Is it possible to have two kernels, each with it's own output scatter stream?
For example:
kernel void foo1(float4 a[], out float4 b[])
{
....
}
kernel void foo2(float4 c[], out float4 d[])
{
....
}
I couldn't get this to work, even if I put these two kernels in separate .br files.
Is this possible? Is there an example that does this?
In this same regard, is it possible to have more than 1 output stream per kernel? I thought I read this was possible, but I keep getting an error when I try this.
Also, for the above, I get an "Invalid Buffer" error. Am I overrunning the stack?