Reducing power consumption while performing collective operations on a plurality of compute nodes
View Patent ↗Methods, apparatus, and products are disclosed for reducing power consumption while performing collective operations on a plurality of compute nodes that include: receiving, by each compute node, instructions to perform a type of collective operation; selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations; and executing, by each compute node, the selected collective operation.
1. A method of reducing power consumption while performing collective operations on a plurality of compute nodes, the method comprising:
receiving, by each compute node, instructions to perform a type of collective operation;
selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations;
executing, by each compute node, the selected collective operation
determining, by each compute node, which hardware resources are not used during execution of the selected collective operation; and
reducing power, by each compute node during execution of the selected collective operation, of the hardware resources that are not used during execution of the selected collective operation.
2. The method of claim 1 wherein reducing power, by each compute node during execution of the selected collective operation, to the hardware resources that are not used during execution of the selected collective operation further comprises powering down the hardware resources that are not used during the selected collective operation.
3. The method of claim 1 wherein reducing power, by each compute node during execution of the selected collective operation, to the hardware resources that are not used during execution of the selected collective operation further comprises throttling operating speeds for the hardware resources that are not used during the selected collective operation.
4. The method of claim 1 further comprising:
determining, by each compute node, which hardware resources are used during execution of the selected collective operation; and
throttling, by each compute node during execution of the selected collective operation, operating speeds for the hardware resources that are in fact used during the selected collective operation.
5. The method of claim 1 wherein:
the method further comprises profiling the plurality of collective operations for power consumption characteristics; and
selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations further comprises selecting the particular collective operation in dependence upon the power consumption characteristics for the plurality of collective operations and current collective operation performance requirements.
6. The method of claim 1 wherein the plurality of compute nodes are connected together for data communications using a plurality of data communications networks, at least one of the data communications networks optimized for point to point operations, and at least one of the data communications optimized for collective operations.
7. A parallel computer capable of reducing power consumption while performing collective operations on a plurality of compute nodes, the parallel computer comprising the plurality of compute nodes, each compute node comprising one or more computer processors and computer memory operatively coupled to the computer processors, the computer memory having disposed within it computer program instructions capable of:
receiving, by each compute node, instructions to perform a type of collective operation;
selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations; executing, by each compute node, the selected collective operation
determining, by each compute node, which hardware resources are not used during execution of the selected collective operation; and
reducing power, by each compute node during execution of the selected collective operation, of the hardware resources that are not used during execution of the selected collective operation.
8. The parallel computer of claim 7 wherein the computer memory has disposed within it computer program instructions capable of:
determining, by each compute node, which hardware resources are used during execution of the selected collective operation; and
throttling, by each compute node during execution of the selected collective operation, operating speeds for the hardware resources that are in fact used during the selected collective operation.
9. The parallel computer of claim 7 wherein:
the computer memory has disposed within it computer program instructions capable of profiling the plurality of collective operations for power consumption characteristics; and
selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations further comprises selecting the particular collective operation in dependence upon the power consumption characteristics for the plurality of collective operations and current collective operation performance requirements.
10. The parallel computer of claim 7 wherein the plurality of compute nodes are connected together for data communications using a plurality of data communications networks, at least one of the data communications networks optimized for point to point operations, and at least one of the data communications optimized for collective operations.
11. A computer program product for reducing power consumption while performing collective operations on a plurality of compute nodes, the computer program product disposed upon a computer readable recordable medium, the computer program product comprising computer program instructions capable of:
receiving, by each compute node, instructions to perform a type of collective operation;
selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations;
executing, by each compute node, the selected collective operation
determining, by each compute node, which hardware resources are not used during execution of the selected collective operation; and
reducing power consumption, by each compute node during execution of the selected collective operation, of the hardware resources that are not used during execution of the selected collective operation.
12. The computer program product of claim 11 wherein reducing power, by each compute node during execution of the selected collective operation, to the hardware resources that are not used during execution of the selected collective operation further comprises powering down the hardware resources that are not used during the selected collective operation.
13. The computer program product of claim 11 wherein reducing power, by each compute node during execution of the selected collective operation, to the hardware resources that are not used during execution of the selected collective operation further comprises throttling operating speeds for the hardware resources that are not used during the selected collective operation.
14. The computer program product of claim 11 further comprising computer program instructions capable of:
determining, by each compute node, which hardware resources are used during execution of the selected collective operation; and
throttling, by each compute node during execution of the selected collective operation, operating speeds for the hardware resources that are in fact used during the selected collective operation.
15. The computer program product of claim 11 wherein:
the computer program product further comprises computer program instructions capable of profiling the plurality of collective operations for power consumption characteristics; and
selecting, by each compute node from a plurality of collective operations for the collective operation type, a particular collective operation in dependence upon power consumption characteristics for each of the plurality of collective operations further comprises selecting the particular collective operation in dependence upon the power consumption characteristics for the plurality of collective operations and current collective operation performance requirements.