AMD: Pirate Islands (R* 3** series) Speculation/Rumor Thread

The question is really how much more efficient can be gained beyond Maxwell (adjusted for process, of course.) For gaming workloads, I doubt there's going to be a huge additional jump, though I've been proven wrong about these kinds of claims in the past. So, for now, I'm assuming just process scaling in terms of perf/mm2. Say a 50 to 60% increase per mm2. That gives a 400mm2 chip roughly the same performance as a 600mm2. We already know that a 600mm2 doesn't strictly need HBM to be competitive...
Isn't a 50-60% increase enough to make HBM a winner for high end?
 
Isn't a 50-60% increase enough to make HBM a winner for high end?
If the next gen 400mm2 chip is 50-60% faster than current gen 400mm2, then it will only be marginally faster than current gen 600mm2. For which HBM is not a necessity...
 
My fear is that the next few generations of PC graphics hardware will be even more depressing. It's shaping up to be a massive leap in performance with HBM and a new process after a long time on 28nm.

Problem is there's nothing to do with all that performance. We're still going to be playing mediocre looking games and bumping up resolution and AA to justify needing all that horsepower.
 
If the next gen 400mm2 chip is 50-60% faster than current gen 400mm2, then it will only be marginally faster than current gen 600mm2. For which HBM is not a necessity...
Yes, but I'm talking about the high end, i.e. the next gen 600mm2 chip.
 
My fear is that the next few generations of PC graphics hardware will be even more depressing. It's shaping up to be a massive leap in performance with HBM and a new process after a long time on 28nm.

Problem is there's nothing to do with all that performance. We're still going to be playing mediocre looking games and bumping up resolution and AA to justify needing all that horsepower.

VR will eat all the horsepower we can throw at it, and then some. I expect high-end GPU sales to tick up in a major way once/if VR headsets become mainstream. Other than that, I see little point in top-end performance nowadays...
 
HBM gen 1 won't be any cheaper than HBM gen 2. Actually, for lower end devices that need more memory than 1GB yet don't necessarily need the BW of many stacks HBM2 could be cheaper.

One thing I noted was a comment made about HBM dies being very thin at 100 microns.
That's notable because a lot of discussion and research prior to Fury's launch about die thinning for stacked DRAM involved silicon thicknesses in the range of 50 microns.
This might be one of the reasons why HBM is half-height relative to HBM2. HBM2 could be an architectural revision and a shift to a new and more polished stack manufacturing process.

If HBM1 is reliant on a process that Hynix wants to move on from, and HBM2's other features are what the broader set of customers want, HBM1's lifespan could match that of the prototype process it comes from.

The additional layers possible with HBM2 might make an 8-high stack more expensive, but it should be possible to just use 4 layers if it's really necessary. Then, it's a case of using a bug-fixed HBM2 process and broader volumes versus single-client HBM1 on something clunkier.
 
VR will eat all the horsepower we can throw at it, and then some. I expect high-end GPU sales to tick up in a major way once/if VR headsets become mainstream. Other than that, I see little point in top-end performance nowadays...

I'll believe it when I see it. They said the same thing about 3D.

I will be very surprised if VR in the form of head mounted hardware ever goes mainstream.
 
I'll believe it when I see it. They said the same thing about 3D.

I will be very surprised if VR in the form of head mounted hardware ever goes mainstream.

Have you tried it?

I honestly believe that VR will (eventually, when it can be provided by lower-cost machines) be bigger than all of gaming. The kind of experiences it can provide are not just a cheap gimmick like 3d displays and are totally outside what can be provided by current platforms -- simply walking about in interesting landscapes without any actual interaction is something that people will pay for.
 
I'll believe it when I see it. They said the same thing about 3D.
In case anyone is willing to listen to anecdotal evidence...
I've always bought the cheapest card I could, as long as it did not fall off the perf/price curve (cards above ~$100 usually have more perf/$ than cards below). Then I stopped gaming. Oculus rift is actually making me consider not only a return to gaming, but also buying a "R9 480"/"GTX 1060" or similar (i.e. the minimum VR card from the next node). I still fondly remember how Crysis gave me an inkling of immersion, and I want to relive that.
I've two gamer friends, one on GTX 570, the other on GTX 680, both are specifically putting off any GPU update until they go for VR. My brother had a 6850 and simply could not put off an update any longer.
 
