Zen 2 samples are with AMD, pumping away with 8C/16T at 4.5GHz .. TweakTown

JPB

Diamond Member
Jul 4, 2005
3,856
18
91
#1
https://www.tweaktown.com/news/63355/amds-next-gen-zen-2-cpu-samples-rtg/index.html

The first Zen 2 samples are with AMD, pumping away with 8C/16T at 4.5GHz.

AMD is preparing its third-gen Ryzen CPU family that will be a true next-gen Zen 2 CPU core, with the original Zen architecture powering both the Ryzen 1000 and Ryzen 2000 series processors. According to a forum post on HardOCP by '_mockingbird', Radeon Technologies Group has "just received its first Zen 2 sample (to optimize for) and it's really impressive".



RTG is a weird place for the very first Zen 2 CPU engineering sample to go, but the posted said in a reply "Zen team brought the processor over so that the RTG teams can make changes to the video card's drivers that make them work better with the processor". This is good news for gamers as it would hopefully mean AMD is keen to see just how their next-gen Zen 2 architecture is holding up with gaming on a prototype Ryzen 3000 series CPU.

The test rig was using a custom motherboard with an AMD logo on it, with the HardOCP forum member saying the "Zen team is being awfully secretive, that's for sure". I'm not surprised given it's an unreleased CPU. The poster added: "It crashes all the time. Some of the tests have to be run multiple times because they crashed before finishing". This isn't uncommon: the first prototype of a new processor, with a heavily tuned architecture, on a totally new 7nm node process... crashes happen.

But, if we're expecting at least one of the new Zen 2 based Ryzen 3000 series processors to offer 8C/16T at 4.5GHz, you can see how Intel is running scared and cranking up the frequency on its very first 8C/16T processor (something AMD has had for over 18 months now). We're looking at the Core i9-9900K boasting 8C/16T at up to 4.7GHz.

This could all be someone joking around, but it would make sense that AMD is getting its new Zen 2 prototypes with the new 7nm node up and running and wafers hitting customers like AMD in the coming months. AMD is also set to launch the world's first 7nm GPU with its Vega 20 GPU going into a next-gen Radeon Instinct accelerator with 32GB of HBM2 memory.

Read more: https://www.tweaktown.com/news/63355/amds-next-gen-zen-2-cpu-samples-rtg/index.html
 

Thunder 57

Senior member
Aug 19, 2007
716
218
136
#2
I do wonder if Intel will end up shooting themselves in the foot by disabling HT on all but the i9.
 
Last edited:

StinkyPinky

Diamond Member
Jul 6, 2002
6,413
101
126
#3
Why would it be a weird place to send it to? Sounds logical to me. You'd want the GPU/CPU teams to be working together.
 

piesquared

Golden Member
Oct 16, 2006
1,650
107
136
#4
I don't think it's much of a surprise or weird at all that their GPU business is getting first samples of Zen2. That seems like they should be one the first to get a sample. We already know that the CPU team has had one in the lab for a couple months now.
 

Thunder 57

Senior member
Aug 19, 2007
716
218
136
#6
That's the 2019 update. HT enabled at each tier. ;)
I'm tired of the extreme segmentation crap. I am almost certainly going with a 6C/12T Zen 2 next year. Then in a few years I should be able to throw a Zen 3 8C/16T. Intel is really off-putting and I thought that they may change their tune some. Sure they have added more cores, but it's the same games elsewhere.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
227
96
#7
I'm tired of the extreme segmentation crap. I am almost certainly going with a 6C/12T Zen 2 next year. Then in a few years I should be able to throw a Zen 3 8C/16T. Intel is really off-putting and I thought that they may change their tune some. Sure they have added more cores, but it's the same games elsewhere.
This one does seem like a missed opportunity on Intels part. It should cost Intel the same with or without HT enabled, and since they don't do both except at 8 cores, it isn't like they get more segments for offering both options at most core counts.

So really they should have just given it everywhere to be more competitive, but then there would be no i9-9900 if the i7-9700k had HT.
 

arandomguy

Senior member
Sep 3, 2013
470
16
116
#8
In terms of a business stand point it's due to the staggered release schedule from both on the consumer end (almost perfectly so at roughly 6 months lead for each side). Due to that neither is fully competing from a generation to generation standpoint.

