New Zen microarchitecture details

Page 158 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

leoneazzurro

Golden Member
Jul 26, 2016
1,099
1,825
136
OK, I'll take that but it anyway says it is higher. As I told before, it must be seen, how the tests are done and in which conditions. It is an ES.
 

KTE

Senior member
May 26, 2016
478
130
76
Here we are talking of the minimum Voltage for a chip to work.
Anyway this chip draw 330mW at 2.41GHz and <1W at 4Ghz.
It has 1/4 of throughput than Zen FPU, so I estimated a full Zen core is 16x this chip, leading to <16W/core @4ghz. And we are close to reality...
Power/clocks/voltage NEVER scale linearly.

Seriously. I mean cmon

So all the rest is wrong due to the inherent incorrect assumptions.

But I will make sure to clock 1 core and measure the current/voltage fully loaded at 4GHz. I will make sure to test and report how close or far your 16W conjectures are

Sent from HTC 10
(Opinions are own)
 

Abwx

Lifer
Apr 2, 2011
11,728
4,647
136
OK, I'll take that but it anyway says it is higher. As I told before, it must be seen, how the tests are done and in which conditions. It is an ES.

You can believe me, while i m at it i would add that a good translation would be "for sure".

Assuming all converters have 90% efficency his sample was consuming 93 x 0.9 = 83.7W@3.3GHz since he did the measurement at the 12V rail that feed the CPU.

AMD s demo also showed 93W, and certainly with more accurate measurement gear.
Given that it s measured at the main the CPU did drain 93 x 0.9 x 0.9 = 75.33W@3.4GHz.

Conclusion is that Canard PC sample has 7% higher voltage at same frequency than the sample used by AMD, wich is not surprising since ESs are using higher voltage margins than final products, in this case the added voltage margin is 7%, wich is quite reasonable for an early ES.
 

leoneazzurro

Golden Member
Jul 26, 2016
1,099
1,825
136
The tweet I'm referring to is speaking about the 3.6/4.0 GHz sample, not the 3.4 GHz. So I don't understand what you are referring to.
 
Reactions: Doom2pro

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
hmm, would 3.4 = lowest base clock translate to all Ryzen cores? I thought this was just for the 8c part?

3.4 @65W would be amazing, especially if we are looking at ~4.2-4.5W @, I dunno, 120W on that 6c part?
Its just guesses for my part but as i recall 3.4 wasnt mentioned directly in relation to the 8c part but for zen in general. We know we have a 32c 180w tdp server variant incomming q2. Its like 45w for 8c. So it seems to me they are at the edge of freq and that it comes at high efficiency cost. But no matter what its imo crazy what we get here anyway...and its nitpicking where we are. Tpd is certainly well within what is acceptable in a desktop. By far imo. Prices are far more important.
 

zinfamous

No Lifer
Jul 12, 2006
111,530
30,729
146
Its just guesses for my part but as i recall 3.4 wasnt mentioned directly in relation to the 8c part but for zen in general. We know we have a 32c 180w tdp server variant incomming q2. Its like 45w for 8c. So it seems to me they are at the edge of freq and that it comes at high efficiency cost. But no matter what its imo crazy what we get here anyway...and its nitpicking where we are. Tpd is certainly well within what is acceptable in a desktop. By far imo. Prices are far more important.

Well, if that 6c comes out near those clocks and @ $200-300, I'll buy one. ....eh, but I'll wait. I don't like first gen stuff.
 

Abwx

Lifer
Apr 2, 2011
11,728
4,647
136
The tweet I'm referring to is speaking about the 3.6/4.0 GHz sample, not the 3.4 GHz. So I don't understand what you are referring to.

https://twitter.com/CPCHardware/status/818428797427052545
Current Ryzen CPUs exceed effectively 95W TDP , (wich is) logical given the results we got on ours tests.

So nowhere he said that he has a 3.6-4.0 sample, the tests he s talking about are the one he published in their latest edition...

OTT he just stated that AMD has delivered B stepping (that he say is the final stepping) 8C/8T samples, and that it s possible that they could find their way as final products as well...

https://twitter.com/CPCHardware/status/818932115270209537
 
Last edited:

bjt2

Senior member
Sep 11, 2016
784
180
86
Power/clocks/voltage NEVER scale linearly.

Seriously. I mean cmon

So all the rest is wrong due to the inherent incorrect assumptions.

But I will make sure to clock 1 core and measure the current/voltage fully loaded at 4GHz. I will make sure to test and report how close or far your 16W conjectures are

Sent from HTC 10
(Opinions are own)

I said core only. Then there is NB, MC and SB...

And power does not scale linearly with frequency, but scales linearly with the core. Because dissipation area is proportional to core number. It's under linear only for limitations of VRMs and tracks...

If you are implying that my forecast is exaggerated, think that 65W Excavator APU with 4 cores at 4/4.2 draw 65W... 16x4=64... At 4.0GHz. I'll give 1W to NB, SB, MC and GPU. On the 28nm BULK. We already discussed that: AMD said that Zen draw same energy/core/clock.

