Nvidia GeForce GTX 900 Series MegaThread: FAQ and Resources

Page 7 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Nvidia's Pascal 'GP100 GPU' Reportedly Taped Out on TSMC 16nm Node - 'Big' Pascal and GP104 Arriving Q2 2016
http://wccftech.com/nvidia-gp100-chip-taped-gp-104-arriving-q2-2016/
...now it's gp100...

edit:
Geforce GTX 970 issue destroyed sales in February
http://www.fudzilla.com/news/graphics/37938-geforce-gtx-970-issue-destroyed-sales-in-february
turns out there is such a thing as bad publicity...

edit2:
NVIDIA Shows Off Advanced DirectX 12 Features, Ray Tracing, Volume Tiled Resources
http://wccftech.com/nvidia-shows-advanced-directx-12-features-ray-tracing-volume-tiled-resources/

 
Got my Gigabyte GTX 980 G1 Gaming a few days ago.

I think my Card is borked 🙁 Does anyone know of a site I can go to that lists different cards, their brands, and the known issues related to them? If not, hopefully a fix to my problem will be discovered soon D: I made a thread about it, hopefully someone with know-how can answer it when it's seen.
 


you can either start playing soccer to gain that knowledge from stephen chow or you can try browsing online retailers like amazon or newegg to read reviews of their products.

 
Can someone explain how on the earth this is possible?!

GTX 970 & GTX 980 does not have full DirectX 12 support. They are DX12 API (which means feature level 12_0) while their notebook series, GTX 970M and GTX 980M have full DirectX 12 (Feature Level 12_1) support.

GTX 980 Ti also uses same architecture and it has full DirectX 12 (Feature Level 12_1) support.

Nvidia's new marketing cheat to sell new graphics? :|

http://www.geforce.com/hardware/desktop-gpus/geforce-gtx-970/specifications

http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-970m/specifications
 


Probably just an error.
http://www.extremetech.com/extreme/207598-demystifying-directx-12-support-what-amd-intel-and-nvidia-do-and-dont-deliver

Maxwell GM2xx has 12.1 support
 


sometimes the info inside geforce.com are not updated to reflect current API support. for example OpenGL. when the card first comes out it list the latest support for the latest version of OpenGL at that time. but the doesn't mean it cannot support newer version of OpenGL (as long as the required hardware exist).

All second gen maxwell based chip have DirectX 12_1 feature level support.
 
Couple of months ago that was just "DirectX : 12"

These info became more specific after 980 Ti release.

Cause 970 and 980 sold quite well, I think they made this to sell their new cards.
 
Yes that's correct. There ie only DX12 right now (no such thing as Directx 12.1). In directx 12 there are feature level 12_0 and 12_1. It is a bit confusing since MS did not come more info about DX12 until recently. But feature level 12_1 are known to exist on maxwell V2 ever since the launch of GTX 980 and 970. If you read on the architecture analysis there was mention about this (on this matter i like to read anandtech review because they usually go a bit deep on architecture analysis compared to other reviewer)
 


AMD started loosing market share to Nvidia at a considerably steep rate sometime in 2014 and has been trying to win it back ever since.

I would need something a bit more official from someone who can spell.
 


if that's true AMD probably cursing nvidia right now.
 

it's not unprecedented that nvidia will initiate a price war...except amd shot first with fury launch and caused nvidia to price 980TI at a lower point than previous titan-derivative 780TI launched. according to at,
titan launched at $999, 780 TI at $699
titan x at $999 again, but 980TI at $649.
it's just $50 cheaper from 780 TI's launch price but the feeling is obvious. now nvidia is rumored to be considering further lowering the prices despite the 980ti being built using current, cheaper parts and fury being early adopter, similarly-to-lesser-performing, very (speculating) expensive new vram. 780ti's amd counterpart(!) was the hawaii r9 290x and it has $550 price at the time of 780ti's launch. add to that, nvidia just took a hit with gtx 970 sales due to the outcry over vram. to me it looks like nvidia has the room to lower prices but it's still the one cursing at amd.

