nVidia confirms it is unveiling new Pascal GPUs tomorrow

[bortas1]

  this good news. may have to wait and see how the cards are going to work out in me plans 

[Chalenge]

Chalenge, I haven't seen any release dates for B-E, I've been hoping it would be simultaneous with Pascal, just for my own reasons, but first week in June would absolutely work for me.

KitGuru and WCCF have both stated Q2, and now the June Computex as the big "reveal," even though the stats are already known and benchmarks posted. Probably just a rumor, right?

[Gman]

Maybe, I will say quite a few of the rumors, not all, but a surprising amount, have turned out to be reasonably accurate, at least in terms of Pascal etc.  Some not, but more than usual I would say.

Chalenge, you`ll like this, as you run SLI I know - apparently the new SLI system for the 1080s and 70s doubles the available transfer bandwidth compared to Maxwell and other predecessors.  If that turns out to be even 1/2 true, being a fellow SLI enthusiast (right back to the 12 meg Creative Labs Voodoo2 cards I bought specifically to play Janes F15E), I'm pretty happy with that right now.

[Chalenge]

I'm going to pull the trigger on the AUA MB and then try to get it stable with maxed out memory. I want to find a setup that runs stable with 256GB of memory and four way SLI. I can stick with the M6000s for now, but I am hoping Nvidia introduces Pascal to the WS line soon.

[Gman]

Some interesting stuff regarding the SLI bridges is coming out, looks like they are only making them for SLI 2 way, and not 3 way+.  Some rumbles here about higher than 2 SLI cards not being all that great.  We'll see at the end of the month I guess.

http://wccftech.com/nvidia-geforce-gtx-1080-pcb-gp104-gddr5x/

As predicted, a flood of 970 and 980/ti cards for sale on used forums, Hardware Canucks up here has at least 1/2 dozen popping up every day, and it's a relatively small site for used HW. 

I wonder what HBM2 will bring, if Pascal is as much a leap ahead of the last generation as predicted, HBM2 is going to boggle brains.

[Chalenge]

16k resolutions for multi-monitor setups (4x4), or 16k DSR is what I am hoping for gaming.

On the WS side just speeding up HD renderings by a factor of 1600 is a nice dream.

[bortas1]

16k resolutions for multi-monitor setups (4x4), or 16k DSR is what I am hoping for gaming.

On the WS side just speeding up HD renderings by a factor of 1600 is a nice dream.

4 monitors?

[Chalenge]

16 monitors (4 high x 4 wide), or the equivalent in resolution.

EDIT: Let me explain why.

At 8k you have reached the level that 'tech gurus' claim is the pinnacle beyond which humans cannot perceive any difference. So, at that point you can remove anti-aliasing and start pumping in extra resolution to provide greater clarity, while also enhancing shadows and anisotropy (where now a single pixel color represents the color fall-off of a leaf, for instance), and more important to complex models would be things like environmental effects upon reflectivity drop-off.

I think DirectX is limited in some of these areas. Reflectivity, for instance, seems to have a single value, rather then a maximum-minimum magnitude (falloff) which is a closer representation of reality.

[BoilerDown]

Have a GTX 980, and unless I make a killer deal to sell it to someone and the 1080 benches really well, I'm waiting on the 1080Ti.

P.s.  From what I read AMD is completely sitting out the high end.  Again.

[Gman]

Don't count AMD out yet, HBM2 is coming from them as well - plus they have stuff to counter the 1080, I'll just be really surprised if they beat it, or tbh match it.

Boiler, the 1080 is going to be a huge leap from even the 980ti, the 980 for certain, a single 1080 is supposedly faster (hard to say in what or how exactly) than 2 980 in sli but a leap.  1080ti - you may as well wait for the HBM2, in fact, IMO, the 1080ti will likely BE the HBM2 first card they release, but that's just a guess, but a measured one.  It'll be well into 2017 most like.

It's personal pref, but I'm selling my 980sli equipped system for a cheap deal and moving to 1080sli in a new box, and will possibly sell all the vid cards in our other systems, 980ti sli and a Titan for 1080s as well.

1070 will be the card that really sells I'll bet, better than current Titan performance for 379 bucks, crazy.  Too bad VR and SLI don't get along, as a 1070 sli system would be a very cost effective solution, it may in fact be the best bang/$ out there for guys running 4k or 1440p 144hz LCDs.  I'm going to run a 34" 1440p 100hz panel with my 1080s, and my ultimate goal would be to run 3, or 3 27" Asus ROGs, as I have 2 already.  Apparently they've done a pile of improvements on the nvidia surround stuff for 3 lcds, and I know a TON of Aces High players run 3 panels, HT even said so based off their hardware survey that the program does when you install/update/etc. 

Good year for Aces High 3 to come out IMO, big time.  Fantastic new hardware, and a lot of really capable current gen stuff that will have prices drop like a rock, and which will also have a massive buyers used market. 

[FLOOB]

You're talking toejam. I know Pascal and he knows diddly all about microchips and toejam.

[Shuffler]

I'd be happy if NVidia could figure out how to quit the flickering jiggy shadow in the GTX 960 - 980 cards

[Gman]

Shuffler, Gsync does help with that, it doesn't completely eliminate it, but it does improve a lot.

AMD's Polaris 10 looks like it'll be 299$ and possibly be a 1070 competitor, about 80$ cheaper with similar performance.  Likely won't touch the 1080 though. 

[hgtonyvi]

I'd be happy if NVidia could figure out how to quit the flickering jiggy shadow in the GTX 960 - 980 cards

This

[bortas1]

  so gents should we really believe all the hype about the 1070 and 1080? is anything really proven about them yet?