Joining two 32 bit variables into a 64 bit one in inline CUDA ptx
int lo, hi;
unsigned long long x;
asm volatile("mov.b32 {%0, %1}, %2;" : "=r"(lo), "=r"(hi): "l"(x));
Can anyone pointed out why NVCC report the following error with this
simple inline ptx code?
error : Arguments mismatch for instruction 'mov' ptxas : fatal error :
Ptx assembly aborted due to errors
No comments:
Post a Comment