memory allocation

Abstract

Describes memory allocation methods. The first example method maps registers referenced by different instance groups of the same task to separate logical memory. Other example methods described herein describe mapping registers referenced by tasks to different banks within a single logical memory, and in various examples, this mapping can take into account which bank is likely to be the primary bank for a particular task and which one or Allocation of multiple other tasks.

Description

technical field Background technique [0001] In a processing system, when a task is created, a portion of the memory is allocated to the task. Next, the address generation unit maps the registers referenced within the task to actual memory addresses within the allocated memory portion. Two tasks can be assigned a memory address within the same memory. Conflicts can occur when multiple access requests are made to memory at the same time. For example, two tasks can each request a value from memory, or a single task can request two values ​​from memory. This causes one access to have to stop until another access is complete. [0002] To increase read / write throughput (by reducing the occurrence of stalls), memory can be arranged into separate groups, and in any cycle, data can be read from each group. In this way, the processing system can simultaneously access different banks within the same memory (for example, read register values ​​from row 0 in bank 0 of memory and read...

AI Technical Summary

Problems solved by technology

In this way, the processing system can simultaneously access different banks within the same memory (for example, read a register value from row 0 in bank 0 of memory, and read a register value from row 2 in bank 1 of memory), but Whenever simultaneous attempts are made to access the same group, a conflict occurs and one of the accesses must stop

This affects the performance of the processing system

Images