Zen needed to look good (in terms of capabilities and pricing) against Kabylake. Coffeelake needed to look good against Zen. Zen+ needed to look good against Coffeelake. Coffeelake-R needs to look good against Zen+. Zen 2 needs to look good against Coffeelake-R. Whateverlake needs to look good against Zen 2.

Until Zen 2 an Intel 8c/16th (even without the supply issue) will be appealing even at a significant premium over Zen+. Either they greatly increase i7 pricing to take advantage of it or use the i9 branding. In practice either way it'd cost more than 6c/12t because it can sell for more.
 
Feb 23, 2017
597
544
106
#9
Such a delayed reaction.
That thread was posted nearly 3 weeks ago, and the sample has since been returned.
The thread even jokes about people linking it as new info even after it's mentioned in the thread that it has been returned.
Seems as though people get dazzled by the headlights, and all of a sudden they are being probed by aliens.
 
Mar 11, 2004
18,598
896
126
#10
https://www.tweaktown.com/news/63355/amds-next-gen-zen-2-cpu-samples-rtg/index.html

The first Zen 2 samples are with AMD, pumping away with 8C/16T at 4.5GHz.

AMD is preparing its third-gen Ryzen CPU family that will be a true next-gen Zen 2 CPU core, with the original Zen architecture powering both the Ryzen 1000 and Ryzen 2000 series processors. According to a forum post on HardOCP by '_mockingbird', Radeon Technologies Group has "just received its first Zen 2 sample (to optimize for) and it's really impressive".



RTG is a weird place for the very first Zen 2 CPU engineering sample to go, but the posted said in a reply "Zen team brought the processor over so that the RTG teams can make changes to the video card's drivers that make them work better with the processor". This is good news for gamers as it would hopefully mean AMD is keen to see just how their next-gen Zen 2 architecture is holding up with gaming on a prototype Ryzen 3000 series CPU.

The test rig was using a custom motherboard with an AMD logo on it, with the HardOCP forum member saying the "Zen team is being awfully secretive, that's for sure". I'm not surprised given it's an unreleased CPU. The poster added: "It crashes all the time. Some of the tests have to be run multiple times because they crashed before finishing". This isn't uncommon: the first prototype of a new processor, with a heavily tuned architecture, on a totally new 7nm node process... crashes happen.

But, if we're expecting at least one of the new Zen 2 based Ryzen 3000 series processors to offer 8C/16T at 4.5GHz, you can see how Intel is running scared and cranking up the frequency on its very first 8C/16T processor (something AMD has had for over 18 months now). We're looking at the Core i9-9900K boasting 8C/16T at up to 4.7GHz.

This could all be someone joking around, but it would make sense that AMD is getting its new Zen 2 prototypes with the new 7nm node up and running and wafers hitting customers like AMD in the coming months. AMD is also set to launch the world's first 7nm GPU with its Vega 20 GPU going into a next-gen Radeon Instinct accelerator with 32GB of HBM2 memory.

Read more: https://www.tweaktown.com/news/63355/amds-next-gen-zen-2-cpu-samples-rtg/index.html
Wasn't mockingbird the guy that said Vega was delayed "indefinitely" and I think even claimed or insinuated that it was full on cancelled (and never released to market) on here?

Not that I'm terribly surprised or there's anything particularly noteworthy there, and he was partially right about some things, so it is possible he knows someone at RTG. Well I guess if its doing all core/thread at 4.5GHz on early silicon that would be fairly noteworthy (normally aren't early samples often lower clocked? Like wasn't Zen 1 early samples at like 3/3.2GHz?). I won't take it as gospel, but RTG having one wouldn't be surprising at all.

