New Zen microarchitecture details

[deasd]

Seriously? That's what is difficult with these large threads (1500+ posts!) - most people don't read them from the beginning.

I don't force anyone reading 1500+ posts, but if you tell me you don't read the thread's starter, the first post of this thread before you comment, then I don't understand.
http://forums.anandtech.com/showpost.php?p=38065073&postcount=1

 

[JDG1980]

Lisa Su is an incredibly bad presenter. When she was discussing Zen at Computex it sounded like she was delivering the news that someone had died.

On the plus side: working silicon!

 

[Shakabutt]

They confirmed 8 cores / 16 threads, thats kinda exciting, and 40% IPC. Its looking great

 

[exar333]

Lisa Su is an incredibly bad presenter. When she was discussing Zen at Computex it sounded like she was delivering the news that someone had died.

On the plus side: working silicon!

Yeah, the 'Zen is alive and on track' doesn't inspire much enthusiasm. Was expecting more on Zen than a footnote. Hoping we get more concrete info soon.

 

[therealnickdanger]

40% IPC 8c16t? Yes pls.

 

[Elixer]

I AM ZEN.
Um.. Yeah.
Can they get anymore vague?

 

[ZGR]

I have a feeling the 8c/16t Zen will be priced as a premium product, competing with the 5960X/6900X in performance and price.

This won't be cheap. Would be awesome if it was. I hope IPC and clockspeed are comparable to at least Haswell-E.

 

[Justinbaileyman]

Lisa Su, talked about summit ridge and Zen but nothing was mentioned about Bristol Ridge?? No early Am4 motherboards like everyone thought either. I guess we have to wait till the end of the year Only Fx-7th/gen and A8,A10,and A12 Mobile apu's. I am pretty pissed off right now!! Total bullshit...


cussing is not alllowed in the technical forums
Markfw900

 

[csbin]

 

[exar333]

I have a feeling the 8c/16t Zen will be priced as a premium product, competing with the 5960X/6900X in performance and price.

This won't be cheap. Would be awesome if it was. I hope IPC and clockspeed are comparable to at least Haswell-E.

I guess...I have splurged on AMD CPUs before, but how the heck do they go from ~$99-150 CPU all of a sudden to $1000? If it is competitive with Intel's 6C/12T and/or 8C/16T parts, and is not $800+, I will get one...

 

[dark zero]

HOLY COW!

Ok, jokes aside, this is very interesting. AMD Zen is about to come sooner than expecting... so AMD is about to be ready to go with everything?

 

[Stuka87]

Ok, jokes aside, this is very interesting. AMD Zen is about to come sooner than expecting... so AMD is about to be ready to go with everything?

I would say that its coming exactly when it was expected too, not sooner.

 

[ET]

Zen is never late, nor is it early. It arrives precisely when it means to.

 

[Dresdenboy]

Zen is never late, nor is it early. It arrives precisely when it means to.

+1
I had the same thought.

 

[JimmiG]

They confirmed 8 cores / 16 threads, thats kinda exciting, and 40% IPC. Its looking great

Everyone keeps bringing up the 40% IPC improvement as if it's a good thing, but I'm feeling it might not be enough.

http://www.anandtech.com/bench/product/698?vs=836
Just multiply the AMD numbers by 1.4. This is not enough to even match Haswell from 2013. There would have to be a 60% improvement. They'd still be at roughly Sandy Bridge performance from 2011, or slightly below. Makes you wonder what the AMD engineers have been doing for the past 5 years if they can't catch up with a nearly stationary target.

I would love to switch to AMD next time I upgrade, but it should not be a step back in single-threaded performance compared to my 4770K from 2013. Of course, if Zen reaches 5+ GHz like Vishera, the gap would close, but probably at the cost of much higher power consumption and heat.

Also, did anyone really expect a proper launch of Zen? I thought it was common knowledge that Zen and Vega would be revealed later.

 

[nenforcer]

Everyone keeps bringing up the 40% IPC improvement as if it's a good thing, but I'm feeling it might not be enough.

http://www.anandtech.com/bench/product/698?vs=836
Just multiply the AMD numbers by 1.4. This is not enough to even match Haswell from 2013. There would have to be a 60% improvement. They'd still be at roughly Sandy Bridge performance from 2011, or slightly below. Makes you wonder what the AMD engineers have been doing for the past 5 years if they can't catch up with a nearly stationary target.

I would love to switch to AMD next time I upgrade, but it should not be a step back in single-threaded performance compared to my 4770K from 2013. Of course, if Zen reaches 5+ GHz like Vishera, the gap would close, but probably at the cost of much higher power consumption and heat.

Also, did anyone really expect a proper launch of Zen? I thought it was common knowledge that Zen and Vega would be revealed later.

It's suppose to be 40% > than the most recent CPU stepping released which is Excavator. Other reports have said that a single Zen core is twice as fast as a Piledriver core. This would put them just behind the new Broadwell-E I believe.

