AMD: Speculation, Rumors, and Discussion (Archive)

IIRC Radeons are all thought to be Cxxx's, and Dxxx would be something completely different (especially since Dxxx's only appeared quite recently)

 

Maybe SoC or Apple related?
Cheers

 

Then again, after C9xx the next obvious would be D0xx.
Anyway, just double checked, D0xx is indeed RX 480, it's visible in official renders

 

So there have been C98x(Baffin), C99x, C95x, C94x, C92x, C91x and finally the D00x from shipping manifests and it's D00x that's the Polaris 10 which was developed earlier than the smaller Polaris 11(C98x) but shows up way late to the party. So there's some truth to the delay rumor?

And even the C993 passed the certification earlier, is it a different chip or just a souped up Polaris 10?

 

A designation to avoid somewhat. Or really better not have a driver security issue anyway

 

D0 is for Polaris 10 based boards. See here for example: http://www.blogcdn.com/slideshows/images/slides/391/001/8/S3910018/slug/l/rx480-hands-on-gallery-5-1-1.jpg (D00947)

 

Must be the full or dual Polaris 10. Price is more than twice.

 

just my perception but it seems that many people liked the 480. youtube is full of content creators saying how good it is(talking about performance per dollar) so it seems that AMD could be successful with it. 21 days and counting.

 

Fudo's sure about Apple going for Polaris instead of Pascal. This, if correct, is a major win for AMD.
http://www.fudzilla.com/news/graphics/40835-all-apple-2016-laptops-come-with-amd-gpus

 

Has OpenCL support gotten better on Nvidia? If not it's unlikely Apple will ever seriously look at Nvidia for supplying its graphics hardware.

And if Apple ever plans to support adaptive sync on their displays in the future (would be nice but not holding my breath) that would also be a barrier as they wouldn't want to be tied down by Gsync.

Regards,
SB

 

They could presumably convince nvidia to support adaptive sync on macs. As long as they call it iSync or somesuch it wouldn't look too silly for nvidia that they'd only support it on macs .

 

Actual chip thinner than normal? Just a packaging thing or like thinner wafers or something?

 

Anyone else super intrigued about the lack of actual Polaris news / leaks since the announcement a week ago?

All we have is a Polaris 10 RX 480 reference card that will have 4-8GB of GDDR5, will cost between $199 and $249. single 6-pin power connector, up to 5.5TFLOPs and that's it.

I thought Computex would revolve around Polaris 11 design wins on laptops, but so far Polaris 11 is a complete no-show.


Apart from that DX12 AoTS mGPU benchmark against a GTX 1080, we've got practically zero leaks on benchmark numbers for final hardware.

 

I hope they have a surprise in stock, like a RX 480X or 490 with Fury performances, around 300$...

 

The pricing doesn't seem to be linear there, the dual Fury card was 120k to 80k for Fury X and it doesn't make sense for the dual card to come up before the single one.

There have been rumors of AMD delays and P10 not achieving the high clocks, so maybe the old prices were before nvidia were in the equation and now AMD are going $200 route. We'd know soon enough.

 

Much as people like to ridicule AMD narketing, I have to say that they managed to reveal just enough information to get everyone with an active interest intrigued, and create a lot of anticipation and demand for in depth reviews and information. The tech websites will get their page hits!
Also, forum speculation has run rampant, firm positions have been taken, so once harder data is available, there will be another round of forum activity and product buzz.
Plus, a healthy bit of hesitation has been instilled among prospective GTX 1070 buyers. Pretty well played, as far as I can see.

 

To further that we still haven't seen what the board or chip look like to the best of my knowledge. No architectural details, processor counts, etc.

 

There's some small amelioration, but honestly i will not consider an Nvidia gpu for OpenCL.. ( Pascal still seem to be OpenCL 1.2 ). But well, can only that it change.

Well, thoses prices are for insurance, and are not forcibly fixed on the true value of a proudct( and in this case it seems just to be the PCB ).

 

I've seen leaks indicating a clock speed of 1266MHz, which would give a raw power of 5.83TFLOPS, but nothing double-confirmed yet.

By the way, what's the basis for the 110~130W estimate in this thread's title?

 

Guess work, considering Max power available is 150 Watts, no card would ever draw the absolute max available and would have safety margins available, and public statements by AMD saying 150Watts Max and not saying under 100 Watts and not saying just over 100 Watts.