I multiplied for 16 (4x4) the consumption of a NEON FPU: 4 to infer Zen FPU power draw and 4 supposing that a core draw 4x of its FPU (and it's not true)...
 

leoneazzurro

Golden Member
Jul 26, 2016
1,099
1,825
136
He said some time time ago that he had a ES 3.6/4.0 with F4 stepping, it's written previously in this thread.
 

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
Well, if that 6c comes out near those clocks and @ $200-300, I'll buy one. ....eh, but I'll wait. I don't like first gen stuff.
Haha. I have it the same way about new stuff. I will not buy it... But always end up buying it anyway. Except routers. I reject beeing beta tester for some new chipset and crazy buggy bios.
Actually ended up buying a haswell cpu and msi mb so early win 7 couldnt boot without drivers. Drivers that wasnt included with the mb btw.
But am4 is not so new so its probably far more proven stuff.
 

krumme

Diamond Member
Oct 9, 2009
5,956
1,595
136
https://twitter.com/CPCHardware/status/818428797427052545


So nowhere he said that he has a 3.6-4.0 sample, the tests he s talking about are the one he published in their latest edition...

OTT he just stated that AMD has delivered B stepping (that he say is the final stepping) 8C/8T samples, and that it s possible that they could find their way as final products as well...

https://twitter.com/CPCHardware/status/818932115270209537
AMD a commencé à sampler des Ryzen 4C B-Step (prod) avec HT désactivé. Possible qu'on les retrouve dans la gamme commerciale.

And

"

@Ceralion1 It's all about yield and I have very few information about it. Pure speculation, but mass availability seems hard before may."
 

Abwx

Lifer
Apr 2, 2011
11,728
4,647
136
He said some time time ago that he had a ES 3.6/4.0 with F4 stepping, it's written previously in this thread.

He said that there are already 3.6/4.0 with F4 stepping running, but he didnt state that he has one, more surely that he s aware of the fact and has a reliable source, the same source that gave him access to an early plateform for a few hours...
 

lixlax

Senior member
Nov 6, 2014
202
190
116
RyZen has managed to exceed excavators IPC with 55%, way better then the estimated 40% presented last summer.

http://www.sweclockers.com/forum/post/16600771
So it should be almost 80% higher than Vishera, their last high end design. That's not too far off what Conroe brought compared to the latest Netburst designs. I wonder if there are going to be only 65W hexa cores or 95W as well.
I'm prbobably going to upgrade to Rysen at some point this year and a 6 core is the most probable core count for me (just interested because of that).
 

Ajay

Lifer
Jan 8, 2001
16,094
8,111
136
Yep. So 3.4 3.8 for the 6c 65w tdp part? Wasnt 3.4 said to be the lowest base clock?
Pretty darn low tdp for 6 cores...now sell it for 199 usd

Well, again, if the hexacore Zen clocks that well (and is a good overclocker), then I would expect the price to be closer to the i7-6800K (but with a brand discount, maybe -$50US). It would be dumb of AMD to sell these cheap, especially as they are currently production limited.
 

imported_jjj

Senior member
Feb 14, 2009
660
430
136
OTT he just stated that AMD has delivered B stepping (that he say is the final stepping) 8C/8T samples, and that it s possible that they could find their way as final products as well...

https://twitter.com/CPCHardware/status/818932115270209537

You are mixing things up:
"AMD a commencé à sampler des Ryzen 4C B-Step (prod) avec HT désactivé. Possible qu'on les retrouve dans la gamme commerciale."
So AMD has started sampling B-stepping for 4 cores with HT disabled and it's possible we'll see them in retail.
later on - Le stepping est marqué comme final. Les specs peuvent encore évoluer.
It's the final stepping but specs can evolve.
The bit in English about yields and May is speculation and seems to be about all Ryzen and that is a bit odd.

Anyway AMD said the launch is before end of March ,that there won't be a paper launch and Zen hits retail from day 1 and that it's a "complete stack of chips at release".
So ,contrary to rumors,seems more likely that 4 cores will hit retail from day one too.
Maybe 4C no HT arrive later at a lower price point as it is normal to add more SKUs over time.
 

Abwx

Lifer
Apr 2, 2011
11,728
4,647
136
Ryzen 4C B-Step (prod) avec HT désactivé


You are mixing things up:
"AMD a commencé à sampler des Ryzen 4C B-Step (prod) avec HT désactivé. Possible qu'on les retrouve dans la gamme commerciale."

Maybe 4C no HT arrive later at a lower price point as it is normal to add more SKUs over time.

Right for the core count...

As for pricing it should be in the same range as Bristol Ridge, this latter compensating the lower CPU perfs by an iGPU.
 

