OC3D Forums

OC3D Forums (https://forum.overclock3d.net/index.php)
-   OC3D News (https://forum.overclock3d.net/forumdisplay.php?f=47)
-   -   AMD's RX Vega Nano has been pictured (https://forum.overclock3d.net/showthread.php?t=81406)

WYP 31-07-17 07:56 AM

AMD's RX Vega Nano has been pictured
 
Will AMD release a small form factor RX Vega?

https://www.overclock3d.net/gfx/arti...034010616l.jpg

Read more on AMD's RX Vega Nano.

AlienALX 31-07-17 08:50 AM

Not really much point unless Nano was a hot seller. Lots has changed since then, you can get 1070 minis now.

Korreborg 31-07-17 09:32 AM

The 1070 has 60w lower tdp than the 56.
So a card like that needs to be dirt cheap to make sense.

tgrech 31-07-17 09:45 AM

It's important to remember than the TDP of AMDs new cards includes the TDP of the memory, while non-HBM cards don't. It's already fairly pointless to compare different manufacturers TDPs anyway and they're not an accurate indicator of real world power consumption(And nor are they meant to be).

The previous Nano card was at times efficient than Maxwell and Polaris despite being on 28nm thanks to power envelope tweaking and good binning. Given Vega will be in APUs soon in fairly reasonable core counts I'd expect this can be an incredibly effecient architecture when it needs to be. I'm sure this card will be a demo of what Vega can really do in perf./watt with high core counts and low clocks. Expect it to be just as revolutionary compared to both Nvidia and AMD cards as the last Nano was in terms of system power(And remember, HBM2 uses significantly less power than GDDR5X).

AlienALX 31-07-17 09:46 AM

Quote:

Originally Posted by tgrech (Post 954549)
It's important to remember than the TDP of AMDs new cards includes the TDP of the memory, while non-HBM cards don't. It's already fairly pointless to compare different manufacturers TDPs anyway.

The previous Nano card was at times efficient than Maxwell and Polaris despite being on 28nm thanks to power envelope tweaking and good binning. Given Vega will be in APUs soon in fairly reasonable core counts I'd expect this can be an incredibly effecient architecture when it needs to be. I'm sure this card will be a demo of what Vega can really do in perf./watt with high core counts and low clocks. Expect it to be just as revolutionary compared to both Nvidia and AMD cards as the last Nano was in terms of system power(And remember, HBM2 uses significantly less power than GDDR5X).

What the RAM pulls is irrelevant because they were measuring at the wall. And when review time comes that is how they will be measuring the overall power draw.

tgrech 31-07-17 09:55 AM

Quote:

Originally Posted by AlienALX (Post 954550)
What the RAM pulls is irrelevant because they were measuring at the wall. And when review time comes that is how they will be measuring the overall power draw.

Who were measuring at the wall? No one has tested this model yet have they?. Just pointing out that having more of the graphics card integrated into the GPU package itself will raise the GPU package TDP without raising overall power consumption(Which is why you can't compare Vega's package TDP with other models when trying to compare total power consumption- a mostly unrelated metric besides some vague correlation between he two).

Plus, using TDPs for the FE model to try and estimate or compare the Nanos power consumption is a horrifically flawed and inaccurate way of going about it. With HBM2, TDP for 16GB cards will be different from 8GB cards, and memory clocks now also affect package TDP significantly more as well.

AngryGoldfish 31-07-17 12:53 PM

Quote:

Originally Posted by tgrech (Post 954553)
Who were measuring at the wall? No one has tested this model yet have they?. Just pointing out that having more of the graphics card integrated into the GPU package itself will raise the GPU package TDP without raising overall power consumption(Which is why you can't compare Vega's package TDP with other models when trying to compare total power consumption- a mostly unrelated metric besides some vague correlation between he two).

Plus, using TDPs for the FE model to try and estimate or compare the Nanos power consumption is a horrifically flawed and inaccurate way of going about it. With HBM2, TDP for 16GB cards will be different from 8GB cards, and memory clocks now also affect package TDP significantly more as well.

Good points.

AlienALX 31-07-17 02:25 PM

We'll see I guess.

NeverBackDown 31-07-17 05:17 PM

I bet it will be between a heavy OC'd 1060 and a stock 1070. I hope the power consumption drops by at least 100watts if that is the case.

AngryGoldfish 31-07-17 05:32 PM

Quote:

Originally Posted by NeverBackDown (Post 954658)
I bet it will be between a heavy OC'd 1060 and a stock 1070. I hope the power consumption drops by at least 100watts if that is the case.

But that's already pretty close a Fury Nano.


All times are GMT. The time now is 04:04 AM.

Powered by vBulletin® Version 3.8.7
Copyright ©2000 - 2020, vBulletin Solutions, Inc.