AMD: Navi Speculation, Rumours and Discussion [2019-2020]

If you believe rdna1 is better than turing, I guess anything is possible.

 

yes. my bank account after the upgrade i guess...

 

150% less of a measure, no, but technically you can compare the difference of a percentage change...

"Growth was 350% in the first month and 200% in the second - growth was 150% less in the second month."

But, those would be dumb, confusing numbers.

In this context, 150% is likely a mistake.

 

Albeit that kind of sentence does happen in common parlance, it's also considered less than ideal. When comparing percentages like that, one should use the term "percentage points"

 

Or basis points ...

 

That would be 100% less money and x% more debt

 

Yes, I was in a hurry and mistakenly put 150%, instead of 50%... (so mock me)

Vega20 is GCN, and rdna1 is 50% greater performance per watt. While rdna2 is another 50% improvement. Though I suspect many of you know what I was hinting at, but choose to chide me, instead of engage in a discussion.



There is no mistaken the gains AMD has made. Only thing left, is to see rdna2..

 

* When compared to Vega 10, not 20. And Vega itself is hardly a great example of GCN in power efficiency. So maybe you should look at the available data instead of blindly listening to what Lisa Su says from stage to investors.

 

Again, you point is extremely weak, because you can see there are clear gains... that you are 100% dismissing.

ed: When have you know dr su to lie..?

 

The other measure is, RDNA2 provides 225% performance improvement over GCN.

 

Correct and that is a good place to start, for speculation and non-rumor. We already know where the 5700xt stands...



So, getting back to my earlier statement:

Navi10 is pretty small die (251mm^2) and if you make a chip the size of vega20 (331mm^2), using rdna2 architecture you have essentially a 2080ti's performance, using about 50% less power than Radeon Vii.

 

FTFY.

 

Hey, there's no rule that says you can't put a diesel generator on a graphics card.

 

It's relevant on what performance would be expected within a console's (usually) more constrained TDP.
Is it following Navi 10 performance/power curves, or Navi 10 * 1.5?

 

Both the official Xsx design and the apparent PS5 devkit design seem centered around a lot of airflow, so I'm guessing both realized they can get much more out of the chips they buy if they have a good cooling system. My guess is both will have a decently higher TDP than the PS4/XBO this time around.

 

From the leaked pictures, Xsx uses a 2pin AC connector that can't go over 200W...
pics here:
https://www.tweaktown.com/news/70109/xbox-series-prototype-shown-again-latest-renders/index.html

Edit: After checking these plugs are 10A rated so no real power limit on the APU, except the size/noise/cost of the cooling solution

 

You sure about that? I have a figure 8 power cable with a tag on it right in front of me. It says 10amp.

 

after checking, you are right, these plugs are 10A rated. I don't know why I was thinking 3A... please forget my comment

 

The architecture forum is confused about ... percentage calculations?
.
.
.