Broadcast asynchronous loads to shared local memory

A method comprising:

• receiving, at cluster of graphics cores, designation of a graphics core as a producer graphics core;
• initiating a memory load to from a surface in memory via a direct memory access engine of the producer graphics core;
• loading data from the surface into a memory of the producer graphics core via a cache memory associated with the cluster of graphics cores;
• and
• transmitting loaded data from the memory of the producer graphics core to consumer graphics cores within the cluster of graphics cores.

A data processing system comprising:

• a memory device;
• and
• a graphics processor coupled with the memory device, the graphics processor including a cache memory and a graphics core cluster coupled with the cache memory, the graphics core cluster including circuitry configured to:
• receive designation of a first producer graphics core and first consumer graphics cores for a first multicast load;
• receive designation of second producer graphics core and second consumer graphics cores for a second multicast load;
• load data from the memory device into a shared local memory of the first producer graphics core via a cache memory associated with the graphics core cluster;
• load data from the memory device into a shared local memory of the second producer graphics core via the cache memory associated with the graphics core cluster;
• and
• multicast loaded data via the first producer graphics core and the second producer graphics core to the first consumer graphics cores and the second consumer graphics cores.