Will the C870 ever support cuda 1.1? My program is a pretty big program that compiles fine under 1.1 but when I downgrade to 1.0 to get the tesla to work I get:
1># --error 0xc0000005 –
I’ve heard that this error is due to the fact that cuda 1.0 isn’t as good at register allocation. If I upgrade to 1.1 it works fine.
How can I get this to compile under 1.0? Any tips or tricks?