AMD Won't Be Selling 20 nm GPUs This Year

Status
Not open for further replies.

sha7bot

Distinguished
Dec 12, 2009
36
0
18,540
Did AMD give a timeline for FinFET? Will they be using Global Foundries for this? Why is Intel not using this, but instead working on 14nm via Tri-Gate? Moar info!!
 

sha7bot

Distinguished
Dec 12, 2009
36
0
18,540
Did AMD give a timeline for FinFET? Will they be using Global Foundries for this? Why is Intel not using this, but instead working on 14nm via Tri-Gate? Moar info!!

Guess I should do some research before I post... Tri-Gate is Intel's version of FinFET. So AMD is will be three or four generations behind when they finally manufacture on it.
 
G

Guest

Guest
The fab process is not the only path to better products. I'm sure between now and then, especially with the onset of Mantel and DX12, both AMD and Nvidia would be offering competitive offerings with more features and improved thermal/power efficiency.
 

aggroboy

Distinguished
Sep 17, 2010
197
0
18,680
Did AMD give a timeline for FinFET? Will they be using Global Foundries for this? Why is Intel not using this, but instead working on 14nm via Tri-Gate? Moar info!!
If anything, it seems like AMD may skip planar 20nm and wait for FinFET.
 

jasonpwns

Distinguished
Jun 19, 2010
415
0
18,790
This is a shame, but I guess it was expected. I was looking forward to a 20mm 800 series this year from nvidia and I'm sure AMD fans wanted 20mm too. Disappointing for customers on both ends, but I suppose better to have it done "right" than completely wrong.
 

Shin-san

Distinguished
Nov 11, 2006
618
0
18,980
I'm not surprised that both AMD and nVidia are affected. They use the same fabs

amd should ask ibm, they have a working 22nm soi process that's producing the new power cpu.... :whistle:
IBM is in a sharing alliance with GlobalFoundries. It's definitely possible, though
 

KelvinTy

Distinguished
Aug 17, 2011
194
0
18,690
Does this mean the gap between AMD APU and Intel CPU with igp will have less of a gap?
Damn, AMD is losing battles in all fronts!
 

no, the gap won't lessen. amd has made a big performance jump with kaveri's igpu. intel's igpu is weaker by design and gets worse support from drivers.
that's not the most important reason. the most important reason is price. intel's top desktop sku with iris pro cost more than twice kaveri's price and comes soldered to the motherboard.
 

Haravikk

Distinguished
Sep 14, 2013
317
0
18,790
Personally I'm hoping their focus is on pushing FinFETs to market; a 20nm process would be nice, but FinFETs seem to have much higher potential gains. It's not like we're exactly starved for performance right now, the main issue IMO is the cooling that's required, and while a 20nm process could improve the situation, FinFETs ought to do a much better job of it.
 

Zepid

Honorable
Jan 14, 2014
80
0
10,630
Don't worry, I'm sure AMD will release the 7xxx series again before we get anything at 20nm.

I :lol:'d
(being that nVidia simply OC'd and re-badged the 600-series)

Are you slow or just stupid? It is called the tick-tock strategy and every tech company uses it. You release a new generation of processors on a new die shrink, then you release a 2nd generation of it under optimal conditions/tweaks.

That is how it has always worked, that is how it will always work, until we hit a wall with modern computing and have to move on to the next great thing.

400->500
600->700
800->900

Same thing with CPUs and same thing with AMD. They all do this, it is how you operate.
 

CraigN

Distinguished
I :lol:'d
(being that nVidia simply OC'd and re-badged the 600-series)

Um.. No.
GTX 660: 960 CUDA Cores
GTX 660 Ti: 1344 CUDA Cores
GTX 760: 1152 CUDA Cores. Smaller chip, faster speed than the 660 and 660 Ti, wider memory bus (256 up from 192)
GTX 670: 1344 CUDA Cores
760 is faster than the 660 Ti but slower than the 670, at less cores than the 660 Ti. It's a different chip with optimized throughput, as it has less SMXs but wider memory and better performance.

GTX 770 has 7gbps memory modules, the GTX 680 only has 6 Gbps memory modules. That's not achieved by a "simple overclock".

The GTX 780 and 780 Ti are completely different chips too.
The 750 Ti is a Maxwell chip, the 650 Ti is Kepler, completely different architectures.

You couldn't be making a worse generalization. At least it's not as bad as the fact that the HD 7970 can actually CrossFire with a R9 280X because they're basically the same card...
 

Chris Droste

Honorable
May 29, 2013
275
0
10,810
i think the problem now is we're AT the wall. the only things we've really seen in the past year is AMD factory OCing the hell out of their latest processes yielding 200W CPUs and hotter than hell Hawaii cards in a fit of ...desperation? perhaps to stay ahead in GPUs and to catch up in CPUs. Overall i really don't feel either is paying off particularly well. Intel seems to be in the same boat but since they're already ahead in CPU, they're just continuing to work on power consumption refinements. nVidia is playing the same game, they're not really innovating all that much as it stands either. we're getting Talk of MAXWELL delays now? and all the current gen is just refined previous gen parts?

I've been biding my time for when someone finally makes a leap and heats the market back up before buying a new setup but I'm so disillusioned at this point I feel like a 4130 with a 750Ti is going to be the setup to have...and this is coming from someone who used to PLAN a full high-end build every 4-4.5years almost exactly on the dot to accommodate the hotest, latest games.
 


Actually the GTX760 is a rebadged OEM 660, it's not a different chip.

http://www.geforce.co.uk/hardware/desktop-gpus/geforce-gtx-660-oem/specifications
 
Status
Not open for further replies.