** The Official Nvidia GeForce 'Pascal' Thread - for general gossip and discussions **

[easyrider]

Nvidia are going to sell nowhere near as much this time round. AMD have the superior hardware this time round. End of story. Also, for those waiting to see what's best for VR _ again AMD.

Dude you have 223 posts in 3 years ,so I'm guessing you contemplate each one...and you chose to post this ?

 

[Gregster]

Dude you have 223 posts in 3 years ,so I'm guessing you contemplate each one...and you chose to post this ?

That made me laugh

 

[easyrider]

That made me laugh

 

[r7slayer]

Dude you have 223 posts in 3 years ,so I'm guessing you contemplate each one...and you chose to post this ?

Lol.
Made me laugh! And he's so sure he is stating facts apparently.

 

[CAT-THE-FIFTH]

Technically he is somewhat right though - AMD was utterly destroyed in laptops by Maxwell and having something like Polaris 11 would mean they would claw back sales from Nvidia in that space but by how much is another question.

If you looked at the last 12+ years of units shipped per quarter AMD used to be 35% to 40% of all units shipped not the 18% to 22% they are now in the last year or so.

 

[andybird123]

N. Also, for those waiting to see what's best for VR _ again AMD.

Based on what?
If youre refering to that guy that commented on preemption on nvidia, checkout the oculus blog post from yesterday - preemption is fixed on nvidia.

So far, there are more titles having announced VR-SLI support than liquidVR - in fact i cant find any titles hving announced liquidvr support

 

[CAT-THE-FIFTH]

Based on what?
If youre refering to that guy that commented on preemption on nvidia, checkout the oculus blog post from yesterday - preemption is fixed on nvidia.

So far, there are more titles having announced VR-SLI support than liquidVR - in fact i cant find any titles hving announced liquidvr support

The AMD cards support fined grained pre-emption which is hardware based whereas the Nvidia Maxwell cards support draw called base pre-emption which is far more CPU heavy as it is software based. AFAIK,fine grained pre-emption will only arrive with Pascal(could be wrong about this).

If you look at non-gaming VR usage,ie,commercial content creation and for teaching,AMD has had far more traction than Nvidia it appears which is unusual considering Nvidia is the bigger company and sells more cards too in the commercial sector AFAIK.

 

[andybird123]

That was the case but as i said in my last post CAT, its been fixed;
https://developer.oculus.com/blog/asynchronous-timewarp-on-oculus-rift/

 

[bru]

AMD have the superior hardware this time round. End of story.

Next time you look into your crystal ball can you Msg me the next euro millions wining lottrey numbers please, seeing as you obviously know this for a fact.

More seriously though. Just how anyone can be so sure about this is just beyond me. It's like nobody thinks NVIDIA are capable of building a decent GPU. Pascal has been written off before we actually know anything about it.
Thankfully it is only another week or so till GTC and hopefully some good info on Pascal, so we can see just how good\bad it will be.

 

[CAT-THE-FIFTH]

That was the case but as i said in my last post CAT, its been fixed;
https://developer.oculus.com/blog/asynchronous-timewarp-on-oculus-rift/

I get what you are saying,as I did read the blog post,but a fix means they can get the functionality working - what we don't know is how well. I believe the R9 290/R9 390 are on the same level featurewise as the GTX970/GTX980/GTX980TI but were the first to support many of the VR features.

The software based pre-emption methods will be quite CPU heavy though - it could be why both companies are pushing their own software to help.

Fiji and Tonga support the hardware based pre-emption methods,so I suspect they will be pretty strong longterm for VR support. This is why I suspect AMD is getting traction outside non-gaming companies for VR support.

The question is whether Pascal will support it or not. Polaris and Vega will though.

However,since I have no interest in buying a VR headset soon,it wouldn't bother me if Pascal has it or not. I am more interested in how well Async support is enabled on Pascal. From what I gathered reading on AT forums,from a poster who has a decent track record on predictions(I think they work in the industry),that Pascal show get Async support closer to earlier revisions of GCN.

 

[Rroff]

The question is whether Pascal will support it or not. Polaris and Vega will though.

The only public information on finer grain hardware pre-emption is "future GPUs" and doesn't seem to be referring to Pascal in context if you look at the whole presentation - its looking like the implementation in Pascal while more advanced than Maxwell will still be relatively rudimentary and while capable of beating the AMD equivalents require more dev hands on to do so.

 

[andybird123]

The software based pre-emption methods will be quite CPU heavy though - it could be why both companies are pushing their own software to help.

.

I'll be able to test that next week, elite and pcars are both being updated to SDK1.3 so I can try them both in normal and VR and see what impact it has on CPU usage

 

[lesoo7]

Without reading the entire thread, that's not all about Pascal. Can someone quickly tell me when cards are expected?

Just want to know when to start researching new screens, Cheers

 

[andybird123]

Without reading the entire thread, that's not all about Pascal. Can someone quickly tell me when cards are expected?

Just want to know when to start researching new screens, Cheers

the current rumour is that they will be announcing something at GTC on the 5th April

 

[lesoo7]

the current rumour is that they will be announcing something at GTC on the 5th April

Many thanks, better get my reading glasses on

 

[Rroff]

Its all just speculation at this point - with AMD releasing something (very) late summer you can be sure nVidia will be looking to release something in the same ballpark - probably a couple of months or so earlier.

 

[lesoo7]

From what pages I have read in this thread, most seem to think the little cards will come first. Nv last two ticks brought out the big boys first and I'm hoping they do the same again!
If the gains are not spectacular and it's a hard sell, then the titan's will emerge first to set the NEW ball park.
£800 - £900 GPU'S are good for defining perf / price and creating a sense of value for the smaller cards that follow. If it happened the other way round then nobody would buy Titan's we would all buy 1080's.
CHEER'S FOR THE INFO!!

 

[andybird123]

From what pages I have read in this thread, most seem to think the little cards will come first. Nv last two ticks brought out the big boys first and I'm hoping they do the same again!
If the gains are not spectacular and it's a hard sell, then the titan's will emerge first to set the NEW ball park.
£800 - £900 GPU'S are good for defining perf / price and creating a sense of value for the smaller cards that follow. If it happened the other way round then nobody would buy Titan's we would all buy 1080's.
CHEER'S FOR THE INFO!!

eh?
with Kepler they released the 680 first, then with Maxwell they released the 750 and 750ti first then the 980... the "big" Kepler and Maxwell debut was the Titan cards but they came a year after K/M originally launched

with AMD saying their mid/low cards are coming first, I find it highly unlikely NV will start with the Titan

 

[Rroff]

with AMD saying their mid/low cards are coming first, I find it highly unlikely NV will start with the Titan

Yeah any true Titan card would really need 16GB of HBM2 to be truly a Titan - which isn't likely to happen this side of the end of summer - most likely 2017.

 

[CAT-THE-FIFTH]

If AMD release their cards after Nvidia it is not going to end well for them,unless Polaris is much better than Pascal in performance and price-performance.