Here's that Mockingbird thread:
https://forums.anandtech.com/thread...a-indefinitely-delayed.2504201/#post-38853390

Sounds more reasonable (basically that the price of Vega made it near impossible to sell for $499 and less so they were reconsidering it as the performance wasn't competitive with the 1080Ti and the lower priced 1080). I think they were also right about the other Vega versions being delayed/cancelled. There was I believe supposed to be a 14nm refresh of Vega that got canned to focus on the 7nm Vega 20 instead. So they very well do know someone working at RTG. Cryptocurrency changed Vega from a failure to a success for AMD, although it hurt their rep in the consumer gaming market.

Why would it be a weird place to send it to? Sounds logical to me. You'd want the GPU/CPU teams to be working together.
Yeah, I don't think it's odd, especially since RTG is doing all their custom/embedded stuff (so consoles for instance). I think quite a few of the engineers that worked on Zen are there now too. So it could be prepping dev kits for the next consoles for instance.

I believe mockingbird was pretty consumer focused too, so the talk about this and gaming might not be on point, and this might actually be more about testing their new interconnect aimed at helping GPUs (that will be mostly there for the HPC market, with Vega 20 being the first GPU to have it, and Zen 2 the first CPU).
 
Last edited:

coercitiv

Diamond Member
Jan 24, 2014
3,321
822
136
#11
Wasn't mockingbird the guy that said Vega was delayed "indefinitely"
Yes it was, then came back after Vega launched to say "see I was right", as if "indefinitely" might as well have meant anything between 3-4 months and completely canceled.

Imagine someone came on this forum in April and announced Turing is "indefinitely" delayed. It would have been technically true, but of little if any value.

PS: I don't know what's going on at TweakTown, but after that Turing review and this piece of AMD fluff I think they're ready for a "wccftech" quality tag, which is funny considering wccftech is trying to get away from their old bad habits.
 
Last edited:
Mar 10, 2004
28,522
238
126
#12
Yeah, WCCF is almost starting to look reliable in comparison to some sites.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
227
96
#13
Yeah, WCCF is almost starting to look reliable in comparison to some sites.
It's not like they didn't also cover this one:
https://wccftech.com/amd-zen-2-ryzen-8-core-16-thread-cpu-leak/

We live in a click bait world, so small forum posts becomes a big rumor news story.

It's like the recent Polaris 30 story. It comes from a 3 line forum post on Chiphell.

Seeing the difference between the original forum posts, and how big the word count is inflated in the final story, gives you a good idea how much the stories were embellished.
 
Aug 17, 2013
1,804
27
106
#14
This is pretty exciting news if true..
I will be more then happy to upgrade to one of these at release.
Pretty cool to see tech advancing at such a fast rate, but wonder how much longer AMD will cling on to the X86 arch??
Wonder what will be the next big thing after Zen?
 

scannall

Golden Member
Jan 1, 2012
1,402
253
136
#15
This is pretty exciting news if true..
I will be more then happy to upgrade to one of these at release.
Pretty cool to see tech advancing at such a fast rate, but wonder how much longer AMD will cling on to the X86 arch??
Wonder what will be the next big thing after Zen?
As long as Microsoft Windows is primarily x86, then Intel and AMD are locked into that.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
227
96
#16
As long as Microsoft Windows is primarily x86, then Intel and AMD are locked into that.
Also in x86, AMD has only one real competitor, in ARM, they would have MANY.

So AMD should hope that x86 keeps being a the huge market with only one competitor, because they would have a much rougher time in ARM, where every niche already has multiple competitors.
 

EXCellR8

Diamond Member
Sep 1, 2010
3,126
120
126
#17
Is it possible that RTG is experimenting with onboard graphics?

Regardless, I am planning to build with Zen 2; I think I've decided not to invest in a 'ripper build since I don't have any practical use for it. Sign me up for a 3700 though.
 

SPBHM

Diamond Member
Sep 12, 2012
4,843
69
126
#18
Is it possible that RTG is experimenting with onboard graphics?
they are probably busy working on the consoles chip, people expect a new gen of consoles late 2019 or 2020
 

Enigma-

Junior Member
Feb 1, 2017
9
4
51
#19
First I would like to point out that 15% IPC is very possible for AMD to extract. Zen is having ALOT of low hanging fruit. Compared to Zen, AMD will focus on extract higher integer execution capability and I think we can see changes or enhancements like this if we look just at the execution unit of the Matisse Zen2:

* Caches in Zen is quite good already, but some enhancements is to be seen to latency and L3

* Enhanced branch and fetch unit

* More fully symmetrical ALU's to support more branch/SMO/CRC operations as Zen already having PLENTY of power for floating point calculations (except AVX niché)

* Higher entry Physical Register File, and the Retire Queue is already plenty as it is I think.

