OAWS: Memory Occlusion Aware Warp Scheduling

@article{Wang2016OAWSMO,
  title={OAWS: Memory Occlusion Aware Warp Scheduling},
  author={Bin Wang and Yue Zhu and Weikuan Yu},
  journal={2016 International Conference on Parallel Architecture and Compilation Techniques (PACT)},
  year={2016},
  pages={45-55}
}
We have closely examined GPU resource utilization when executing memory-intensive benchmarks. Our detailed analysis of GPU global memory accesses reveals that divergent loads can lead to the occlusion of Load-Store units, resulting in quick consumption of MSHR entries. Such memory occlusion prevents other ready memory instructions from accessing L1 data cache, eventually stalling warp schedulers and degrading the overall performance. We have designed memory Occlusion Aware Warp Scheduling (OAWS… CONTINUE READING

Citations

Publications citing this paper.

References

Publications referenced by this paper.
Showing 1-10 of 11 references

Similar Papers

Loading similar papers…