Memory-aware TLP throttling and cache bypassing for GPUs

Memory-aware TLP throttling and cache bypassing for GPUs
复制标题

GPU 的内存感知 TLP 节流和缓存绕过

DOI:
10.1007/s10586-017-1396-0
复制
发表时间:
2017-11
期刊:
Cluster Computing
影响因子:
--
通讯作者:
Hai Tan
Hai Tan
中科院分区:
其他
文献类型:
--
作者:
Jun Zhang;Yanxiang He;Fanfan Shen;Qing'an Li;Hai Tan

文献摘要

参考文献

相似文献

通用图形处理器(GPGPU)已成为面向高吞吐量应用的重要高性能平台之一。然而,由于GPGPU内部有大量并发运行的线程,因此经常会发生片上资源争用。它已成为影响GPGPU性能的重要因素。我们提出了内存感知的TLP节流和缓存旁路(MATB)机制,该机制可以利用数据的时态局部性和内存带宽。它的目的是使数据局部性好的缓存块在L1D缓存中停留更长时间,同时保持片上资源的利用率。一方面,在发生缓存竞争和片上网络拥塞时,通过限制具有不良数据重用的内存偏差来调度缓存竞争,从而缓解缓存竞争。另一方面,它可以通过缓存旁路使内存带宽得到更有效的利用。实验结果表明,与GTO和DYNCTA相比,MATB在硬件成本较低的情况下,性能平均分别提高了26.6%和14.2%。
General-purpose graphics processing unit (GPGPU) has become one of the most important high performance platforms oriented to high throughput applications. However, on-chip resources contention can often occur as there are large amounts of concurrent running threads inside GPGPU. It has become an important factor affecting the performance of GPGPUs. We propose memory-aware TLP throttling and cache bypassing (MATB) mechanism, which can exploit data temporal locality and memory bandwidth. It aims to make those cache blocks with good data locality stay inside L1D cache longer while maintaining on-chip resources utiliza- tion. On one hand, it can alleviate cache contention via limiting the memory warps with bad data reuse to be scheduled while cache contention and on-chip network congestion occur. On the other hand, it can make memory bandwidth be utilized more effectively via cache bypassing. Experimental results show MATB can achieve 26.6% and 14.2% performance improvement respectively on average relative to GTO and DYNCTA with low hardware cost.
DOI: 10.1109/mc.2007.59
发表时间: 2007-02
期刊: Computer
影响因子: 2.2
作者:
D. Luebke;G. Humphreys
通讯作者: D. Luebke;G. Humphreys
DOI: 10.1109/mm.2010.41
发表时间: 2010-03-01
期刊: IEEE MICRO
影响因子: 3.6
作者:
Nickolls, John;Dally, William J.
通讯作者: Dally, William J.
DOI: 10.1145/3037697.3037742
发表时间: 2017-04
期刊: Proceedings of the Twenty-Second International Conference on Architectural Support for Programming Languages and Operating Systems
影响因子: --
作者:
Bo Wu;Xu Liu;Xiaobo Zhou;Changjun Jiang
通讯作者: Bo Wu;Xu Liu;Xiaobo Zhou;Changjun Jiang
DOI: 10.1145/2678373.2665702
发表时间: 2014-10
期刊: 2014 ACM/IEEE 41st International Symposium on Computer Architecture (ISCA)
影响因子: --
作者:
Ivan Tanasić;Isaac Gelado;Javier Cabezas;Alex Ramírez;Nacho Navarro;M. Valero
通讯作者: Ivan Tanasić;Isaac Gelado;Javier Cabezas;Alex Ramírez;Nacho Navarro;M. Valero
DOI: 10.1109/tpds.2013.257
发表时间: 2013-03
影响因子: 5.3
作者:
Jianlong Zhong;Bingsheng He
通讯作者: Jianlong Zhong;Bingsheng He