Discussion in 'Frontpage news' started by Rich_Guy, Jul 31, 2014.
Plan on replacing a reference 290 on another rig (its too hot), might get this
I'll skip it. It would be nice if my GTX 780 had more vram (SC totally maxes it, although it runs great - newest patch has some impressive optimisations) but other than that all is good.
256bit memory bus? Not much use for 4K unfortunately.
I wonder what the Titan 2 will bring to the table, otherwise 9xx series will be the next upgrade.
tempting, if ti gonna do to 680 what 780 did to 680 then it might be a good place.780 after good oc is 55%+ faster than 770 which is impressive.... but still waiting for big daddy maxwell.
Now, I'm not much in to 3Dmark scores and such, but I just googled the Firestrike extreme score thing and found Futuremarks top 100 single gpu char. The 6100 points wasn't even enough to enter that. The #100 was a 780Ti running at just under 1200mhz, which scored 6400ish.
Another year...another x70!
I would not go for 28nm at this point in time unless my GPU burns, but in such case I may as well survive on iGP for a while.
As for power efficiency or performance. One should judge based on real world tests not scaling assumptions from stuff like 750Ti which I consider quite good, but we all know from past that beefer GPUs tends to be less power efficient.
Simplest thing is memory power consumption and transistors used for additional memory lanes.
128b GPU itself may eat like 5W less than 256b one and memories power consumption difference may be as high as 45W considering clocks.
In that regards one can say that 290(x) GPU itself is quite more power efficient than 7950/70 as it is ~=35% more powerful while eating 10% more because big portion of additional power consumption comes from additional memory chips.(they improved on memory controller side as it is actually smaller than one in 7970)
And what I meant by this is, that bigger GPU with more memory chips has lower available power for GPU to stay in same TDP as memories eat more.
Seems pretty good, only disappointing thing is the VRAM. Hopefully some non reference cards might come with 8GB.
Maybe the 880, but not the Ti. nv will probably act the same as they did with the 780Ti / Titan.
Looks like I'll be sticking to my overclocked 4GB 760's for the near future, just more time to save up the pennies.
£240 for 880gtx...sure I'll have 2!
Well the GTX 880 is definitely the next card I want as a buyer of the high-end single GPU options but I doubt I will be buying one in September as there are no new demanding games coming out then that would justify it. Besides, I bought a GTX 780 for £550 and then three months later the full-spec GTX 780 Ti was released for the same price so commonsense tells me that I would be better off waiting even if there were any games out that required a card more powerful than the GTX 780 to run it.
Hope the rumours about the GTX 880 being cheaper than the GTX 780 at release are true as well because I must admit that I felt very uncomfortable about paying £550 for a graphics card when £350-£400 was the usual price for new high-end cards. Of course, the GTX 780 dropped in price when the GTX 780 Ti was released...after three months.
P.S. If the GTX 880 is not the high-end card then I might as well wait for the inevitable Ti version that is faster.
Good news for me. I´m using a borrowed 4850 from my brother after my GTX570 died and i need a new card so it´s good that the 880 and 870 are close to being released. If they perform well and don´t cost and arm, leg and a kidney, i may get one, otherwise i will go for a cheaper 770/780.
I'll try to hold out till a Ti version, or follow up of a Titan, whatsoever. I wouldn't buy a reference design anyway...
This might be my next card. I always skip a generation so if it is faster than a 780Ti, then this is the way to go for me.
I'm not going to wait for a possible Ti version. No one knows if they have planned a 880Ti in this series.
Two 880 classy's under water :bannan:
Why would you need a card with 8GB and 256bit?
Anyone else a bit disappointed that graphics seem to be bandwidth-limited? By that I mean the fact that the total bandwidth available for rendering hasn't really increased in the last five years but is limited by either the speed of the GDDR5 memory or the address bus (or whatever it's called). The last thing I expected was for NVIDIA next card to go back to the 256-bit bus of the GTX 600 series. I was expecting 384 or even 512-bit.
Don't 256-bit cards usually struggle with anti-aliasing at higher resolutions (though I guess with MSAA/CSAA becoming less used over cheaper, inferior post-processing AA, it may not be that much of an issue anyway)?