Originally Posted by m3t4lh34d
M3T4LH34D ---- 3930k @ 4.4ghz ---- 4 x GTX Titan 1097/3005 ---- 178.9 FPS --- 4507
There definitely needs to be some driver optimizations with Unigine and Quad Titans in general, as the GPU usage was bouncing between 95% down to the 60s at times.
Going from 4000 to 4500 by adding a 4th Titan is unacceptable. I wont be happy until I'm pushing 6k with these cards once heavily OC'd.
I'll also try to jack my CPU up as high as possible since it's running on its own triple thick 360 Rad with Push/Pull Noctuas.
I might also swap back in my old set of 16GB of 2133mhz memory instead of using this 32gb set of 1600mhz memory.
Either way, Tsm, your #1 spot is about to be gone buddy. It was a nice run though...
Originally Posted by zGunBLADEz
Im sorry but i find very hilarious the titans struggling to get 7970 spot... Even if you do it will be a matter of just ridicule for nvidia part that a 4k gpu setup is struggling to beat 4x7970...
Just the thought of seen tsm rig around titans in different benchmarks even beating them i would think twice to spent that much money on titans...
Wow, there is some serious noob'ness in this thread over the last few pages. Running a 1080P benchmark at 3-4x SLI crossfire is.
Of course all the setups are going to top off around the same as the benchmark bounces off the CPU limit almost the entire run. The only way 3-4 cards gain any lead over 2-way setups is during the few low CPU demand scenes in which the FPS screams to 300+. Like in the Valley bench, TSM's 4x 7970's only bench 21%
faster than my two stock air-cooled Titan's. Does that mean his setup is really only 21% faster than two Titan's? Of course not, just bouncing of the CPU limit more than a 2-card setup. I fail to see how this is a difficult concept to understand.
If the demand (resolution) was increased on the setups to something more appropriate, the gap between the cards would increase. I think the time for a 1440P/1600P/Surround/Eyefinity benchmark with ultra settings and not these silly 1080P ones is upon us with these monster rigs.
Then of course there is something to be said about water cooled cards tweaked over a year on super over-volt runs with the winter windows open to squeek by a run that doesn't crash versus stock air cooled cards sandwiched together on new drivers.
EDIT: I'm going to see if Karlitos wants to do a 5760x1080 Surround/Eyefinity version of his Valley benchmark thread since his is run so well. I'd suspect most 3-4 GPU users could at least run that resolution and it would be much more useful and tell a more realistic story.
There shouldn't be a tone of entries either like the 1080P thread, so shouldn't be too much work.Edited by CallsignVega - 3/9/13 at 1:10pm