Graphics processor with non-blocking concurrent architecture
In some aspects, systems and methods provide for forming groupings of a plurality of independently-specified computation workloads, such as graphics processing workloads, and in a specific example, ray tracing workloads. The workloads include a scheduling key, which is one basis on which the groupings can be formed. Workloads grouped together can all execute from the same source of instructions, on one or more different private data elements. Such workloads can recursively instantiate other workloads that reference the same private data elements. In some examples, the scheduling key can be used to identify a data element to be used by all the workloads of a grouping. Memory conflicts to private data elements are handled through scheduling of non-conflicted workloads or specific instructions and/or deferring conflicted workloads instead of locking memory locations.
1. A computer-implemented method for processing workloads in a computer system comprising a plurality of computation elements, the method comprising:
receiving a plurality of fibres, each of the fibres comprising computer executable instructions, and each of the fibres having an associated priority;
identifying a scheduling key for each of the fibres;
grouping the fibres into one or more groups of fibres based on the identified scheduling key for each of the fibres;
determining a priority for each of the one or more groups of fibres based on the priority of the one or more fibres contained within that group; and
scheduling, for execution, the one or more groups of fibres based on the priority of each of the one or more groups of fibres.
2. The method according to claim 1 , wherein each of the fibres are instantiated individually either by a thread or another fibre.
3. The method according to claim 1 , further comprising determining a packet identifier representing one of the groups of fibres.
4. The method according to claim 1 , wherein the priority of each of the one or more groups of fibres is determined by determining an average of the individual priorities of the one or more fibres contained within that group.
5. The method according to claim 4 , wherein the determined average is a weighted average.
6. The method according to claim 1 , wherein the priority of each of the one or more groups of fibres is determined by assigning a priority to each of the groups of fibres based on the current execution state of the fibres within the computer system.
7. The method according to claim 1 , wherein the priority of each of the one or more groups of fibres is determined based on a storage location of the group of fibres.
8. The method according to claim 1 , wherein the computer executable instructions of a fibre are operable to:
read and write to a fibre storage memory, and
instantiate additional fibres.
9. The method according to claim 1 , wherein the computer executable instructions of a fibre are operable to read data from the fibre storage memory through a Non-Uniform Memory Access (NUMA) architecture.
10. The method according to claim 1 , wherein the fibres grouped together for execution are executed to perform ray tracing.
11. The method according to claim 1 , wherein the fibres grouped together for execution are executed to perform database traversal.
12. The method according to claim 1 , wherein the fibres grouped together for execution are executed to perform sorting.
13. The method according to claim 1 , wherein the fibres grouped together for execution are executed to perform spatial searching.
14. The method according to claim 1 , wherein the computer executable instructions of the fibres grouped together for execution are executed on a GPU.
15. A computing system configured to process workloads, wherein the computing system comprises:
a plurality of computation elements configured to execute fibres that comprise computer executable instructions; and
a controller configured to:
receive a plurality of fibres, each of the fibres having an associated priority;
identify a scheduling key for each of the fibres;
group the fibres into one or more groups of fibres based on the identified scheduling key for each of the fibres;
determine a priority for each of the one or more groups of fibres based on the priority of the one or more fibres contained within that group; and
schedule, for execution, the one or more groups of fibres based on the priority of each of the one or more groups of fibres.
16. The computing system according to claim 15 , further comprising a host processor which is configured to execute threads, and wherein the computer system further comprises a computing module that is a fibre API which is further configured to interface with the host processor to allow threads running on the host processor to instantiate a plurality of fibres for execution on one or more computation elements.
17. A non-transitory computer readable storage medium having stored thereon computer readable instructions, which when executed on at least one processor in a computer system comprising a plurality of computation elements, causes the computer system to:
receive a plurality of fibres, each of the fibres comprising computer executable instructions, and each of the fibres having an associated priority;
identify a scheduling key for each of the fibres;
group the fibres into one or more groups of fibres based on the identified scheduling key for each of the fibres;
determine a priority for each of the one or more groups of fibres based on the priority of the one or more fibres contained within that group; and
schedule, for execution, the one or more groups of fibres based on the priority of each of the one or more groups of fibres.