No one expected a ZEN release - this has always been a Q4 2016 product availability. There were, however, quite a few people (myself included) expecting a desktop Carrizo APU launch alongside some new AM4 motherboards which were no where to be seen. The 7th generation APU are only going in mobile parts with < 35W TDP.

 

[KTE]

'Which diagram'? LOL, you really don't know this thread was started by Dresdenboy. Go to 1 page and post count 1. (hard to believe people replying without reading title and contents from start)

I wouldn't do you the disservice of replying without reading the first page..

But I definitely haven't read a lot of the thread or any related thread as you could probably understand.

I thought there was some other diagram I'm missing here which somehow showed it wasn't designed for high clocks/IPC-ILP monster...

Sent from HTC 10

 

[KTE]

Everyone keeps bringing up the 40% IPC improvement as if it's a good thing, but I'm feeling it might not be enough.

http://www.anandtech.com/bench/product/698?vs=836
Just multiply the AMD numbers by 1.4. This is not enough to even match Haswell from 2013. There would have to be a 60% improvement. They'd still be at roughly Sandy Bridge performance from 2011, or slightly below. Makes you wonder what the AMD engineers have been doing for the past 5 years if they can't catch up with a nearly stationary target.

I would love to switch to AMD next time I upgrade, but it should not be a step back in single-threaded performance compared to my 4770K from 2013. Of course, if Zen reaches 5+ GHz like Vishera, the gap would close, but probably at the cost of much higher power consumption and heat.

Also, did anyone really expect a proper launch of Zen? I thought it was common knowledge that Zen and Vega would be revealed later.

40% average IPC gain is so rare...

Sent from HTC 10

 

[deasd]

I wouldn't do you the disservice of replying without reading the first page..

But I definitely haven't read a lot of the thread or any related thread as you could probably understand.

I thought there was some other diagram I'm missing here which somehow showed it wasn't designed for high clocks/IPC-ILP monster...

Sent from HTC 10

If the diagram of Dresdenboy's one is closed to what AMD do, Zen has to be a design which is aim for ILP/IPC, there's nearly 0 possibility that implementing high clocks/high latency such as Netburst/Power6 did, it would even has lower latency than BD/SR/EXV core.
Since the process node today stall at around 4Ghz, it makes no sense design target higher than this. OTOH more ILP/IPC might use much more complex logic, thus more pipeline stages, but it still far from being called as high clock design.

40% average IPC gain is so rare...

Sent from HTC 10

I think he confused '40% IPC' and '40% performance'.

 

[Dresdenboy]

Sharper pic of the chip at Computerbase:

http://www.computerbase.de/2016-06/amd-zen-gezeigt/

I read:
"ZPAA000121"

This is an early ES, not comparable to the usually distributed ones marked with "ZD.." or "2D.." for desktop.

"ZP" could mean "Zeppelin" (which is assumed to be 16C, but who knows), or also simply "Zen Processor". "ZP" has also been used in some AMD patches.

"AA" or after "ZPA" the "A0" might stand for the stepping, and "121 is just a serial number, I think.

 

[The Stilt]

Zeppelin is the name of the die (like Orochi). Currently there is only one stepping (A0) and I have no idea if there will be a n + nd/rd/th one.

 

[Dresdenboy]

Zeppelin is the name of the die (like Orochi). Currently there is only one stepping (A0) and I have no idea if there will be a n + nd/rd/th one.

This would mean, the APU with ZP + GL will have 8 cores. This leaves more room (thermally and physically) for the GPU. OTOH the slide showed 4 DDR4 channels.

Someone at reddit mentioned there were no pins... :sneaky:. So did Lisa show the small datacenter Zeppelin CPU?

 

[cytg111]

This is my main grief with building a rig right now. And I need a new rig very soon.
If Zen brings a somewhat competetive 8 core mainstream -> Cannonlake will likely bring 6 cores mainstream.
That equals that any quad you're building today is gonna be 2 cores short of a full deck in a years time.

 

[Shamrock]

Don't know if this is worth anything, but AMD updated their cover photo on Facebook.

 

[The Stilt]

This would mean, the APU with ZP + GL will have 8 cores. This leaves more room (thermally and physically) for the GPU. OTOH the slide showed 4 DDR4 channels.

Someone at reddit mentioned there were no pins... :sneaky:. So did Lisa show the small datacenter Zeppelin CPU?

Zeppelin has two memory channels and the AM4 socket itself support only two, obviously.

Even mechanical sample CPUs have pins, so if the CPU Su showed to the public had none... Could have been a AM3 or AM4 mock-up. The package size itself is identical, while the pin count, pitch and gridding is different.

And yes, AM4 is PGA not LGA.

Also there is no way you can confuse AM4 to either of the server variants :sneaky: