In case you expected a lot from the GTX 980

[kuddlesworth9419]

This is speculation isn't it. If you actually had a brain cell you would probably see how Nvidia and AMD have released GPU's pretty much every year and every year they are better than the previous series of cards so what makes this year any different. If anything the gap between the 700 and 800 series cards will be bigger then the gap between the 600 and 700 because of 4K and the need for cards that can run games at 4K playable framerates. 

[Phalolia]

In your mind... Think about who Nvidia actually develops architecture for: corporate users in servers/supercomputers. Performance is less importance than perf/cost and perf/watt which adds to cost. Gamers are so conceited. Nvidia and AMD are wise to ignore you. You're not a market large enough to sustain either company, much less both. If you can pack the same performance into a cheaper package then large-scale architects will buy them.

The 880 will beat the 780 purely due to its much higher core count and low thermal output. That means clock rates can soar back into the 900MHz range and up,

Gotta ask you, tihnk they make the gamer cards for servers? you think the Main part of say 760´s are in servers ? or supercomputers? then apply it to 770 u think those are in ALOT of super computers ? and then finally the 780 ?`do u tihnk that is in alot of super computers if 760 or 770 are viable choices 

 

EDIT: btw just thought if this the 860m-880m these things got to be tailor made for gaming 

 

note im not saying they dont make cards for corporat users just saying thats another lineup

[patrickjp93]

Gotta ask you, tihnk they make the gamer cards for servers? you think the Main part of say 760´s are in servers ? or supercomputers? then apply it to 770 u think those are in ALOT of super computers ? and then finally the 780 ?`do u tihnk that is in alot of super computers if 760 or 770 are viable choices

EDIT: btw just thought if this the 860m-880m these things got to be tailor made for gaming

note im not saying they dont make cards for corporat users just saying thats another lineup

Nvidia starts with server designs and then trickles the tech down. It's the same thing Intel does. Gamers don't get the best. They get "good enough" in the stack of products made, which is a test run of high-end stuff while they work out the kinks.

For mobile you're getting the reject pile: dies that don't have all cores functioning, but are cool enough to be sustained in a laptop setting.

[Techhog]

Nvidia starts with server designs and then trickles the tech down. It's the same thing Intel does. Gamers don't get the best. They get "good enough" in the stack of products made, which is a test run of high-end stuff while they work out the kinks.

For mobile you're getting the reject pile: dies that don't have all cores functioning, but are cool enough to be sustained in a laptop setting.

 

This is completely wrong. Mobile parts are high-binned for lower voltage, and some do use full chips. (The Maxwell version of the 860M is a full GM107, and the 880M is a full GK104)

[patrickjp93]

This is completely wrong. Mobile parts are high-binned for lower voltage, and some do use full chips. (The Maxwell version of the 860M is a full GM107, and the 880M is a full GK104)

If you think the 107 and 104 aren't derivatives of a larger GK110 (Kepler vs. Maxwell I know, but hold on) you're crazy. Nvidia begins with a flagship chip design and then builds lower parts around it. The GK210 or w/e they decide to name it will be going into the 980TI, Quadro K7000, or their Maxwell Tesla or a combination of all 3. While the bugs are being worked out on that much larger design (and while TSMC is ironing out 20nm production), they design low core count chips mainly by slicing off cores and cache. It saves a lot of R&D time because they already need to design high-compute, low-power for the supercomputer/server world, and then sub-designs they get more breathing room to shave off expensive parts (like a 512-bit bus).

 

The binning for mobile parts may or may not be right, especially considering Nvidia dumped maxwell onto the mobile market with no desktop version (750TI aside).

[Techhog]

If you think the 107 and 104 aren't derivatives of a larger GK110 (Kepler vs. Maxwell I know, but hold on) you're crazy. Nvidia begins with a flagship chip design and then builds lower parts around it. The GK210 or w/e they decide to name it will be going into the 980TI, Quadro K7000, or their Maxwell Tesla or a combination of all 3. While the bugs are being worked out on that much larger design (and while TSMC is ironing out 20nm production), they design low core count chips mainly by slicing off cores and cache. It saves a lot of R&D time because they already need to design high-compute, low-power for the supercomputer/server world, and then sub-designs they get more breathing room to shave off expensive parts (like a 512-bit bus).

 

The binning for mobile parts may or may not be right, especially considering Nvidia dumped maxwell onto the mobile market with no desktop version (750TI aside).

This isn't what the other guy said really, but this sounds plausible. You don't need to say it in such a dickish way, though. 

[patrickjp93]

This isn't what the other guy said really, but this sounds plausible. You don't need to say it in such a dickish way, though. 

If my fatal flaw is being dickish I can live with myself. <--Please tell me everyone sees the joke in this.

[Misanthrope]

I'm looking at the eventual 960 to equal or surpass the current 770 but only if it's a 2014 release, if it's not I'd still settle for a 280x or a 290 or if the price is right a 970.