* Enhanced Stack Engine to reduce the need of the quite already busy AGU's in Zen. That should knock of some bottlenecks. I find AMD going with 3 AGU's unlikely in Zen2.

* Very likely to see symmetric load/store capability and more aggressive prefetch


On the topic of clock speed, I would not expect something lower than 5GHz with XFR which would be 15% higher frequency compared to 2700X. If IPC and clock speed scales toghether, we could expect around 30% higher performance. Zen2 on TSMC 7nm is not to be underestimated and very dangerous for Intel.
 
Last edited:

french toast

Senior member
Feb 22, 2017
943
59
106
#20
Personally I would be very happy with single ccx 8/16, improved topology/IMC/caches/memory support.
15% integer PPC, double width SIMD (4x256bit)...~4.8ghz xfr2.
<100w.

I think that is achievable with new architecture and process, i would be disappointed with anything less than that, that would be underachieving imo.
 

Hans Gruber

Senior member
Dec 23, 2006
539
62
116
#21
When Ryzen was released I read that Zen 2 is a significant improvement beyond the die shrink of the CPU. It's supposed to be far more significant than Zen+ was. So an IPC improvement well beyond 15% is highly likely.
 

PeterScott

Platinum Member
Jul 7, 2017
2,605
227
96
#22
When Ryzen was released I read that Zen 2 is a significant improvement beyond the die shrink of the CPU. It's supposed to be far more significant than Zen+ was. So an IPC improvement well beyond 15% is highly likely.
I doubt it. 15% IPC boost on a modern x86 processor is a massive boost. Expecting well beyond that is wishful thinking.
 
Aug 11, 2008
10,457
67
126
#23
Also in x86, AMD has only one real competitor, in ARM, they would have MANY.

So AMD should hope that x86 keeps being a the huge market with only one competitor, because they would have a much rougher time in ARM, where every niche already has multiple competitors.
Yea, always makes me laugh when I see the glee the AMD camp has when some article comes out about how ARM is pushing "intel" out of the server market or Apple is outpacing "intel" in progress, not bothering to think AMD is in the same boat.
 

itsmydamnation

Golden Member
Feb 6, 2011
1,840
223
136
#24
Yea, always makes me laugh when I see the glee the AMD camp has when some article comes out about how ARM is pushing "intel" out of the server market or Apple is outpacing "intel" in progress, not bothering to think AMD is in the same boat.
Never seen this personally, at the same time amd is probably a big reason arm isn't getting a bigger foot hold. All the benifits arm is bring vs Intel amd is already bringing and they beat arm to market.

Now depending how long it takes Intel to get a 10nm high core count chip out vs how long it takes an arm vendor to get a 7nm chip out this dynamic could change, but to paint the risks/rewards in the data entire space as comparable between amd and Intel is to ignore reality.
 

JDG1980

Golden Member
Jul 18, 2013
1,656
172
136
#25
Yea, always makes me laugh when I see the glee the AMD camp has when some article comes out about how ARM is pushing "intel" out of the server market or Apple is outpacing "intel" in progress, not bothering to think AMD is in the same boat.
ARM server chips have been tried multiple times, but despite several well-funded attempts by major corporations like Qualcomm, there has been no success yet. I don't see that changing. The one company that might have the credibility and resources to push ARM-based servers is Apple, but they haven't shown any interest in doing so as far as I can tell.

Apple's current chips compete with Samsung and Qualcomm, not Intel or AMD. There have been rumors about Apple moving their desktops and laptops to in-house ARM SoCs, but I'll believe that when I see it. Apple is willing to ditch backwards compatibility to some extent, but usually not immediately overnight, which is what such a change would entail.
 

Similar threads



ASK THE COMMUNITY

TRENDING THREADS