Intel® FPGA SDK for OpenCL™ Standard Edition: Best Practices Guide

ID 683176
Date 9/24/2018
Public
Document Table of Contents

7.6. Minimizing the Memory Dependencies for Loop Pipelining

ensures that the memory accesses from the same thread respects the program order. When you compile an NDRange kernel, use barriers to synchronize memory accesses across threads in the same work-group.

Loop dependencies might introduce bottlenecks for single work-item kernels due to latency associated with the memory accesses. The offline compiler defers a memory operation until a dependent memory operation completes. This could impact the loop initiation interval (II). The offline compiler indicates the memory dependencies in the optimization report.

To minimize the impact of memory dependencies for loop pipelining:
  • Ensure that the offline compiler does not assume false dependencies.
    When the static memory dependence analysis fails to prove that dependency does not exist, the offline compiler assumes that a dependency exists and modifies the kernel execution to enforce the dependency. Impact of the dependency enforcement is lower if the memory system is stall-free.
    • Write after read operations with data dependency on a load-store unit can take just two clock cycles (II=2). Other stall-free scenarios can take up to seven clock cycles.
    • Read after write (control dependency) operation can be fully resolved by the offline compiler.
  • Override the static memory dependence analysis by adding the line #pragma ivdep before the loop in your kernel code if you are sure that it carries no dependences.

Did you find the information on this page useful?

Characters remaining:

Feedback Message