20240054099. Cluster-Based Placement and Routing of Memory Units and Compute Units in a Reconfigurable Computing Grid simplified abstract (SambaNova Systems, Inc.)
Cluster-Based Placement and Routing of Memory Units and Compute Units in a Reconfigurable Computing Grid
Organization Name
Inventor(s)
Kin Hing Leung of Cupertino CA (US)
Feng Sheng of Palo Alto CA (US)
Ajit Punj of Palo Alto CA (US)
Cluster-Based Placement and Routing of Memory Units and Compute Units in a Reconfigurable Computing Grid - A simplified explanation of the abstract
This abstract first appeared for US patent application 20240054099 titled 'Cluster-Based Placement and Routing of Memory Units and Compute Units in a Reconfigurable Computing Grid
Simplified Explanation
The abstract describes a method for placing, routing, and using compute units and memory units in a reconfigurable computing grid.
- Receiving a placement graph for a computing task with unplaced memory units, compute units, and data connections.
- Forming subgraphs for each unplaced memory unit with a primary connection.
- Placing each formed subgraph as a cluster on the reconfigurable computing grid.
- Potential Applications:**
- Reconfigurable computing grids for high-performance computing tasks.
- Optimizing resource allocation in complex computing systems.
- Problems Solved:**
- Efficient placement and routing of compute units and memory units in a computing grid.
- Streamlining the process of configuring computing resources for tasks.
- Benefits:**
- Improved performance and resource utilization in computing tasks.
- Simplified management of compute units and memory units in a reconfigurable grid environment.
Original Abstract Submitted
a method for placing, routing and using compute units and memory units in a reconfigurable computing grid includes receiving a placement graph for a computing task that defines a set of unplaced memory units, a set of unplaced compute units and data connections between the unplaced memory units and the unplaced compute units, the data connections comprising primary connections corresponding to the primary ports of the unplaced compute units and secondary connections corresponding to the secondary ports of the unplaced compute units. the method also includes forming a subgraph for each unplaced memory unit having a primary connection, each subgraph comprising the unplaced memory unit and each unplaced compute unit connected to the unplaced memory unit via a primary connection. the method also includes placing each formed subgraph as a cluster on the reconfigurable computing grid. a corresponding computer program product and system are also disclosed herein.
