Originally Posted by Zenitur
I found some slow of CUDA. I'm using pyoclbm (sha256 hashes) and nVidia 9800GT with overclock: 700/1070. This is 31600 with 275.09.07 and 31200 with 280.04.
I observed significant slow of all my CUDA applications too (Amber, ACEMD, NAMD, PetaChem and so on). In fact I observe progressive performance degradation since the first 270.xx driver and each new generation of drives are just worse...for what I should to perform any upgrade...? Only minor bug fixes....I really can't understand this politic especially for the high-end cards. How it is possible one that wish to spend money for 10 GTX590 cards then to wait for a bug fix more than 4 months, just some one to explain me that....no comments....