Yeah, no. If Big Pascal releases this year I'm gonna be very surprised. Certainly not the GP100. Maybe they're bringing a 1080ti in form of the GP102 but I don't know if I want that one.

Can u explain how they relate to each other, GP100 and PG102? Is it again that they correspond to Titan and 980Ti somehow (and GP104 corresponds to the GTX 980?).

How do you suppose the 1080ti would be if it came "in the form of GP102" and what do u not like about it? The size of the chip? Number of transistors?

from what i've been hearing by the techrumor circles, the Vega card should be out this fall in time for Battlefield 1. i'm patiently waiting to see the advantages of the new architecture with HBM2 and some decent VR games to make the case for it. I'd say i'm looking for an excuse to upgrade from the r9 290x OC and obviously the rx 480 not it

from what I heard it's rumored the GP100/Tesla P100 will be pro-only as a big portion of the transistors seem to be solely used for double precision calculations which are unnecessary for gaming

Big Pascal will probably hit in Nov/Dec range if not a little earlier, it just won't be 1080ti, it will wear a Titan badge and cost around $1k. Simply because they can, and there's zero competition at the top end. NVidia knows they have a core group of die-hards that pay huge $ to upgrade to whatever the latest is, and right now they're giving away their current top card for 800ish (soon to drop a little bit when the scalping frenzy fades), obsoleting all of their previous Titan products in the process. They won't let that sit for long, it's just leaving $$ on the table. But there are a suprising number of people who simply jump from every top end card to the next, even if it's only a couple of months later. It's not a gigantic market, but it's a profitable one. 

I still think it's premature to call 480 a fail, but AMD's stock cooling/PCB on the thing is pretty meh. I had a stock 290X on release, and it was pretty bad. Then later on I installed some good aftermarket 290X cards that were drastically improved in heat/noise/clock along with drivers. Hopefully the 480 goes through the same improvements. I'm sticking with my 970 and 980ti mix for a while myself though. Vega or 1080ti a year from now maybe. I don't do VR and the current games aren't really an issue at settings I use, so why bother wasting $ on it.

I disagree. I saw the writing on the wall a long time before that, people certainly blew it's expectations up.

It was meant for a good $200 1080P/VR experience and AMD peddled that since it's announcement, people expected more, it was almost a plague on this forum when people expect a 5Tflop GPU to be a 4k gaming chip based on flops alone.

Global Foundries 14nm is *slightly* superior to TSMC 16nm, it has a smaller gate pitch and sram feature size, Global Foundries is just struggling to get acceptable power characteristics at this stage, that will be resolved in time.

And yes, performance from a node can mature, leakage can be reduced which results in less heat for instance.

28nm saw constant improvements. (And still is even today)

^^ Intel's 32mn and 22nm also improved from stellar to incredible. It would be really scary for AMD and nVidia if Intel gave a damn about GPU lol. However, their decision to go from soldered IHS to using mediocre TIM as a profit-boosting routine, I was disappointed. I went from a 5Ghz SB i7 2600k to a 3770K that was hard to get past 4.4Ghz 24.7 stable on thermal virus testing. Decapping and shaving got me back to 4.8Ghz, but never 5ghz totally stable. A more mature stepping indeed got me back to 5Ghz, and then Devil's Canyon got another 5Ghz product for me that was a lot easier to achieve. I only do air cooling as a matter of principle, so it takes a lot of precision and trial and error.

Sandy Bridge was a fantastic chip all around, still is actually, hence why I am still keeping my 3930K, overclocked it's still out-benching allot of people with Broadwell-E at stock, seen no reason to discard her yet.

Yeap, great stuff. I'd probably not notice much had I kept my trusty 2600k to be honest. With Intel focusing on mobile tech and power consumption rather than increasing raw performance, I am skipping everything until there is something that will really make me notice an upgrade in IPC from my 5Ghz 4790K. It's looking like that will be a while, which is great I guess. I do think that by this time an 8C/16T 5Ghz capable '7900K' should be out in the $350-$500 range, but with AMD basically irrelevant in the CPU space, and games not often taking advantage of more than 4 cores, that seems unlikely for at least a couple more years :/ I ran a dual hex Xeon (one of the older socket 1366 setups) for a little while, and even with 12 Cores and 24 Threads at 3Ghz, my Sandy Bridge was a better gaming setup in basically every scenario.

AMD won't be relevant once Zen releases either, but it will make up some ground which will be good.
I wouldn't be surprised if my 3930K @ 5ghz still out benches them though. :P

I'm doing the same, not upgrading my CPU/Motherboard until it's worthwhile... Otherwise you throw away money for nothing.