Techniques for serialized execution in a simd processing system
AssigneeQUALCOMM Incorporated
InventorsAndrew Evan Gruber, Lin Chen, Yun Du, Alexei Vladimirovich Bourd
Life of this patent
Estimating the term…
Abstract
A SIMD processor may be configured to determine one or more active threads from a plurality of threads, select one active thread from the one or more active threads, and perform a divergent operation on the selected active thread. The divergent operation may be a serial operation.
Claims20
A method of performing divergent operations on a multi-threaded processor, the method comprising:
- determining one or more active threads from a plurality of threads;
- selecting one active thread from the one or more active threads;
- and
- performing a divergent operation on the selected active thread.
An apparatus configured to perform divergent operations, the apparatus comprising:
- a multi-threaded processor comprising:
- a control unit configured to determine one or more active threads from a plurality of threads, and to select one active thread from the one or more active threads;
- and
- a plurality of processing elements, wherein one processing element of the plurality of processing elements associated with the selected active thread is configured to perform a divergent operation.
An apparatus configured to perform divergent operations on a multi-threaded processor, the apparatus comprising:
- means for determining one or more active threads from a plurality of threads;
- means for selecting one active thread from the one or more active threads;
- and
- means for performing a divergent operation on the selected active thread.
A multi-threaded processor configured to perform divergent operations, the multi-threaded processor comprising:
- a control unit configured to determine one or more active threads from a plurality of threads, and to select one active thread from the one or more active threads;
- and
- a plurality of processing elements, wherein one processing element of the plurality of processing elements associated with the selected active thread is configured to perform a divergent operation.
Dependencies are inferred from the claim text.