TSMC current 20nm process issues have appeared to reach a crisis point, as current information suggest Nvidia has thrown in the towel over TSMC's stranglehold and development (read TSMC screwups) cost price gouging and redesigned GM20x for the current established 28nm...
 
This means -NO- full powered successor to the 780 till 2Q 2015--> at the earliest
 
Expect Nvidia to shovel 28nm low-mid rage Maxwells at full price without shame until sometime next year
------------------------------------------------------
 
Terrible news indeed
 http://www.techpowerup.co...s-to-launch-in-q4.html

well the first thing is NO SOURCE 
3Dcenter.org is just reposting a post from a person on the forums - and it's pure speculation
3Dcenter - "From our forum " - and they even cite the source as speculative
 
I alluded to this in another post - watch out for FUD
Nvidia earlier this year had said in a conference that they would stick to their Timeline and role out Maxwell in june (Although it could be moved up due to TSMC being early)
 
Nvidia reworked their silicon design - which is why it so efficient - meant for SOC silicon
knowing TSMC fab process ahead of time that they were only going to have SOC planar
 
AMD did not do this - and why they had to trash their GPU and rework it and why they will be late
 
the idea that there is so much re-posting of one source without looking for 2nd or third independent sources -
and "NO" you can't count different sites as new sources that use the same single source
and/or using an opinion "as a source" without some credible backup - a credible source... hard info - a reputable person on record or a site
 
What I did hear is 20nm may be reserved for 870,880 and titan and 28nm for everything below
that's possible - then the shortage of 20nm will be reserved for top end at first
 
As I said before this is production tape-out ... not test tape-out
 
They have 1000's of chips already - more like several - maybe as high as 10k - this was first tape out for testing
and why you see gm200, gm204, gm206 in the drivers already - they have already reworked the drivers to handle Maxwell and are in testing and driver refinement stage
you don't place them in unless you are testing them and have partners reviewing them 
 
why if because they aren't ready you do testing inhouse without release to drivers -
to keep problems out of the release drivers and to not give heads up to your competitors/and Press to how soon you are to go
 
furthermore you have daily rebuilds or multiple rebuilds to get the new drivers working on new silicon
to get rid of bugs in-house - because you'll have plenty and some may bugger up existing drivers and add new problems
 
last nvidia production tape-out - Kepler - led to a release a couple of months later
same will follow here
 
as noted the TE GK210 chips for titan Z and 780Ti were in feb and are being release already
 
you don't order masses of product without wanting to go to release soon
and keep stock sitting on shelves costing $$ and getting no revenue
 
 
 

We are reaching the technological limits of our lithographic capabilities.  In summary, it's almost as simple as that.  I think the popular habit of blaming the chip manufacturer for falling short on demands/expectations is rather childish when the reality is that we need a technological breakthrough instead of a spanking.

Well if that's the case then I'm happy that I ordered my two 780 Tis (I have step up for 2.5 months).

Bruno747
Problems with 20? heck they were having issues galore with 40, and issues with the one before that. Yes we need to make a jump in technology to the next thing, but if its so terribly bad how does Intel do it time and time again WAY ahead of TSMC? I mean Intel is getting ready to launch 14nm chips fairly soon, yes most of them have a smaller die than a gk110 type chip but really? Its all the process and not tsmc?

differnt fab process, and intel measures nod size differently, thare are many many reasons. hell intel is still does not make 20nm in mass quantities yet. 14nm actually got a large delay last i heard.