AMD Previews 7nm Ryzen CPUs at CES

Donald Bell

Limp Gawd
Joined
Apr 24, 2018
Messages
160
Anyone notice the extra room below the CPU chip? Something tells me they are saving the 16c/32t Ryzen monster for a response to Intel depending on what they do. See my doctored photo.

10sdu7b.jpg
 

juanrga

Pro-Intel / Anti-AMD Just FYI
Joined
Feb 22, 2017
Messages
2,730
It clearly stated TBA , that means: To Be Announced.
It is not about faith in AdoredTV it is where the leaks are coming from. Beside his technical guess work pretty much got Ryzen 3000 down.
Even the IO die and the chiplets.

AdoredTV said desktop would be different, only one die. So he failed on that as well.
 

jmnovak

Weaksauce
Joined
Sep 14, 2007
Messages
127
For reference,

A 2700X OCed to 4.4ghz with CL14 3200mhz DDR4 scores 1964 in CB:View attachment 133987
(Guru3d)
The 16 thread Zen 2 CPU scored 2057:
View attachment 133988
(9900k left, AMD 16 thread right)

If your only variable is clockspeed, then a 2700X clocked at 4.6ghz would achieve a similar score that was shown off today as 16 thread Zen 2.

Zen 2 is evidently quite power efficient. I hope that means there is a lot of headroom for overclocking.

Curious to see how this one plays out.
Single core performance is obviously on par with a 9900k but guessing at lower clock speed. IPC efficiency itself has most likely went up 10% clock for clock. A 2700x @ 4.3ghz get 180 in CB15 for single core. Just on IPC improvements alone the new chips should probably get 198 to 200 at the same speed. Add a few hundred megahertz to the top end (4.8ghz) and they should have no problem performing better than Intel's 9000 series. With overclocking you might see 5 to 5.2ghz but I would be very surprised if we see any higher than that. I feel like there's a hard limit starting to rear its ugly head with the chip technology. Intel has billions of dollars in r&d I'm sure if there was a way to push the clock speeds higher they would have done it by now. I think from this point out it's going to have to be a massive change in architecture and materials in order to push speed any higher with the same design.
 
Last edited:

RanceJustice

Supreme [H]ardness
Joined
Jun 9, 2003
Messages
5,972
I'm pleased to see the progress but I really hope they launch much sooner. After all, isn't Q2 supposed to be when Threadripper 3000 7nm Zen2 series arrive? They have Intel on the ropes here, but if they don't launch in stores by March for at least the AM4, that would be disappointing. I hope that neither AM4 nor Threadripper are delayed...
 
  • Like
Reactions: mikeo
like this

hurleybird

Weaksauce
Joined
Jan 14, 2012
Messages
106
Anyone notice the extra room below the CPU chip? Something tells me they are saving the 16c/32t Ryzen monster for a response to Intel depending on what they do. See my doctored photo.

If you watch the part of the livestream where Lisa holds up the processor, you can actually see the attach point of the 2nd chiplet shimmer here and there as the light catches it at certain angles.

 

DNMock

Limp Gawd
Joined
Apr 16, 2015
Messages
399
AMD gifted Intel 6 months to come up with something.

Didn't intel already shoot their wad with their upcoming canyon series CPU's?

Unless they secretly got their 10nm fab producing somehow I think AMD is gonna be just fine in the consumer desktop market in 2019
 

Sindalis

n00b
Joined
Aug 27, 2014
Messages
43
AdoredTV said desktop would be different, only one die. So he failed on that as well.

If you watch his videos he had two thoughts on what it could possibly be, and his hope was for a chiplet based design with an IO die. His original leak/predictions in early December are remarkably similar to exactly how the ryzen 2 die shot looks.

As far as why only one chiplet is being used? Thats actually quite easy to explain. These are 8-core chiplets. It was speculated that the 8-core Ryzen cpu's without integrated graphics would be using 2 4-core chiplets rather than 1 8-core chiplet

It's entirely possible that instead of using 2 dies for a 8 core cpu. Their yeilds on Zen 2 are so good that they can afford to just use a single 8-core chiplet rather than 2 4-core ones and save the 4-core defective chiplets to lower end parts.

