actually its not the firmware its the drivers that tell which card what to do becuase after watching mantle thing from AMD its more clear. What happens is the driver gets commands from the game and it tries to guess what the best order, queue which is best at processing what etc… with mantle the game/application basically becomes the driver it pretty much directly tells the cards what to process what data goes in which memory buffer, the order of the queues, just anything basically the driver has enough abstraction to make some stuff easier but instead of driver doing best guess, the game/application is telling it what to do for the most efficeient way possible.

Now mantle is still very much beta, its got alot to do to become a really good API but MS obviously dont like it cus DX12 announcement the thing with that is though how can they provide such low level access which is literally by architecture cus most are different enough to do a good job. itll take MS years

Dual gpu workload should in theory be just as fast because each card is supposed to render only 50% of the image which would more than make up for travel time of the data. Unfortunately the real world performance just isn’t there.

hmm i dunno its prolly more to do with it doesnt have to leave the card, in dual GPU like 2 physical card it has to leave the card enter the bus then enter the other card and vice versa so it prolly is quicker a little bit maybe not sure how they work tbh but id imagine data being pumped through 1 card with 2 GPU’s is faster cus of smaller nm distance so latency reduced fps increases butwith any dual gpu setup CPU the bottleneck and mantle is the cure, and maxwells denver chip to some extent

Dual GPU cards seem to work better than 2 separate cards. Its probably because he division of work between the chips is handled by the firmware itself and not a game.

ah i mean dual cards for 4K resolution for 1080p maxed 780-780Ti is enough, for 1440p might be a bit dodgy maybe dual GPU there but think dual 770’s will be enough for 1440p

Star citizen is why I’m doing it. My hangar runs at 19fps at sub HD resolution with detail settings on low.

Don’t ever bother with dual cards. A lot of games don’t take advantage of it and the ones that do aren’t worth it. You see a gain of like 5-10 fps on average. The only real benefit is using 1 card just for physics but again the game has to support doing that. Not worth double the price at all.

would you line a refresher to why thr PC will be king again?

http://www.dualshockers.com/2014/03/07/fantastic-star-citizen-video-will-show-you-a-different-kind-of-next-generation/

be more if you goto http://robertsspaceindustries.com if anygame warrants a dual GPU setup its this

ive never ought dual cards i ahd a athlon 64 3200, 4GB ram and hmm notsure the only i really replaced was the gfx card had it for 6 years. I upgraded my PC 3 years agao to i7 950, 6GB triple channel, 1GB 6870 xifi sound card, 5.1 speaker 24ich 1080p monitor. Ive recently replaced it with a 3GB 280x Toxic which failed 3 times in a row so got a MSI 280x gaming edition which is more than upto the job can throw anything at it at max details and performs extremely well. But for Styar Citizen when it comes out ill be looking upto £600 gfx card and a new PSU to make sure i max it out cus its going to be awesome

I used to lol. I’d buy 2 new graphics cards at least once a year. Now I’m a little more tempered in my spending. I bought a gaming class laptop about 2 Years ago. It just barely runs stuff coming out now at medium settings so I stopped buying games for it unless its a great steam sale or a humble bundle. I’ll build or buy a new one probably next year.