AMD RyZen CPU Architecture for 2017

[Picao84]

I wonder why ASUS (my favourite laptop brand) does not seem to be one of the early adopters for their Zenbook line, since they were with Ryzen "non-mobile" with the ROG laptop.

 

[Kaotik]

Ryzen 7 2700U and 5 2500U

2700U: 4c/8t - 2.2/3.8GHz with 10 NCU
2500U: 4c/8t - 2.0/3.6GHz with 8 NCU

TDP: 15W

Actually it's just 15W by default and configurable between 9 to 25 watts

 

[Deleted member 13524]

Slide footnotes:

https://imgur.com/a/BC4Fz

And a Raven Ridge high-quality x-ray, it's 209.78mm^2:

 

[Deleted member 13524]

From TechReport, here's a slide comparing 2700U to a Kaby Lake 7500U + Geforce 950M

 

[Picao84]

Its looking good but I need to see actually independent measurable benchmarks on energy performance in real devices. November 2nd cannot come soon enough.

 

[Deleted member 13524]

In the meanwhile, it looks like the Smach-Z will soon post an update about release dates for their handheld/console, as well as performance numbers for it:

https://www.kickstarter.com/projects/smachteam/smach-z-the-handheld-gaming-pc/posts/2014071

 

[Picao84]

So its November 2nd. Has anyone seen any reviews?

 

[Picao84]

No launch reviews, this does not bode well :/

 

[el etro]

I have a feeling that the Zen2 CCX will have 8 cores. I'm speaking this after the Carnard PC leak for Zen2 Epyc.

 

[Kaotik]

I have a feeling that the Zen2 CCX will have 8 cores. I'm speaking this after the Carnard PC leak for Zen2 Epyc.

Unlikely, either 4 or 6 core CCX's and possibly more than 2 CCXs per chip, regardless if the supposed leak has any credibility

 

[Alexko]

Unlikely, either 4 or 6 core CCX's and possibly more than 2 CCXs per chip, regardless if the supposed leak has any credibility

Sam rarely leaks anything he's not really confident in. And he was the first to publish Ryzen benchmarks—that I know of.

 

[Kaotik]

Sam rarely leaks anything he's not really confident in. And he was the first to publish Ryzen benchmarks—that I know of.

The "5GHz Zen" wasn't worth anyones confidence, even though he claims it's true it's so far from reality at least I can't just take his word for it

 

[el etro]

Transistor density has gone up FOUR times. sRAM cell is the smaller on the 7nm market and power reduction is awesome.

Two 8-core Zen CCX2 with double the cache(thanks to higher sRAM density) gives us 64MB of L3 cache in a 16-core Ryzen. Four of those Ryzen dies gives us the 256MB of L3 cache is leaked for EPYC by CPChardware.

 

[Alexko]

Do we even have actual density figures for GloFo's 7nm process? Or even TSMC's, for that matter?

 

[Bondrewd]

Do we even have actual density figures for GloFo's 7nm process? Or even TSMC's, for that matter?

Ye.

 

[mfaisalkemal]

Do we even have actual density figures for GloFo's 7nm process? Or even TSMC's, for that matter?

source:link

 

[Arnold Beckenbauer]

Some results of HP Envy x360 with R5 2500U: http://apusilicon.com/hp-envy-x360-w-ryzen-first-impressions/

https://imgur.com/LOCjX2S

~600 Points in MT-Test are extremely good.

 

[Alexko]

Some results of HP Envy x360 with R5 2500U: http://apusilicon.com/hp-envy-x360-w-ryzen-first-impressions/

https://imgur.com/LOCjX2S

~600 Points in MT-Test are extremely good.

Cinebench is notoriously Zen-friendly, but that's very good indeed, close to a desktop Ivy Bridge—possibly better than some Ivy i5s. Beating Iris Pro in graphics is quite nice too. I honestly didn't expect that much of a leap in graphics, and I wonder why AMD didn't announce more than marginal improvements on that front. Perhaps we ought to wait for more reviews to verify those results, but they do seem very promising.

 

[Arnold Beckenbauer]

You know: Who cares about CineBench?
But on the other hand: It's a huge improvement compared to my A10-9600P.

 

[monstercameron]

FYI, apusilicon is my shitty blog.
I added a gameplay video with instantaneous power.
It will drink around 40W +/-3W when going all out but with framerate capping and lower resolution I was able to get around 20W +/- 3W.