GTX 680 PPBM5 Bench Test

[mikecLA]

First, I've never done this and have no experience in benching. The 680 was the final component I was waiting for to build my LGA 2011 box, which I finally got together today. I was anxious to see if the cuda card hack would work and how much difference Mercury playback would make on this card. I did notice that rendering the H.264 test timeline in CS5 was too quick to measure by eye with MPB on and real slow with it off.

It seems much faster than my GTX 470, but then again, that card is in my Q6600 box.

Anyway, the #s from the output.txt file with MPB On:
"-1237","-1226","-1357","5"

With it off, the last # changes to 69, rest are the same.

I have no idea why the first three are negative and how this would compare to a 580 (since I don't have one to test). Perhaps someone with a 580 can compare?

Running on a 3930K, 32 gigs ram, stock clock for now. Will play with OC'ng the box later.

 

[Gazza238]

Surfing the net for details on the 680
came upon your post

Not sure what is causing the problem.
The 5 at the end of the output file would be correct. 5 seconds
however a work around for the encode times would be to
run the test and once completed
(a) place mouse over the tick on each item and this will give
you time
(b) click on the tick and it will bring up AME Encoding log
check the end of report as it will show all AME tasks

note the timings for your posted output file should be in the log

If you tally all three times plus the 5 it will give you and approximate
score +/- 2 second (i found slight difference between AME times and Output file)