There is something here that I don’t understand. It is said in the guide (and seems to be true when compiling) that atomic operations are only available for 1.1 architecture -arch sm_11 and not 1.0 with -arch sm_10…but the 8600 (1.1) is in middle range of G80 gpus and cheeper than 8800 (1.0) ones.
I don’t understand … I have a 8800 as you have understood and need to use atomic functions.
What is the reason ? is it because 8600 GPUs have been released later than 8800 ones ??