There is clearly space on the die for a second CPU chiplet or a GPU chiplet. I would expect for us to see chips in the near future that would use it. Whether those are going to be part of the Ryzen 3000 series is yet to be seen.
 

oldmanbal

2[H]4U
Joined
Aug 27, 2010
Messages
2,319
Interesting... they went with a chiplet design, with 1/4 I/O die, as I suspected...

...but this ES only has one chiplet, despite the clear room for a second chiplet.

I wonder if AMD plans to release the 8 core, single chiplet first... and a 12 and 16 dual chiplet (and/or 8c + Vega GPU) later. Or if the best working engineering samples are only the single chiplet ones, for now.

Maybe they've decided not to cannibalize their Threadripper product stack, since there's no competition from Intel to warrant a serious jump in cores.
 

alxlwson

You Know Where I Live
Joined
Aug 25, 2013
Messages
7,538
I wonder if separating IO from the compute section allows them to crank the clocks up and not have to deal with the extra heat from the IO section>?
 

Pieter3dnow

Supreme [H]ardness
Joined
Jul 29, 2009
Messages
6,785
I wonder if separating IO from the compute section allows them to crank the clocks up and not have to deal with the extra heat from the IO section>?

You say that because you know how high Intel clocks their chips. The designs and the process could be vastly different Intel has been with 14nm for 5 years now? That process has been golden for them the last few improvements.

Between design and process difference it is incredibly hard to tell why when where what with frequency in mind until both AMD and Intel engineers compare notes.
 

DuronBurgerMan

[H]ard|Gawd
Joined
Mar 13, 2017
Messages
1,340
AdoredTV said launch. And WCCFTECH said launch. And several sites above the hype train said launch

https://gadgets.ndtv.com/laptops/ne...ces-2019-launch-rumours-16-core-9-cpu-1972085

AdoredTV put out a video specifically saying ANNOUNCED, not launch. He called this out, specifically. I still think it wasn't even much of an announcement, though. More like a teaser. So he's kinda-sorta wrong on that.

And I don't care what Wccftech says. It's a rag. If they were a print publication, I'd use it for toilet paper.
 

Sikkyu

I Question Reality
Joined
Jan 21, 2010
Messages
2,878
I wonder if separating IO from the compute section allows them to crank the clocks up and not have to deal with the extra heat from the IO section>?
It allows them to bin the cores without worrying about good cores and bad IO. You can bin good cores and good IO without any issues at all and make way more on it.
 

rtangwai

[H]ard|Gawd
Joined
Jul 26, 2007
Messages
1,369
Back in my day we had 1 core, 1 thread and cooled everything with 8 fins of aluminium without these fancy "fans". Lot of innovation since my 2700+ days.

You're making me feel old - I remember when Intel put a heatsink on the 486DX, it was considered sacrilege: "WHAT?!?!?! Heatsink *AND* active cooling?!?!?! If you need that you aren't building the chip right!!!!"
 
Joined
Mar 9, 2004
Messages
3,318
16 core am4 has to mean 64 core thread rippers with IO die coming down the pipe right? I can't have my rig be bested by some B350
 

Neapolitan6th

[H]ard|Gawd
Joined
Nov 18, 2016
Messages
1,114
16 core am4 has to mean 64 core thread rippers with IO die coming down the pipe right? I can't have my rig be bested by some B350
They confirmed 64 core EPYC in the press conference so I'd imagine Threadripper would be an option if they wanted to.
 

Meaker

Official representative of powernotebooks.com!
Joined
Jan 10, 2004
Messages
924
Back in my day we had 1 core, 1 thread and cooled everything with 8 fins of aluminium without these fancy "fans". Lot of innovation since my 2700+ days.

