for my Tesla K80

-b 13 -t 256 - p 1200  (basic settings)  ==> 55MKey/s
-b 26 -t 256 - p 1200 (double SM) ==> 99Mkey/s
-b 39 -t 256 - p 1200 (triple SM) ==> 108Mkey/s
-b 52 -t 256 - p 1200 (quad SM) ==> 117Mkey/s

-b 13 -t 512 - p 1000  (basic settings) ==> 99MKey/s
-b 26 -t 512 - p 1000 (double SM) ==> 125Mkey/s
-b 39 -t 512 - p 1000 (triple SM) ===> alloc error and get 110Mkey
-b 52 -t 512 - p 1000 (quad SM) == alloc error and get 110Mkey

Here are my results, thanks for you support