Lack of 8gbit gddr5 chips?If you can have in theory 8 GB with a 384bit bus, what would then stop you to go to 16GB again in theory?
The article I saw from the last week or so stated that TSMC was anticipated ~20% of their wafer revenue from 20nm in Q4 2014.
That is in-line with how 28nm ended up in Q4 2012, 22% of their wafer revenue.
Edit- Found the article.
Article from 2013 with quarterly wafer revenue broken down by node.
Last edited by a moderator:
D
Deleted member 7537
Guest
So no maxwell at computex. 
AFAIK there has never been a major GPU announcement from AMD or Nvidia at Computex/CES/Cebit/E3/...So no maxwell at computex.
This year's CES had NVidia's launch of the K1. That wasn't previously a secret, but the big surprise was Jen-hsun even showed off the first Project Denver version, appearing over a year earlier than the roadmap had indicated.
NVIDIA's big shows are CES (Tegra), their own GTC event (pro GPUs), any "out of band" events (consumer GPUs), and then possibly NAB and SIGGRAPH depending on what it is you're looking for.
I meant 'big GPU' announcement, not big 'GPU announcement'.
A1xLLcqAgt0qc2RyMz0y
Veteran
Doesn't stop the V I from spouting his nonsense again because Nvidia didn't release a new GPU at Computex:
It's pretty much pointless to debate an article behind a paywall.
Either we can't see it and are making stuff up, or people have read it and are violating subscriber agreements to discuss it openly.
Whether the lack of a big GPU announcement at an event that doesn't always have them is the justification for the claim, or running into insiders at an event full of insiders provided opportunity for leaks is unknowable from the teaser text provided.
Either we can't see it and are making stuff up, or people have read it and are violating subscriber agreements to discuss it openly.
Whether the lack of a big GPU announcement at an event that doesn't always have them is the justification for the claim, or running into insiders at an event full of insiders provided opportunity for leaks is unknowable from the teaser text provided.
This article has nothing to do with the lack of a Computex release. There also happens to be a Charlie love/hate thread for that sort of thing.
You heard the details from me first: gm100, 101, 102, 103, 104, 105, and 106 were on the roadmap and had working silicon but they were all canceled because... wait for it... bad yields!
You heard the details from me first: gm100, 101, 102, 103, 104, 105, and 106 were on the roadmap and had working silicon but they were all canceled because... wait for it... bad yields!
lol i will not comment it .. but bad yiels on 28nm or 20nm ?
( i tend to think that with Semiaccurate, there's allways bad yields somewhere, even when Jen hsun goes on toilet, semi accurate see bad yields there. ).
Many rumors are running around.. Sweclockers ( who can be trusty i can think, because they never really goes far on speculations ) say Q4 ..
But i read a lot of rumors thoses days, including a studio demo ( Epic ? ) on an Maxwell gpu (( not the 750 ), and report from here and there with 256 or 512bit or .. well impossible to know the true or the fake so i give up before we get real info . Because nearly the 100% is made by guys on forum (( expreview, xtremesystems etc ), and reported then . Its too much viral for me for believe then anything without real source and information . ( I dont say what they say is wrong, but without sources, just compilation of phrase from here and there, impossible to know ) ..
This said Q4 seems a plausible date.. but well it was allready plausible 6 months ago, and even 1 year ago .
https://translate.google.com/transl...0-med-maxwell-anlander-till-hosten&edit-text=
Last edited by a moderator:
http://www.google.com/patents/US20140118393
http://www.google.com/patents/US20140118380
http://www.google.com/patents/US20140118362
2nd gen is tile based rendering ?
http://www.google.com/patents/US20140118380
http://www.google.com/patents/US20140118362
2nd gen is tile based rendering ?
What makes you think this is for Maxwell? Although it could make sense given its emphasis on "built for mobile"...
A1xLLcqAgt0qc2RyMz0y
Veteran
Megadrive1988
Veteran
Could be for Pascal, Volta, or Nvidia's GPU architecture that is the furthest off, Einstein.
