New Zen microarchitecture details

[TheELF]

Except probably >90% of the world doesn't give 2 craps about 3D graphics performance but needs a power efficient processor (i.e. desktops in professional environments).

And that helps the APU to sell how exactly?

 

[ElFenix]

90% of the world's computers aren't desktops in professional environments.

 

[iBoMbY]

Except probably >90% of the world doesn't give 2 craps about 3D graphics performance but needs a power efficient processor (i.e. desktops in professional environments).

Probably 90% of the world are idiots, and don't give 2 craps about anything, including power efficiency. They buy whatever their "experts" (who are mostly part of the 90%) tell them to buy.

 

[StrangerGuy]

Probably 90% of the world are idiots, and don't give 2 craps about anything, including power efficiency. They buy whatever their "experts" (who are mostly part of the 90%) tell them to buy.

Yup, you are truly a unique snowflake over there.

 

[iBoMbY]

Yup, you are truly a unique snowflake over there.

I'm the snowflake, because you can't accept the truth? Okay ...

 

[krumme]

For mid and high end laptops and servers why not use separate cpu gpu hbm2 in various sizes and bins on an organic interposer?
What is the packaging cost?
The entire idea of one size fits all is stupid and the segmentation we see now on mobile with next to no difference is artificial.

 

[scannall]

After reading Mark Papermasters interview that covered APU's along with other things, the chances of a HBM2 APU in 2017 is zero. Unlikely for 2018, but possible. It's something they want to do, but the cost isn't where they need it to be yet.

I do have a HBM question though. Could HBM be used in an APU where only the GPU portion used it, then the CPU portion would go out to DDR4 for example?

 

[KTE]

There is a test chip of TWO years ago that runs at that frequency. And it's a synthetized ASIC design, with an high FO4 (>30). Zen is a low FO4 (<30) custom design. And it has AVFS and DFVS that let you avoid the 10% voltage margin. I think that at 0.8V Zen will go at least at 2.4GHz...

Precisely. A random TEST vehicle part.

Means nothing for mass manufacturing.

And what's its total chip power AND IPC vs Skylake?

Sent from HTC 10
(Opinions are own)

 

[bjt2]

Precisely. A random TEST vehicle part.

Means nothing for mass manufacturing.

And what's its total chip power AND IPC vs Skylake?

Sent from HTC 10
(Opinions are own)

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...

 

[Atari2600]

And that helps the APU to sell how exactly?

Its cheaper than a separate CPU+GPU solution.

 

[ButtMagician]

https://twitter.com/CPCHardware/status/818429683633127424
https://twitter.com/CPCHardware/status/818428797427052545

Canard PC mentions 6c/12t version of Ryzen consuming 65W of power (8c/16t a bit over 95W). Pretty hyped.

 

[TheELF]

Its cheaper than a separate CPU+GPU solution.

"probably >90% of the world doesn't give 2 craps about 3D graphics performance"
but zen+vega+hbm will be cheaper then intel+crappy igpu?

 

[zinfamous]

https://twitter.com/CPCHardware/status/818429683633127424
https://twitter.com/CPCHardware/status/818428797427052545

Canard PC mentions 6c/12t version of Ryzen consuming 65W of power (8c/16t a bit over 95W). Pretty hyped.

what clocks I wonder? sounds awesome.

 

[ecogen]

what clocks I wonder? sounds awesome.

Their 8c/16t is the 3.6/4.0 sample, no idea if that's final clocks. Would be surprised if they weren't though.

 

[krumme]

We need some french to translate "dépassent bien" but i think the translated meaning is "well within" the 95tdp for the 8c 3.6 4.0 part and not slightly surpass?. Bodes very well imo.

 

[krumme]

Their 8c/16t is the 3.6/4.0 sample, no idea if that's final clocks. Would be surprised if they weren't though.

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

 

[leoneazzurro]

It means: it quite surpasses, so it's above. but it must be seen how the tests were done

 

[Abwx]

We need some french to translate "dépassent bien" but i think the translated meaning is "well within" the 95tdp for the 8c 3.6 4.0 part and not slightly surpass?. Bodes very well imo.

It means "actually"...

But he s not right, because he took is sample as basis, this latter did display 93W at the CPU 12V rail while AMD s demo showed the same power but at the main, wich mean that his sample not only had a 3% (at least) lower frequency but also 7% higher voltage (= +15% TDP at same frequency) than the sample used by AMD.

It means: it quite surpasses, so it's above. but it must be seen how the tests were done

See above, it means "actually", or "effectively", the "bien" does not mean "well", "good" or "quite" in this case but is used as confirmation about something that was not clear previously.

 

[ecogen]

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

Why would the 6c be clocked lower?

 

[leoneazzurro]

It means "actually"...


See above, it means "actually", or "effectively", the "bien" does not mean "well" or "good" in this case but is used as confirmation about something that was not clear previously.

Nope, it means "surpasses by a margin". I studied French. BTW, they are referring to theid 3.6/4GHz sample

 

[Doom2pro]

Here is Reddit's take on the french translation...

https://www.reddit.com/r/Amd/comments/5n65aj/ryzen_8c16t_f4_3640_engineering_sample_over_95w/

Personally I could care less if it's slightly over 95W TDP, I'm buying the high end X370 board anyway so it shouldn't even care.

 

[krumme]

Why would the 6c be clocked lower?

As the cores is 3/4 of the 8c part but the tdp is a bit lower more like 2/3?
Anyway hope oc tools is as easy as the new software they use for their gpu. That would be nice. Its also nice all cpu is unlocked. No more paying tax for a k designation on a cpu. Never more. I hate such segmenting practice.

 

[ecogen]

As the cores is 3/4 of the 8c part but the tdp is a bit lower more like 2/3?
Anyway hope oc tools is as easy as the new software they use for their gpu. That would be nice. Its also nice all cpu is unlocked. No more paying tax for a k designation on a cpu. Never more. I hate such segmenting practice.

Ah, didn't really think too much about it

But yeah at the end of the day we're gonna be OCing these so screw tpd and base clocks

 

[Abwx]

Nope, it means "surpasses by a margin". I studied French. BTW, they are referring to theid 3.6/4GHz sample

And i m french..

Translated accuratly it s " they exceed effectively", the term "bien" is used to increase the positiveness and sureness of the affirmation, said otherwise he could have stated that it s 200% sure that it exceed 95W, but this is in no case a quantitative statement that would say that it exceed it significantly, contrary to what can be understood at first glance by a litteral translation.

Besides he has no 3.6 sample, he said that the excess is not surprising given what he measured with the sample he used for his review...

 

[zinfamous]

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

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?