Intelligent thread dispatch and vectorization of atomic operations

One or more non-transitory computer-readable storage mediums having stored thereon executable computer program instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:

• detecting a plurality of threads corresponding to a plurality of workloads associated with tasks relating to one or more processors including a graphics processor, the one or more processors including a plurality of compute blocks;
• scanning the plurality of workloads for spatial locality, and comparing detected spatial locality with the plurality of threads;
• based at least in part on the comparison, determining whether threads of the plurality of threads have adjacent surfaces in memory;
• and
• upon determining that a first set of threads of the plurality of threads have adjacent surfaces in memory, launching threads of the first set of threads using a first set of compute blocks of the plurality of compute blocks, compute blocks within the first set of compute blocks being adjacent to each other.

A method comprising:

• detecting a plurality of threads corresponding to a plurality of workloads associated with tasks relating to one or more processors including a graphics processor, the one or more processors including a plurality of compute blocks;
• scanning the plurality of workloads for spatial locality, and comparing detected spatial locality with the plurality of threads;
• based at least in part on the comparison, determining whether threads of the plurality of threads have adjacent surfaces in memory;
• and
• upon determining that a first set of threads of the plurality of threads have adjacent surfaces in memory, launching threads of the first set of threads using a first set of compute blocks of the plurality of compute blocks, compute blocks within the first set of compute blocks being adjacent to each other.