New Zen microarchitecture details

Page 61 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

sirmo

Golden Member
Oct 10, 2011
1,012
384
136
Thats really disappointing. Been waiting forever for zen and polaris.
We don't know if they are "skipping" Polaris on Computex (don't think they would invite the press all the way to Macau for just the Bristol Ridge), and Zen was always going to be the end of the year 2016 not middle of the year.
 

CentroX

Senior member
Apr 3, 2016
351
152
116
We don't know if they are "skipping" Polaris on Computex (don't think they would invite the press all the way to Macau for just the Bristol Ridge), and Zen was always going to be the end of the year 2016 not middle of the year.


isn't computex like 3 days away? I thought we'd know for sure what they will show there by now (and not WCCTECH rumors)
 

Tuna-Fish

Golden Member
Mar 4, 2011
1,311
1,420
136
isn't computex like 3 days away? I thought we'd know for sure what they will show there by now (and not WCCTECH rumors)

The invite that went to the press was really vague, and AMD isn't interested in talking more about it. We won't know more until the event starts.
 

majord

Senior member
Jul 26, 2015
433
523
136
You have to put integer loads here too majord. Loads like 7-zip and WinRar are important because most of consumer loads are integer!

I just used whatever was available on bench that scaled well with cores. I think 7zip may be a different version between Anand's 8150 tests, and the i7's so I couldn't include.

So, on majords graph, I'm pitting the release Zeppelin flagship model in heavily ILP code @100-115% 6700K mark, and in most other serially limited code to be average around the 2700K mark.

With Turbo, that is. I also can't see release frequencies being high either. Max base 2.8-3.0GHz for Zeppelin all cores loaded.

I left the raw data at bottom of graph. The theorized zen would be avg 25% faster than a 6700k in MT, not 10-15.

It's funny you say 2.8-3ghz, because if you work it out using the claimed IPC, and afew assumptions about SMT scaling, the performance painted above should be achievable at just that. As a worse case though - I'd bet they won't launch top SKU at anything below 3ghz base, and the max turbo frequency will have a 3 as it's integer digit..

I also quite sure there will be a STAPM style power managment , to push all-core frequences above base

It's probably worth mentioning at this point that, theoretically, you could build an 8 module/16thread Excavator based "xen" CPU on 28nm ( with some L3 thrown in for good measure), and achieve close to,( if not bang on) 3Ghz @ ~95w, thus giving double FX-8150 performance. So really, I think for them not to achieve that kind of performance (double 95w Vishera or FX8150, whatever) at a minimum would be the 'disaster' scenario.. It would mean they would have been better off shrinking Excavator.
 

coffeemonster

Senior member
Apr 18, 2015
241
86
101
It's probably worth mentioning at this point that, theoretically, you could build an 8 module/16thread Excavator based "xen" CPU on 28nm ( with some L3 thrown in for good measure), and achieve close to,( if not bang on) 3Ghz @ ~95w, thus giving double FX-8150 performance.
Haha. I'd love to see what something like that could do.
 

DrMrLordX

Lifer
Apr 27, 2000
21,528
10,714
136
Be careful, all you have to do is bring up 22FDX and Nosta will show up again!

*ducks*

But yes Zen really does have to outdo a hypothetical XV 8m/16t chip to really justify the entire effort. Anything less and it's a bit of a joke.
 

krumme

Diamond Member
Oct 9, 2009
5,952
1,585
136
It's true. I haven't followed computing since I left the field circa 2010. But I'm now working back in computing for the past year. Bear with me whilst I refresh but I do remember some basics


Throwing more cores at a problem like AMDs at the time is never a solution, only a temporary band-aid - a stop gap. I had plenty of private chats with John Fruehe at the time (explaining this).

Let's talk about today and the future. Right now, all the big businesses have moved or are in the middle of migrating to the cloud models; Iaas and Paas mainly (that I'm seeing). Licensing is per Core+GB of RAM and OS dependent. Wintel is FAR cheaper than the rest, I mean 1/50 to 1/3 some others. The major hosting companies right now use 4-8 Cores. Above that is rare, except when you really need AIX boxes or HP-UX. Even if you ask for more cores (VMs) from cloud providers, they won't offer you more than 4 on Wintel platforms. Something that'll be in the contracts.

AMD needs that +90% market Wintel has fortified. In corps, people in purchasing and infrastructure right now don't even know of AMD being available in this [server] market.


New process is always finicky, hence why MFGs tend to use it with matured archs first: To iron out the problems. New process + new complex arch far more so. It's rarely achieved without problems (clocking and power primarily). It's just part of the process learning curve.

But any 8C/SMT chip will if properly gated and modulated present a major thermal budget for n-4/8 Cores to tap into. I suppose the process performance at such MHz will primarily dictate that due to PVT sensitivities but clock jitter, skews and distribution become the limiting factors in conjunction with the delays. It's not so simple to keep multiple high clocking and voltage domains, on large chips, in sync with stability, even if you posses a vast thermal budget. Gosh, even hot spots wreck havoc on gate and wire delays which directly affect the clocking.

Anyone know the type of clock distribution BD/PD uses? Or how many drivers and buffers are used?

What about FO4 delays for any of AMDs recent archs?

