to cap off the Ryzen launch, get the APU's out. Probably a good idea for low voltage 8 6 and 8 core Ryzen laptop part to go with the laptop APU's. Release Naples to force IT departments to look their way. Then next gen offer a workstation platform. If they have decent penetration into a couple of these markets and still need growth. Then maybe Zenver 3 they can look at HEDT platform and only if they have made up ground lite and mildly threaded tasks.
 

otinane

Member
Oct 13, 2016
68
13
36
Rumors say end of 2017/early 2018 for Zen 2

AMD-Zen-Zen-2-Zen-3.jpg
 
  • Like
Reactions: lightmanek

lopri

Elite Member
Jul 27, 2002
13,209
594
126
Unlikely as it is all it takes is diverting a server platform and dressing it up as an "HEDT" or something that sounds just as intimidating. People who need that many cores will have gone with an appropriate server setup and simply installed their preferred OS/applications whether it is called a server or an HEDT. There is no technical barrier and thus the market demand will dictate the supply. Personally I do not think the client desktop is going to need more than 8 cores for a very long time.

Edit: imported_jjj beat me to it.
 
Last edited:

Atari2600

Golden Member
Nov 22, 2016
1,409
1,655
136
If I were in charge at AMD, I'd be viewing it as a relatively low-risk investment which should generate decent returns.

Get the motherboard manufacturers to produce a lower quality board which can take a modified form of the SP3 Die - i.e. move a pin around so it can't take an Opteron, but can take a (hypothetical) "Ryzen9".
 

mikk

Diamond Member
May 15, 2012
4,133
2,136
136
At least unlike Intel, AMD has ambitions to actually update its CPU cores over a reasonable period of time.


This is questionable because the leaked Roadmap from yesterday indicated that AMD stays with Zen cores for the next year, means Zen 2 isn't coming before 2019. But in this year Intel is updating its lineup from mobile to server with Icelake as well. And for the entry desktop Raven Ridge isn't even coming this year, there is no Zen competitor for Intels Core i3 and Pentium (G4560!) lineup. These marketing Roadmaps from AMD were never useful anyways.
 
Mar 10, 2006
11,715
2,012
126
This is questionable because the leaked Roadmap from yesterday indicated that AMD stays with Zen cores for the next year, means Zen 2 isn't coming before 2019. But in this year Intel is updating its lineup from mobile to server with Icelake as well. And for the entry desktop Raven Ridge isn't even coming this year, there is no Zen competitor for Intels Core i3 and Pentium (G4560!) lineup. These marketing Roadmaps from AMD were never useful anyways.

Yeah, looks like Zen2 isn't coming in 2018.
 

french toast

Senior member
Feb 22, 2017
988
825
136
This is questionable because the leaked Roadmap from yesterday indicated that AMD stays with Zen cores for the next year, means Zen 2 isn't coming before 2019. But in this year Intel is updating its lineup from mobile to server with Icelake as well. And for the entry desktop Raven Ridge isn't even coming this year, there is no Zen competitor for Intels Core i3 and Pentium (G4560!) lineup. These marketing Roadmaps from AMD were never useful anyways.
I cant find the roadmap, but i thought pinnacle ridge comes Q1 2018? That is zen2.
RR will be on zen1 for ages though and stoney ridge use excavator.
 
Mar 10, 2006
11,715
2,012
126
I cant find the roadmap, but i thought pinnacle ridge comes Q1 2018? That is zen2.
RR will be on zen1 for ages though and stoney ridge use excavator.

AMD-Pinnacle-Ridge-Raven-Ridge.jpg


Pinnacle Ridge is Zen, not Zen2. Looks like another Bristol Ridge/Kaby Lake kind of deal here.
 

french toast

Senior member
Feb 22, 2017
988
825
136
AMD-Pinnacle-Ridge-Raven-Ridge.jpg


Pinnacle Ridge is Zen, not Zen2. Looks like another Bristol Ridge/Kaby Lake kind of deal here.
Thanks, but im sure its zen 2, or zen +.
Its another bulldozer - piledriver, or in amds own words "tock tock tock".
Unless on 14nm lpu amd would have no reason to tape out another 8 core processor if using the same cores and process, if they are not taping a new die then why would they call it pinnacle ridge? They would just respin summitridge and add a 50 on the end or something as there is nothing else on die to change unlike apus which have graphics.

I expect pinnacle ridge to fix some low hanging fruit, Either amd themselves or publications the last week with knowledge mention 15% ipc, whether this is accurate i dont know, i expect 5-10% ipc with slightly better clocks.
 
Mar 10, 2006
11,715
2,012
126
Thanks, but im sure its zen 2, or zen +.
Its another bulldozer - piledriver, or in amds own words "tock tock tock".
Unless on 14nm lpu amd would have no reason to tape out another 8 core processor if using the same cores and process, if they are not taping a new die then why would they call it pinnacle ridge? They would just respin summitridge and add a 50 on the end or something as there is nothing else on die to change unlike apus which have graphics.

I expect pinnacle ridge to fix some low hanging fruit, Either amd themselves or publications the last week with knowledge mention 15% ipc, whether this is accurate i dont know, i expect 5-10% ipc with slightly better clocks.

If it were Zen 2, it would say something like:
  • Up to 8 "next generation Zen" cores
  • Up to 8 "enhanced Zen" cores
  • Up to 8 "Zen 2" cores
The fact that it just says Zen means exactly that -- it's the OG Zen.
 
  • Like
Reactions: mikk

french toast

Senior member
Feb 22, 2017
988
825
136
If it were Zen 2, it would say something like:
  • Up to 8 "next generation Zen" cores
  • Up to 8 "enhanced Zen" cores
  • Up to 8 "Zen 2" cores
The fact that it just says Zen means exactly that -- it's the OG Zen.
No the clue is in the die code name, if you look at raven ridge and stoney ridge they keep the same code names through to 2019, why? Because it is the same die/generation, if you look at bristol ridge it was the same processor cores but had updated IMC for ddr4 amongst a few other changes, because it was an apu they could change other things on die and get away with calling it a new generation, likewise with kabylake which had a few changes and a new process variant.
If you you look at RR and SR there isn´t a new name because it is the same processor, so with nothing else to change on die and presumably exactly the same 14nm lpp process why would they name it pinnacle ridge instead of summitridge? :/

Looking at the first official roadmap in this thread longside AMD tock tock comments as well as reason stated above we must surely conclude pinnacle ridge = zen 2.
 
Mar 10, 2006
11,715
2,012
126
No the clue is in the die code name, if you look at raven ridge and stoney ridge they keep the same code names through to 2019, why? Because it is the same die/generation, if you look at bristol ridge it was the same processor cores but had updated IMC for ddr4 amongst a few other changes, because it was an apu they could change other things on die and get away with calling it a new generation, likewise with kabylake which had a few changes and a new process variant.
If you you look at RR and SR there is new name because is the same processor, so with nothing else to change on die and presumably exactly the same 14nm lpp process why would the name it pinnacle ridge instead of summitridge?

Looking at the first official roadmap in this thread longside AMD tock tock comments as well as reason stated above we must surely conclude pinnacle ridge = zen 2.

I respectfully disagree. Pinnacle Ridge could be a circuit/process optimization of Summit Ridge.

Bristol Ridge in terms of the functional blocks on the die was practically identical to Carrizo, the main improvements were in the process and I believe firmware.
 

french toast

Senior member
Feb 22, 2017
988
825
136
I respectfully disagree. Pinnacle Ridge could be a circuit/process optimization of Summit Ridge.

Bristol Ridge in terms of the functional blocks on the die was practically identical to Carrizo, the main improvements were in the process and I believe firmware.
Ok, but that optimisation would surely include some low hanging fruit ipc improvements no? = zen2.

I dont expect zen2 to be a massive jump, maybe 5% plus process optimisation for a net 15% would easily qualify for zen2 imo.
I dont see another big expensive change to the die if they are not going to add improvements beyond circuitry and process, they would be better off with a respin and calling it summitridge refresh.
That it why i believe PR is zen2.
Besides wasn't bristol ridge a different size and or transistor count? Thought i read that somewhere?