Have you tried it?

Nope, haven't had the chance to.

I honestly believe that VR will (eventually, when it can be provided by lower-cost machines) be bigger than all of gaming. The kind of experiences it can provide are not just a cheap gimmick like 3d displays and are totally outside what can be provided by current platforms -- simply walking about in interesting landscapes without any actual interaction is something that people will pay for.

The idea of VR is amazing. I'm not sold on the implementation. For one I don't think a lot of people would enjoy wearing headsets for an extended period of time. Not to mention the social awkwardness of the whole thing.

Anyway, sorry for the OT.
 
Nope, haven't had the chance to.



The idea of VR is amazing. I'm not sold on the implementation. For one I don't think a lot of people would enjoy wearing headsets for an extended period of time. Not to mention the social awkwardness of the whole thing.

Anyway, sorry for the OT.

I've found this is a common opinion of people that haven't tried it yet. Skepticism is good and all, but there's a reason Facebook spent billions on Oculus. It's the first new way of presenting media that's gotten this dramatic of a reaction in a century. A lot of people literally have the same absolutely amazed/believing its real reaction that people that saw the first movie had, when people fled the first makeshift theaters because they thought they were going to be run over by a train.

Sure it's got a long way to go from a headset sitting on your face to The Matrix, but movies caught on damned quickly, and that was over a century ago, before mass communication and easy international trade made things fantastically easy to spread like wildfire.
 
For one I don't think a lot of people would enjoy wearing headsets for an extended period of time. Not to mention the social awkwardness of the whole thing.

The reason neither of these is as much of a problem as people who haven't tried it think is that when you set up a swivel chair in the corner of the room and put on a head set, it doesn't actually feel to you as if you were alone, sitting in a chair with a headset on. It actually fundamentally feels like being transported somewhere else. As far as your brains know, you won't be antisocially sitting in a corner with a bag over your head, you'll be interacting with other people in the forests of Pandora or whatever.
 
Sure it's got a long way to go from a headset sitting on your face to The Matrix, but movies caught on damned quickly, and that was over a century ago, before mass communication and easy international trade made things fantastically easy to spread like wildfire.

Watching a movie with friends/family is still a very interactive and social activity. It's no different to watching a play or other performance which humans have been doing for thousands of years.

Not to mention you only need one screen/stage cause everyone has eyeballs.

The reason neither of these is as much of a problem as people who haven't tried it think is that when you set up a swivel chair in the corner of the room and put on a head set, it doesn't actually feel to you as if you were alone, sitting in a chair with a headset on. It actually fundamentally feels like being transported somewhere else. As far as your brains know, you won't be antisocially sitting in a corner with a bag over your head, you'll be interacting with other people in the forests of Pandora or whatever.

I don't doubt that the people wearing the headset will have a good time. To everyone else in the room it may be quite strange though. Today if you're playing a game other people can still watch/listen and share the experience.

There's no activity today that removes you from the real world the way VR could. The big visor on your head just adds to the awkwardness. To me that's extra-ordinarily anti-social :)

But maybe I'm just a luddite....
 
The reason neither of these is as much of a problem as people who haven't tried it think is that when you set up a swivel chair in the corner of the room and put on a head set, it doesn't actually feel to you as if you were alone, sitting in a chair with a headset on. It actually fundamentally feels like being transported somewhere else. As far as your brains know, you won't be antisocially sitting in a corner with a bag over your head, you'll be interacting with other people in the forests of Pandora or whatever.

Yarr, saying VR won't work because you don't interact with those immediately around you is like saying smartphones will never catch on because you won't interact with those around you.
 
Yarr, saying VR won't work because you don't interact with those immediately around you is like saying smartphones will never catch on because you won't interact with those around you.
If you have actually tried VR glasses and still think the equivalency above is valid, there's something seriously wrong with the way you use a smartphone.

The difference wrt the ability to interact with those around you is huge. With VR glasses, the only practical way to do that would be to render those around you into the rendered virtual world. With a phone, it can be a simple as rotating your eyeballs.
 
Back
Top