HIGHLIGHTS
SUMMARY
One can capture the class by value in lambda expressions, making the this-> reference valid inside parallel_for kernels (meaning you can you class methods and data inside that parallel_for call). Member functions called by parallel_for kernels still cannot use class member data or functions because those will still invoke this-> references that are only valid in host memory. The order of parallel_for calls on the device is respected, meaning that a subsequent call to parallel_for call will not start until all previous device work has been completed. YAKL_SCOPE and Using Non-Local Data Inside . . .
If you want to have access to all the content you need to log in!
Thanks :)
If you don't have an account, you can create one here.