• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD VEGA confirmed for 2017 H1

Status
Not open for further replies.
opethdisciple said:
See the one thing AMD kept banging on about yesterday was trying to come to market with solutions to problems rather than just meeting their competition square on.

For this reason maybe 4k is what they are aiming for with Vega as they seem to think that currently there isn't really a 4k card in the market. (I'm sure the 1080ti lot have something to say about that)
Click to expand...
Plenty of cards can do 4k, all depends on the settings used and the frame rate you expect.
 
I'm still on scabby 1080P @ 60Hz (play on a 40" TV), I still want Vega now! :D Feel like I've owned this 290 forever!
 
shankly1985 said:
All about high bandwidth cache controller... This my most exciting feature I can't wait to test and see more results.
Click to expand...
From what was said about it, as long as your game/task fits in the memory, HBCC will not give any advantage. It becomes magical only when constant RAM (or NVMe in case of SSG) access is necessary
 
nashathedog said:
Same here.



All we can do is wait and see, It's disappointing if true though.

I get the feeling this might actually be a last minute change of plan with AMD needing to get these work cards out as soon as possible because of the V100 announcement.
Click to expand...


AMD can't suddenly change plans last minute like that, plus they knew Volta was coming, NVidia had talked about before and it was public knowledge that Nvdia would be delivering big Volta for various supercomputer contract e.g. at Oakridge by the end of the year. Nvidia also publicly announced back at CES that a smaller Volta would be in Xavier, the drive-PX unit for autonomous cars shipped Q3.



I thin what has actually happened is supply of HBm2 is incredibly low so AMD are concentrating on a deep-learning part that can be sold at far higher profit margins. Vega yields might also be low. There were some pretty solid rumours of Vega only having 16000 units to ship, well that is pretty dire considering one of the SKUs must be fighting 1070 and another 1080. but if they can sell those 16,000 at $3-5000 a go then why wouldn't they?



AMDs announcement looks pretty bad in light of the Volta announcement though. AMD are carefully ignoring the fact that Vega only does 1:16 FP64 compared to the 1:2 of Pascal GP100 and Volta GV100. They aren't going to get any traction in the HPC market with that. For deep-learning it stacks up OK against Pascal, but then Volta could be around 8x faster due to the Tensor Cores. And clearly AMD don't want to draw any comparisons to Volta.
 
alec said:
From what was said about it, as long as your game/task fits in the memory, HBCC will not give any advantage. It becomes magical only when constant RAM (or NVMe in case of SSG) access is necessary
Click to expand...
It is something that could really open new rendering techniques and push technologies like ID's megatextutre to whole new levels, but until it features on every GPU form mainstream up on both IHVs then that wont happen.

Until then, yes it only works if you run out of VRAM.
 
old gamer said:
depends really, if hbcc cant do its stuff then I think I will be happy with a 8gb card. but then I don't plan on getting a 4k montor
Click to expand...

Any extra layer in the process by nature of being there will slow the process down, whether that is noticeable or not is another matter.
 
Status
Not open for further replies.
Back
Top Bottom