Pragma that maps a variable to the constant memory on GPU
coder.gpu.constantMemory(
maps the
variable v
)v
to the constant memory space on the GPU device. Place
this pragma within a parallelizable loop. If GPU Coder™ generates a kernel for the loop, it loads v
to a
device constant memory variable. It replaces any access to this variable within the
kernel by access to the constant memory variable. Within the kernel, the variable
v
must be read-only. Otherwise, GPU Coder ignores this pragma. Use this pragma when every thread accesses every
element of the parameter array or matrix.
This function is a code generation function. It has no effect in MATLAB®.
codegen
| coder.gpu.kernel
| coder.gpu.kernelfun
| coder.gpu.nokernel
| coder.gpu.persistentMemory
| gpucoder.stencilKernel