Doom2pro

Senior member
Apr 2, 2016
587
619
106
Haha. I have it the same way about new stuff. I will not buy it... But always end up buying it anyway. Except routers. I reject beeing beta tester for some new chipset and crazy buggy bios.
Actually ended up buying a haswell cpu and msi mb so early win 7 couldnt boot without drivers. Drivers that wasnt included with the mb btw.
But am4 is not so new so its probably far more proven stuff.

I don't like buying freshly new hardware either, I also like to wait until bugs have been smoothed out... However I'm making an exception this time. 8c/16t CPU, X370 Motherboard, 2x Polaris OR 2x Vega GPU (depending on release date), and HTC Vive + the rest.

AMD shut up and take my money!
 
Reactions: krumme

AtenRa

Lifer
Feb 2, 2009
14,003
3,362
136
Well, again, if the hexacore Zen clocks that well (and is a good overclocker), then I would expect the price to be closer to the i7-6800K (but with a brand discount, maybe -$50US). It would be dumb of AMD to sell these cheap, especially as they are currently production limited.

Yea im expecting the 6c 12T at close to $299.
 

Atari2600

Golden Member
Nov 22, 2016
1,409
1,655
136
but zen+vega+hbm will be cheaper then intel+crappy igpu?

There will be 3 market segments addressed by the Zen+Vega APU as I see it:

1. Low Graphics / Low Compute requirements || power sensitive (i.e. granny's internet laptop or a secretary's word/excel machine*) ----- Zen + Vega, no HBM (can be served by Intel solution)
2. Med Graphics / Low Compute requirements || power sensitive (i.e. CAD/CAM workstation, ultrabooks) ----- Zen + Vega w/ HBM2
3. Low Graphics / Med Compute requirements (Workstation that forms part of an overnight distributed HPC grid or "isolated" workstation that uses OpenCL applications) ----- Zen + Vega w/ HBM2

2 & 3 cannot be served by intel's current CPU+iGPU solutions. So its a bit of a mix up. In the professional space, you may see IT depts go for the HBM2 option as it gives them more flexibility in their deployments, streamlines their support and makes it simpler for dealing with others (i.e. procurement/admin).


*I think Cray are still working on something that can cope with outlook
 

CatMerc

Golden Member
Jul 16, 2016
1,114
1,153
136
Here's a thought. Maybe the F3 stepping and the F4 stepping are identical in terms of bug fixed, the only difference being clocks.

In that case they could have built up a stockpile of the F3 stepping, while the F4 stepping will be more limited by launch but will serve as the highest end SKU. Overtime F4 will replace the F3 stepping across the entire product range, but initially there will be a large difference in Fmax between top SKU Ryzen and the lower SKU's.
 
sale-70-410-exam    | Exam-200-125-pdf    | we-sale-70-410-exam    | hot-sale-70-410-exam    | Latest-exam-700-603-Dumps    | Dumps-98-363-exams-date    | Certs-200-125-date    | Dumps-300-075-exams-date    | hot-sale-book-C8010-726-book    | Hot-Sale-200-310-Exam    | Exam-Description-200-310-dumps?    | hot-sale-book-200-125-book    | Latest-Updated-300-209-Exam    | Dumps-210-260-exams-date    | Download-200-125-Exam-PDF    | Exam-Description-300-101-dumps    | Certs-300-101-date    | Hot-Sale-300-075-Exam    | Latest-exam-200-125-Dumps    | Exam-Description-200-125-dumps    | Latest-Updated-300-075-Exam    | hot-sale-book-210-260-book    | Dumps-200-901-exams-date    | Certs-200-901-date    | Latest-exam-1Z0-062-Dumps    | Hot-Sale-1Z0-062-Exam    | Certs-CSSLP-date    | 100%-Pass-70-383-Exams    | Latest-JN0-360-real-exam-questions    | 100%-Pass-4A0-100-Real-Exam-Questions    | Dumps-300-135-exams-date    | Passed-200-105-Tech-Exams    | Latest-Updated-200-310-Exam    | Download-300-070-Exam-PDF    | Hot-Sale-JN0-360-Exam    | 100%-Pass-JN0-360-Exams    | 100%-Pass-JN0-360-Real-Exam-Questions    | Dumps-JN0-360-exams-date    | Exam-Description-1Z0-876-dumps    | Latest-exam-1Z0-876-Dumps    | Dumps-HPE0-Y53-exams-date    | 2017-Latest-HPE0-Y53-Exam    | 100%-Pass-HPE0-Y53-Real-Exam-Questions    | Pass-4A0-100-Exam    | Latest-4A0-100-Questions    | Dumps-98-365-exams-date    | 2017-Latest-98-365-Exam    | 100%-Pass-VCS-254-Exams    | 2017-Latest-VCS-273-Exam    | Dumps-200-355-exams-date    | 2017-Latest-300-320-Exam    | Pass-300-101-Exam    | 100%-Pass-300-115-Exams    |
http://www.portvapes.co.uk/    | http://www.portvapes.co.uk/    |