• 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.
I still have my tinfoil hat on that the frontier edition is using 8GB HBM2 and 8GB GDDR5/X to get their claimed 16GB HBC.
That seems feasible as the stated plan anyway is to layer memory types using HBCC. All the way down to SSD I think or that fancy onboard storage mentioned

A couple people were building ryzen rigs with 8gb main memory, I knew that was wrong at the time and with this card it might be really bad possibly

I’d suggest waiting just a little while longer
So thats June/July availability. A little while is not August September :|
 
Aren't they both aspects of 21:9 though?

2.35:1 becomes 21.15:9

2.37:1 becomes 21.57:9

Or is that wrong and it's not how it's summed up?

I'm only guessing here.
Not sure myself but I have watched a few documentaries with the director stating that they chose "X" resolution to portray the landscapes etc. It has nothing to do with getting more people in the cinema though lol.
 
That seems feasible as the stated plan anyway is to layer memory types using HBCC. All the way down to SSD I think or that fancy onboard storage mentioned

Raja actually came out during the AMA and stated they're using full 16GB HBM2, so 2x8GB HBM2. That's rather mindblowing, considering neither Samsung, nor Hynix have showed 8GB modules working, and couldn't even manage fullspeed 2.0GBps 4GB modules.

Even NVIDIA's Volta Tesla V100 is only using 4x 4GB modules; same for Tesla P100.

Vega Frontier Edition is using 2x8GB @1.88Gbps to get their 480GB/s Bandwidth.

Sadly Hynix failed to deliver their 2.0Gbps as planned for Q3 2016; which would have given vega 512GB/s bandwidth.

The whole Radeon Pro SSG, with HBM2 + 1TB NVMe SSD is still fascinating; and the HBCC controlling it all is rather awesome.

Raja also talks about "primitive batch binning [...] increases memory bandwidth efficiency over 2x" (up to 8x)

https://www.youtube.com/watch?v=590h3XIUfHg&feature=youtu.be&t=2865
 
Raja actually came out during the AMA and stated they're using full 16GB HBM2, so 2x8GB HBM2. That's rather mindblowing, considering neither Samsung, nor Hynix have showed 8GB modules working, and couldn't even manage fullspeed 2.0GBps 4GB modules.

It's quite possible that Samsung/Hynix aren't publicly listing their 8GB modules because it's all been allocated to AMD for the next couple of quarters.
 
So thats June/July availability. A little while is not August September :|
"A little while" could be anything. Even if AMD knew it was still 4+ months away, they might say "a little while" in the hope that people wouldn't decide to give up and buy an nV card.

It's out when it's out, and unless AMD give a concrete release date, that could be any time at all.
 
But 7nm and Navi are coming in 2H 2018....so why would they bother?

"Vega 2.0" card is meant to just be a refresh of Vega frontier, with 1:2 FP64, higher clocks, and 32GB 4-stack HBM2 at 1TB/s. So it's just a pro compute card.
do you really think glofo would have 7nm ready for next year ?
 
do you really think glofo would have 7nm ready for next year ?

Yes?

Their roadmap says it's on track, and IBM originally designed the process, GloFo just bought the IP and are reusing a lot of their 14nm equipment for it.

And TSMC also say they're starting mass production of their 7nm by the end of 1H 2018.
 
Keep the chat to Vega please guys.
I'll second that....amazing how the Vega thread becomes a wasteland for any old topic that decides to drop by when it feels like it. I know we have all been waiting around a while to see what Vega is all about and news is a little thin at the moment....also the debates about Ultrawide/Aspect ratio's and the like are all fascinating (totally subjective...just like lots of other stuff BTW) and all, but I am sure those would be better in a thread all to themselves where likeminded folks could debate and argue all day long.....yes, just like we do in this thread, but about different shapes of computer tellies. So please keep the topic about Vega please (This thread will now go dead until the end of May) LOL :p
 
Yes?

Their roadmap says it's on track, and IBM originally designed the process, GloFo just bought the IP and are reusing a lot of their 14nm equipment for it.

And TSMC also say they're starting mass production of their 7nm by the end of 1H 2018.
7nm next year will be risk production and small SoCs for phones. Back won't launch until 2019 most likely
 
Status
Not open for further replies.
Back
Top Bottom