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

Page 134 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.


at least i think so because AMD is quite successful with their althon before. also in 2008 AMD is just start feeling the heat from the massive cash they spend in order to get ATI under their roof in 2006. it is not until a few years later that we see the true impact.
 



Just a bit confusing, because low profile has always meant half height slot.

The reason workstations cards can be single slot more often is because the clock speeds will usually be running much lower, so the heat and power requirements are lower.
 


they point some interesting point about this upcoming volta. they said there won't be any massive change to the architecture like how maxwell to pascal is not a massive change. though right now i see some people speculate that volta could be nvidia massive architecture change since Fermi. and this rumor also talking about even more expensive Geforce? meaning GTX2080 or whatever nvidia want to call it will be even more expensive than $700 1080ti? the reasoning is pascal did not generate much profit like nvidia expect. make no mistake though with pascal generation nvidia did generate more revenue than they was before but it seems nvidia expect get even more. so this early volta release is set to increase nvidia revenue by selling them at much higher price.

personally i still can't believe this at all. usually my stance is "don't deny the possibilities of a rumor end up being true" but not on this one. i will take it more seriously when VCZ start talking about it.
 

Some more news and analysis here from financial site The Motley Fool:
https://www.fool.com/investing/2017/04/17/report-nvidia-prepping-new-volta-architecture-gami.aspx
 


right now the only way to use NVlink on consumer grade product is for multi gpu setup. and it is setup that many people not interested with. NVLink can also improve CPU to GPU communication but that means nvidia have to replace PCI-E interface on their GPU with NVLink. and that also going to affect motherboard side. need mobo that an support NVLink specifically just for nvidia GPU? that will be a though sell.

 


in any case more info will probably going to surface next month with nvidia main GTC event.
 


I bet you would have said the same thing about GSync.

It's hardly about the expenses, but how much more performance/efficiency brings to the table. Part of the information I was told basically explained how there were going to implement it next to the PCI and not *through* it. I can say this much; if nVLink is all I was told, it might have a market.

Cheers! 😛
 


there is tons of GTX1080ti review out there. go read for the reference design review and then read at least one of the custom card review. then you'll know where the "sweet spot" for the card.
 
The 1080 and are still in fairly short supply on Amazon UK. I guess they just are not selling that many because of the high prices. Usually there are all makes and all models of the cards. Many are stocked but some are only a few left.

The 1080 Ti are barely in stock.

It's a bit of a mess really with these high prices. (I sometimes think that because places like the US sell electrical goods so cheap (meaning no tax). Maybe it increases prices in the UK. I might be wrong. However if Nvidia call sell so many in the US, then they have made their profit. They are not fighting for sales, so no need to be competitive over prices.)
 
Just switched yesterday from a KFA2 1060 OC to a KFA2 1070 EX that I bought for €350. I have changed my display couple of weeks ago, it's a 2560x1440, so the 1060 was fighting a bit I guess to fill all those pixels 😀 The 1070 does a great job, got it at a very good price I think. It's totally silent also. I really vouch for this brand :)
 
so nvidia using TSMC 12nm process is rumor no more. Volta GV100 will be built using this new node. and GV100 is the biggest "GPU" nvidia ever created today. 815mm2. and here i thought that 600mm2 will be the absolute limit they can go regardless of process node.
 


I want to buy a Vega card BUT I decided long ago that playing the waiting game is just silly. You can wait until you're in your grave if you really want the most bang for your buck. Same goes with cars, you can wait for a 2018 model and it'll probably be better than your 2017 model and possibly for the same price. Point being, why wait? You can only 'future proof' for so long.
 


LOL, we pay tax on everything. However, we can get out of paying tax sometimes buying through E-tailers.
 
That is indeed impressive, but a wash all the same. I'm sure it'll make some mini case owners happy, but they won't perform/boost as well as their big 2 and 3 slot brothers. If I had an itx, I'd just stick with a 1060. No point in getting something like that(mini 1080ti) when I can OC a 1060 to about the same performance level for much cheaper. Zotac shot themselves in the foot on this one.
 


even if the mini will not boost as the other after market 1080ti the base performance should not be slower than stock 1080ti. even Oced to 2.1ghz there is no way 1060 can reach 1080ti base performance.