But that's not due for mass production end of this year is it? I thought Tegra was first for mass production 16nmFF+ this year and Pascal mass production much later on in 2016. Have things changed or have I got it wrong?
Last edited:
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
** The Official Nvidia GeForce 'Pascal' Thread - for general gossip and discussions **
- Thread starter Boomstick777
- Start date
Soldato
- Joined
- 30 Nov 2011
- Posts
- 11,358
Theres rumours of a Q1-Q2 Pascal release, which would mean production starting end of this year.
TSMC are in volume production on 16ff now and plan to be starting volume of ff+ in Q3/15
Tegra X1 is 20nm and they've not announced a replacement yet, Pascal they have actually announced.
Last edited:
Soldato
I'm going to assume Pascal is going to be HBM.
Do we think Nvidia will use the same cooling setup on these cards that the Fury X uses?
I personally hope not.
Do we think Nvidia will use the same cooling setup on these cards that the Fury X uses?
I personally hope not.
http://www.extremetech.com/computing/212221-tsmc-quickly-ramping-up-16nm-volume-production
And it loos like TSMC will try to bring that back by a quarter or so.
But even the original timeline makes perfect sense for Pascal which is expected Q2. "Volume" production of 16mmFF+ should begin soon in theory,for smaller parts, wide availability volume production of big chip s by q1 next year looks incredibly likely.
It looks like the biggest bottle neck will be HBM and interposers. There's is 6 months or so for that production to get up to speed. For Hynix that should be enough but is tight for samsung.
No, they will be air cooled.
AMD went with AIO because their chip runs fairly hot (not entirely their fault, we are stuck on 28nm, but they lack the power efficiency of nvidia and the extra 2billion transistors pulls more power). Also to reduce power consumption they wanted to run the chip much cooler, higher temperatures creates electron leakage which increases power consumption and generates heat. The a furyX saves 20-30W by being water cooled.
They also wanted a really cool, quiet, silent card with no heat problems to hide the 290X hair drier fiasco. They got most of that right, and (their supplier) messed up with the pump whine.
With the next node chips will be smaller and cooler.
An AIO adds a lot of fixed costs and increases risks of failure. The FuryX costs a lot more to produce than a 980Ti so nvidia profit margins are going to be much higher. I don't think nvidia would go down an AIO path unless they had no choice.
Yeah 2 aspects to that - none the least the 28nm<>16nm FF+ change but assuming the focus is on producing something viable for commercial compute use they won't want a design that relies on AIO cooling to perform so probably won't be a pre-requisite for a gaming card either.
Soldato
The new GCN architecture is supposed to be twice as efficient at the same clocks. And that is before Finfets come into the equation from what has been rumoured.
All while having IPC improvements.
Associate
Probably Q1-Q2 next year, but i don't know whats with the rumours that NV will start with HBM a year later
(Realistic?)
Associate
I'm thinking Q3 next year at the earliest for the big boys
A year later from when?
What rumors, Have you got a link as it would bean interesting read?
Associate
Well i'd wait for some more credible source but on chinese/taiwanese sites there are news that NV will probably only switch to HBM in 2017
Soldato
Permabanned
I think the big Pascal gaming chip will be at the end of Q2 at the earliest
Announced end of March/April, with availability in April/June. This will all depend on the availability of the HBM2 and the interposers, ie: I don't think the main chip will be the thing that holds these up.
This will arrive as big Pascal first, probably in Quadro form, with a very expensive Titan style card very shortly after. next will come middle Pascal, which will be the mainstream chip ala 980, with the 980ti equivalent coming much latter in the year.
Disclaimer, this is all guesswork on my part. Hell what is the worse that can happen, I can be wrong.
This will arrive as big Pascal first, probably in Quadro form, with a very expensive Titan style card very shortly after. next will come middle Pascal, which will be the mainstream chip ala 980, with the 980ti equivalent coming much latter in the year.
Disclaimer, this is all guesswork on my part. Hell what is the worse that can happen, I can be wrong.