sblantipodi
2[H]4U
- Joined
- Aug 29, 2010
- Messages
- 3,765
Hi all,
should we expect high end pascal for christmas?
If not, when?
should we expect high end pascal for christmas?
If not, when?
Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
I would expect it by 2050. I think it is a safe bet it'll be out by then.
Seriously, nobody outside nVidia knows, and even they might not. Depends on how soon they can get all the issues in the design worked out, how soon it can be produced in sufficient quantity, and when they feel is a good idea market wise to release it.
Is Pascal going to be 16nm?
Hi all,
should we expect high end pascal for christmas?
If not, when?
Dont forget that AMD have "priority" in the HMB2 parts.
Most likely, yes. It's been linked to TSMC 16nm, and TSMC would be ready to produce these 16nm chips anyway, so it wouldn't make sense for nVidia to stick to 28nm.
Probably 3Q16 for 1070 and 1080 (or whatever they designate them as), late 1Q17 for Titan, and 2Q17 for the Ti parts is my guess. They'll probably follow the same release schedule as Maxwell.
NVIDIA has said that they're more interested in putting out the full enthusiast chip first this time, so I would expect the Pascal Ti or Pascal TITAN first (GP100).Probably 3Q16 for 1070 and 1080 (or whatever they designate them as), late 1Q17 for Titan, and 2Q17 for the Ti parts is my guess. They'll probably follow the same release schedule as Maxwell.
Given NVIDIA's release history, I would expect a release during the third quarter of calendar year 2016 (July - September).
By historically, I am talking about the amount of time that passes between major releases, which has historically been 14-16 months.What release history are you talking about?
GTX 750: February 2014
GTX 680: March 2012
GTX 480: March 2010
8800 GTX: November 2006
I'm not seeing any tendency historically to launch a major new architectureline around late summer.
Really, Nvidia has long since gotten as many sales as they can out of Maxwell. The prices have not fallen because there's no more costs they can really cut (same reason AMD ddidn't start a price war with the 390 series). So if they want to keep up the impressive product sales of earlier this year, it will benefit them to get it out asap next year.
So they'll get one last holiday season of Maxwell for the last of the holdouts to feast upon, then it's in with the new. One other benefit is that being released in spring allows plenty of time for OEMs to design new platforms showcasing the new hardware before the holidays.
By historically, I am talking about the amount of time that passes between major releases, which has historically been 14-16 months.
6 Series = April 2004
7 Series = June 2005 (+14 months)
8 Series = November 2006 (+17 months)
9 Series = February 2008 (+15 months)
200 Series = June 2008 (+4 months)
400 Series = March 2010 (+21 months)
500 Series = November 2010 (+8 months)
600 Series = March 2012 (+16 months)
700 Series = May 2013 (+14 months)
900 Series = September 2014 (+16 months)
So I can concede that this would point to a Q1 release. But the fact that it was rumored that Pascal just taped out in May and it's on a completely new node, I would give it at least a year before production volume is at the point where NVIDIA can launch it.
By historically, I am talking about the amount of time that passes between major releases, which has historically been 14-16 months.
6 Series = April 2004
7 Series = June 2005 (+14 months)
8 Series = November 2006 (+17 months)
9 Series = February 2008 (+15 months)
200 Series = June 2008 (+4 months)
400 Series = March 2010 (+21 months)
500 Series = November 2010 (+8 months)
600 Series = March 2012 (+16 months)
700 Series = May 2013 (+14 months)
900 Series = September 2014 (+16 months)
So I can concede that this would point to a Q1 release. But the fact that it was rumored that Pascal just taped out in May and it's on a completely new node, I would give it at least a year before production volume is at the point where NVIDIA can launch it.
The two real fixes, doubling the vias and redesigned circuits to minimize the impact of transistor variances both require a full re-layout. Both also cost time and die area. The time is at least six months from tapeout, if you recall. Fermi taped out in late July and was only slated for small numbers at retail in late November, a very unrealistic goal. A B1 spin of the chip would be at least Q3 of 2010 if it tapes out today, and it won’t have a useful life before it is supplanted by the next generation of 28nm chips.
Will be a huge improvement over the 980ti
I dunno, their worst-case chip on TSMC's most temperamental process ever was taped-out and released in in just 8 months.
http://techgage.com/news/is_nvidias_gf100_broken_and_unfixable/
You are truth bending, they knew that they werent going to have fully working chips at launch. Which means that they were working on it longer than 2 weeks. On top of that it was a revision of an existing chip not a brand new design. A better comparison would be the GK110, the chip took about an entire year before fully functional chips were made available to the consumers(dont remember how long for enterprise). That is the level of difficulty that should be expected on top of the fact that GK100 was never released and had to be redesigned into GK110.
I dunno, their worst-case chip on TSMC's most temperamental process ever was taped-out and released in in just 8 months.
http://techgage.com/news/is_nvidias_gf100_broken_and_unfixable/
You are truth bending, they knew that they werent going to have fully working chips at launch. Which means that they were working on it longer than 2 weeks. On top of that it was a revision of an existing chip not a brand new design. A better comparison would be the GK110, the chip took about an entire year before fully functional chips were made available to the consumers(dont remember how long for enterprise). That is the level of difficulty that should be expected on top of the fact that GK100 was never released and had to be redesigned into GK110.
Meh. NVidia executes--that's what they did then, that's what they will continue to do.
Or in the words of Heartbreak Ridge: "We improvise! We adapt! We overcome!"
Meh. NVidia executes--that's what they did then, that's what they will continue to do.
Or in the words of Heartbreak Ridge: "We improvise! We adapt! We overcome!"
I am not saying they don't, but I find it hard to believe on the consumer side that GP100 will be the first card to drop. All available cards will go to filling government/enterprise contracts and then after that it will become available to us as a Geforce. My guess is a repeat of Kepler.
On another note any guesses on whether the Pascal is Fermi/Kepler/Maxwell derived or is it going to be a completely new arch?
I am not saying they don't, but I find it hard to believe on the consumer side that GP100 will be the first card to drop. All available cards will go to filling government/enterprise contracts and then after that it will become available to us as a Geforce. My guess is a repeat of Kepler.
On another note any guesses on whether the Pascal is Fermi/Kepler/Maxwell derived or is it going to be a completely new arch?
from what I have been reading its a new design.