2700+ would have melted through the motherboard with a fan less little block. Had an slk-800 on my 2500+ Barton core ;)
 

mikeo

Gawd
Joined
May 17, 2006
Messages
622
You say "only 8c/16t," its not long ago that 4c/8t was the best desktop chip you could get. How far we have come when 8c/16t is considered to be nothing special.

If more games could make use of my 6c/12t 3930k from 2012 that would be great lol.
 

CentroX

n00b
Joined
Feb 2, 2017
Messages
47
People really expected amd to launch ryzen 3000 series this soon? It was 8 months ago that 2000 series was launched. I cannot believe anyone thinking this was q1. I expected q4 and got q2.

Intel wont be able to do something amazing in 5 months.
 

Galvin

2[H]4U
Joined
Jan 22, 2002
Messages
2,697
Intel is already ready and anyone who put faith in AdoredTV is an idiot.

Intel doesn't have anything right now. All I been reading is they'll have something for the holidays. Which places us around thanks giving or later. So 11 months from now
 

steakman1971

2[H]4U
Joined
Nov 22, 2005
Messages
2,433
AMD is quite the tease. I'm still waiting...the 2700x is still a compelling option!
I also decided to pick up some AMD stock recently. Not that it means anything to the rest of the world, but I'm bullish on them. I'm thinking their bad news is behind them.
 
  • Like
Reactions: N4CR
like this

velusip

[H]ard|Gawd
Joined
Jan 24, 2005
Messages
1,577
Apparently the successor to silicon is silicon plus glue.

Intel didn't see that coming.
 

psyclist

Gawd
Joined
Jan 25, 2005
Messages
844
Leather was a funny underhanded dig at Huang. They are doing single chiplet design first because of supply constraints I bet. The first priority is Epyc using 4 of em and then desktop follows, so keeping that constrained to 1 chiplet for now makes the most sense for early 7nm yields, and looks to be at parity in performance parity with Intels top offerings! This also allows them to get them out to the most amount of customers growing that AMD userbase . They will lay the smack down later on with a second chiplet, once production and yields ramp. Good play id say, and gives them something new to release later on. Looking forward to the review guys!
 

KazeoHin

[H]F Junkie
Joined
Sep 7, 2011
Messages
8,242
Leather was a funny underhanded dig at Huang. They are doing single chiplet design first because of supply constraints I bet. The first priority is Epyc using 4 of em and then desktop follows, so keeping that constrained to 1 chiplet for now makes the most sense for early 7nm yields, and looks to be at parity in performance parity with Intels top offerings! This also allows them to get them out to the most amount of customers growing that AMD userbase . They will lay the smack down later on with a second chiplet, once production and yields ramp. Good play id say, and gives them something new to release later on. Looking forward to the review guys!
I think that AMD laid the breadcrumbs, they can't give tons of details this early as they aren't exactly on schedule and don't want to Osborne themselves. So they make it clear, VERRY Clear that they are showing the performance of their 8-core,16-thread part. This part matches performance with Intel's 8/16 part, only using a lot less power. Then Lisa shows a naked CPU with OBVIOUS room for exactly 1 more Chiplet. The power budget is there, the physical room is there, and the design allows for it...

So without saying anything about release specs, AMDs basically teased >8 core AM4 chips.
 

variant

Gawd
Joined
Feb 17, 2008
Messages
906
What AdoredTV said is that his sources said they did not know about an IO die for Ryzen 3000 .


AdoredTV put out a video specifically saying ANNOUNCED, not launch. He called this out, specifically. I still think it wasn't even much of an announcement, though. More like a teaser. So he's kinda-sorta wrong on that.

And I don't care what Wccftech says. It's a rag. If they were a print publication, I'd use it for toilet paper.

It's juanrga being juanrga.
 

Qthulu

Weaksauce
Joined
Apr 19, 2018
Messages
103
Did you seriously expect AMD to launch Matisse (desktop Ryzen 3000 Series) at CES?

I have been saying that whole time that the launch is probably going to happen sometime around May.
We've already heard the x570 chipset will be released around may so the 3000 series at that time makes sense.
 
Top