edit: the winner here will be the people who's buy fury and co and gtx 970+ cards amidst the price war, as long as they're okay with stopgaps.
 
it's not unprecedented that nvidia will initiate a price war...except amd shot first with fury launch and caused nvidia to price 980TI at a lower point than previous titan-derivative 780TI launched

it was clearly the other way around. how can AMD make the first shot when nvidia launch their 980Ti first? if anything it was nvidia that taking preemptive strikes towards AMD and forcing AMD to price Fury at $650. and when talking about Fury pricing it is interesting. why 650 but now lower to undercut nvidia 980Ti? no doubt Fury prompt nvidia to priced 980Ti at 650 though. but it is not necessarily bad for nvidia long term. make a bit of sacrifice now to secure long term profit. after all nvidia use quite similar strategy with GTX580 launch.

add to that, nvidia just took a hit with gtx 970 sales due to the outcry over vramf

except nvidia market share continue to climb. and to a level it was never before.the fiasco might hit 970 sales a bit but it did not affect other card. and when we look at the forum some people willing to spend more to get 980 instead of getting 290X which depends on the deal can be had cheaper than 970 itself.

to me it looks like nvidia has the room to lower prices but it's still the one cursing at amd

nope they are not. they are killing AMD slowly. nvidia well aware about AMD financial situation. looking at booth Fury X and 980ti the former most likely costing more to produce. AMD most likely want to make as much profit as they can but nvidia move to price 980ti at 650 will going to narrow that margin. heck that even make AMD mission to gain back the lost market share even more difficult. right now intel probably going to pose more of a treat to nvidia than AMD
 

it's been bugging me since fiji's specs started leaking out: how much does amd save from HBM's implementations and does it offset the cost? simplified card design, lower cost due to on-package memory, oem-designed cooler (unlike nvidia's fully-custom one) all point to amd recouping cost of HBM implementation. with fury non-x and the nano on the way amd will have both higher volume upper-mid and high end parts and higher margin, flagship parts in fury x and the dual fury (if/when it launches). otoh, nvidia seems to have show all it's cards and will be forced to fight on pricing alone. i reckon it has a dual GM200 card waiting but other than that..
edit: wait, it can also release a dual GM204 card to push fury x/xx down but that move seems like a last resort thing and something amd has done in the past.
 
otoh, nvidia seems to have show all it's cards and will be forced to fight on pricing alone.

if only it was that simple. if price advantage alone will determine the winner AMD would leave nvidia in the dust long ago in term of market share. right now there are still some people that decided to take nvidia card because of stuff like PhysX. you know what does it mean right?

i reckon it has a dual GM200 card waiting but other than that

wait, it can also release a dual GM204 card to push fury x/xx down but that move seems like a last resort thing and something amd has done in the past.

GM204 can be a very good candidate for dual GPU but i doubt nvidia will make dual gpu using GM204. some board partner might be interested though. as for dual GM200...i don't know. nvidia probably prefer to sold full fledged GM200 as Quadro instead of Geforce. dual GM200 might not going to happen (this is what i believe right now) unless nvidia intend to break 375w limit. without doing so they most likely will not going to beat Fury X2.
 

my entire hypothesis about amd making moniez on fiji hinges on the assumption that amd is somehow recouping cost from BoM, OEM cooler and an old GPU. if my assumption turns out to be wrong, and if amd really is selling fiji cards at a lower-than-anticipated margin, nvidia will keep it's advantage.

are you talking about dgfx marketshare or overall gpu marketshare?

as for physx - it was more prevalent before but in maxwell's context it's not entirely accurate. the reason for that is that users don't necessarily have to choose a maxwell card for physx let alone a high end one. this is why i didn't bring up physx as i was talking about mostly high end vs high end.
generally, high end physx card users are far outnumbered by dual gfx card (SLI) owners and entry level/ recycled physx card owners who are in turn far outnumbered by single gfx card owners (as most oems use single card).