Sent from HTC 10
The clock distribution in bd and pd is done a bit differently. Pd uses a resonant clock mesh with ip from Cyclos. Pretty interesting tech.

http://hothardware.com/news/amd-wil...ond-4ghz-using-resonant-clock-mesh-technology

If i remember correctly it was sa that broke the news but perhaps that was behind the paywall.

http://www.semiwiki.com/forum/content/917-clock-design-socs-lower-power-better-specs.html
 
Last edited:

CentroX

Senior member
Apr 3, 2016
351
152
116
Seems like both zen and polaris are disapointments. Intel and nvidia won.

Threadcrapping and trolling are not allowed
Markfw900
 
Last edited by a moderator:

prtskg

Senior member
Oct 26, 2015
261
94
101
But yes Zen really does have to outdo a hypothetical XV 8m/16t chip to really justify the entire effort. Anything less and it's a bit of a joke.
I don't think so. It brings 40% IPC improvement. Hence it'll be more efficient. Even if 14nm LPP don't hit high clocks to make it worthwhile for average consumers, it'll be great for sub 3ghz server market. This is what AMD's life depends upon. Also down the line foundries will have higher performance finfet nodes which will make consumer chips from AMD more competitive.
 

ET

Senior member
Oct 12, 1999
521
33
91
Pray tell, explain to us how they are disappointments?

It's pretty obvious:

- They're not here yet. Not that it was expected that they'd be here, but it's still disappointing. Especially considering that Bristol Ridge is ahead of schedule, and even that is late compared to rumours pegging AM4 motherboards for March. So it's all very disappointing on all levels.

- Their imagined performance isn't as good as the imagined performance of other chips which aren't them, and their real performance is certainly not as good as real performance of other chips, since they're not yet available, which makes their real performance a big zero. When you take imagined pricing into consideration, it's even more disappointing.

- They will come from AMD, which means they won't be Intel or NVIDIA products, which means that they are inferior by design, or at least not as well considered, which is rather disappointing. They will also not be "the way it's meant to be played", won't support PhysX, not be 'Game Ready' and a lot of other stuff that's very important.
 

Dresdenboy

Golden Member
Jul 28, 2003
1,730
554
136
citavia.blog.de
It's pretty obvious:

- They're not here yet. Not that it was expected that they'd be here, but it's still disappointing. Especially considering that Bristol Ridge is ahead of schedule, and even that is late compared to rumours pegging AM4 motherboards for March. So it's all very disappointing on all levels.

- Their imagined performance isn't as good as the imagined performance of other chips which aren't them, and their real performance is certainly not as good as real performance of other chips, since they're not yet available, which makes their real performance a big zero. When you take imagined pricing into consideration, it's even more disappointing.

- They will come from AMD, which means they won't be Intel or NVIDIA products, which means that they are inferior by design, or at least not as well considered, which is rather disappointing. They will also not be "the way it's meant to be played", won't support PhysX, not be 'Game Ready' and a lot of other stuff that's very important.

Are you complaining, that reality might not match your expectations?
 

chrisjames61

Senior member
Dec 31, 2013
721
446
136
It's pretty obvious:

- They're not here yet. Not that it was expected that they'd be here, but it's still disappointing. Especially considering that Bristol Ridge is ahead of schedule, and even that is late compared to rumours pegging AM4 motherboards for March. So it's all very disappointing on all levels.

- Their imagined performance isn't as good as the imagined performance of other chips which aren't them, and their real performance is certainly not as good as real performance of other chips, since they're not yet available, which makes their real performance a big zero. When you take imagined pricing into consideration, it's even more disappointing.

- They will come from AMD, which means they won't be Intel or NVIDIA products, which means that they are inferior by design, or at least not as well considered, which is rather disappointing. They will also not be "the way it's meant to be played", won't support PhysX, not be 'Game Ready' and a lot of other stuff that's very important.

Keep trolling. You are good at it.


Personal attacks are not allowed.
Markfw900
 
Last edited by a moderator:

dark zero

Platinum Member
Jun 2, 2015
2,655
138
106
I don't think so. It brings 40% IPC improvement. Hence it'll be more efficient. Even if 14nm LPP don't hit high clocks to make it worthwhile for average consumers, it'll be great for sub 3ghz server market. This is what AMD's life depends upon. Also down the line foundries will have higher performance finfet nodes which will make consumer chips from AMD more competitive.
I feel that Samsung processes and fabs (I feel that Samsung will help AMD in this one) will make the AMD Zen and Vega 10, while GloFo will do the low tier ones.

Also.. remember that Samsung needs a guinea pig to test their processes at their maximum, so AMD is the chosen one for that. AMD has a wide range of products: ARM K12 (about time to see a 3.0 Ghz+ ARM chip?), X86 Zen and Vega 10....

Are you complaining, that reality might not match your expectations?

I feel that my glooming was bad.... that is worse.... consider that if AMD quits the market the prices will likely skyrocket to new levels... they are literally abusing the customers....
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
Samsung isn't manufactoring anything for AMD on 14nm from the looks of it.

The big issue isn't if they make it on Glofo or Samsung. But that they dont make it on the better TSMC 16FF+.