I just replaced my old AMD R9 380 with an 1080 from Nvidia. Result seems to be good in games I got double performance or maybe even more.
But then it comes to my software I have been written using opencl there is a large performance drop.
It seems to be the method clEnqueueNDRangeKernel() now is 200-300% slower for some unknown reason. I have installed the latest drivers from Nvidia and removed all AMD drivers but still same strange result.
The code is 100% the same between my tests.
The AMD R9 380 is 3.5 TFLOPS and Nvidia 1080 is 9 TFLOPS.
What Im a missing here?