Nvidia GeForce GTX 1000 Series (Pascal) MegaThread: FAQ and Resources

Page 123 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
yah not a lot of new news to talk abut with pascal. other than folks waiting on a 1080 ti, there is little going on. still keeping 1st page updated with new models but little else to do with it. finally got time after holidays to get back at it and am setting up time this weekend to go through those posts again.
 
I decided a 1080 was way off being needed. My 980 is doing fine anyway. I did just buy The Witcher 3 Season Pass, though and it makes the 980 struggle a bit. It can do 1080p at a constant 60fps I think. (Not rightly sure as I can't remember if it dropped a bit under now and again.) However it won't hold 1440p at a constant 60fps. In the end I would rather play at 1080p than buy a GTX1080 to play it at 1440p anyway.

I didn't know that the 1080 Ti was late so to speak. When it arrives though, I bet it will push down the 1070 and 1080 prices a bit.

I have not been intensely gaming recently either. I played Dark Souls (1) which kicked my head, and that's putting it politely. I am replaying a bit of Hard Reset so no need for more GPU power really. No need for £650 worth of GPU power anyway.
 
For anyone interested I did add that waterblock to my EVGA 1080 Gaming ACX3.0 SC. Was only able to get another 100Mhz or so out of it, and '400' on the memory. But it does stay at that max clock under all conditions. So no throttling.

2121Mhz, 10388 effective memory is what I settled on for 100% stability. 2134Mhz was about 99% stable, tended to crash on closure of games or alt-tabbing for some reason.
 
Well the card at stock settings would thermal throttle, and I never changed that or made an attempt to overclock, there wasn't much need. It was set to back off at 75C, but still hovered around 1900Mhz under load. Super quiet though.

I pretty much did it because I already bought the block, I wasn't expecting a big performance boost.

Even at 5Ghz the 7700k only beats my i7-4770k @4.3Ghz by about 20% in cinebench.

I built it because it looks cool, and now I have to figure out what to do with the old system. i7-4770k, Hero VI, 2x8GB 1866, GTX980 (Single reference or water cooled, or 980 SLI watercooled, haven't decided) pair of 256GB OCZ Vertex 4, Evo 212 with SP120 in push pull, and a pile of corsair fans in a Bitfenix Ghost. Probably set it up this weekend.
 


There are a LOT of pascal notebooks with 1080P screens that can easily handle 1440P. It's almost as if they WANT you to pay another $500 for a good 1440P panel to play on. :lol:
 


Yeah I guess for a good while. However I was thinking about generally games coming out in 2017. I bet many will be utilising a lot of the 1070/1080 8GB vRAM. Quite a lot of games now use all the 4GB of my 980.

Many gamers too are using higher resolutions, in either better monitors or Nvidia DSR.

Whatever though, I'd be happy if a bought a 1080 laptop, as long as paying the premium didn't bother me.
 
GP100 finally used in Quadro:

http://www.anandtech.com/show/11102/nvidia-announces-quadro-gp100

to some this might be unexpected. because there are speculation that GP100 is a pure compute chip that the chip does not even have things like ROP. turns out GP100 is still a graphic card after all. single precision wise it is slower than than GP102 but GP100 was expected to have more ROP than GP102. it will be interesting to see if those extra ROP can out gun GP102 in rendering (or gaming) or not. though make no mistake. this is not a gaming product. Quadro GP100 will be offering a few things that are not offered with Quadro P6000 such as ECC memory. so we better don't expect nvidia to release geforce based GP100 in the future. one interesting feature of this new Quadro is the NVlink finger at the side of the card. this will allow the GPU to communicate using NVLink despite having PCI-E interface (this does not exist in tesla P100 PCI-E version). will nvidia upgrade their SLI tech with NVLink in the future? although the prospect of multi gpu are not really bright as of late.
 


Not until AMD gives them a reason to put HBM2 in consumer cards, I'd say. Plus, GDDR5X doesn't seem to be holding the consumer GPUs back.

I personally rather have a cheap card than an expensive one that doesn't benefit from HBM2. Case in point: GDDR5 was used for the RX480 and 1060 (even 1070) for performance and cost reasons. It would have been great to have HBM2 in them (or even HBM1), but the performance and consumption benefits would not justify the price increase in production, I'd say.

Cheers!
 


tesla P100 has been using HBM2 since last year. but on nvidia side of things they have no reason to put HBM on their gaming card. at least not yet. also there is no absolute advantage having HBM on gaming card right now.
 
also noticed a new cooler from EVGA http://www.evga.com/articles/01084/evga-icx/ called ICx. saw something about it a long while ago but it's here now.

saw a review for the 1080 card on the main page a half hour ago but it's been removed for some reason. i expect it to be back soon enough. reading now to see what the changes are supposed to be for the FTW2 model other than the cooler

new pcb with lots more sensors plus independent fan control depending on what's hot or not. seems interesting. bottom of the link lists 10 models with this new set-up 1060/70/80
 


VCZ just pick MSI advertisement for their article but they said this new mini ITX card was coming directly from nvidia. meaning this idea is coming from nvidia and they want all their board partner to do it. though they also said it is based on existing gpu so it will not something that unheard of before.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom