TY - JOUR
T1 - Dissecting GPU Memory Hierarchy Through Microbenchmarking
AU - Mei, Xinxin
AU - CHU, Xiaowen
N1 - The authors would like to thank the anonymous reviewers for their valuable comments. This work is partially supported by Hong Kong GRF grant HKBU 210412, HKBU FRG2/14-15/059, Shenzhen Basic Research Grant SCI-2015-SZTIC-002.
PY - 2017/1/1
Y1 - 2017/1/1
N2 - Memory access efficiency is a key factor in fully utilizing the computational power of graphics processing units (GPUs). However, many details of the GPU memory hierarchy are not released by GPU vendors. In this paper, we propose a novel fine-grained microbenchmarking approach and apply it to three generations of NVIDIA GPUs, namely Fermi, Kepler, and Maxwell, to expose the previously unknown characteristics of their memory hierarchies. Specifically, we investigate the structures of different GPU cache systems, such as the data cache, the texture cache and the translation look-aside buffer (TLB). We also investigate the throughput and access latency of GPU global memory and shared memory. Our microbenchmark results offer a better understanding of the mysterious GPU memory hierarchy, which will facilitate the software optimization and modelling of GPU architectures. To the best of our knowledge, this is the first study to reveal the cache properties of Kepler and Maxwell GPUs, and the superiority of Maxwell in shared memory performance under bank conflict.
AB - Memory access efficiency is a key factor in fully utilizing the computational power of graphics processing units (GPUs). However, many details of the GPU memory hierarchy are not released by GPU vendors. In this paper, we propose a novel fine-grained microbenchmarking approach and apply it to three generations of NVIDIA GPUs, namely Fermi, Kepler, and Maxwell, to expose the previously unknown characteristics of their memory hierarchies. Specifically, we investigate the structures of different GPU cache systems, such as the data cache, the texture cache and the translation look-aside buffer (TLB). We also investigate the throughput and access latency of GPU global memory and shared memory. Our microbenchmark results offer a better understanding of the mysterious GPU memory hierarchy, which will facilitate the software optimization and modelling of GPU architectures. To the best of our knowledge, this is the first study to reveal the cache properties of Kepler and Maxwell GPUs, and the superiority of Maxwell in shared memory performance under bank conflict.
KW - cache structure
KW - CUDA
KW - GPU
KW - memory hierarchy
KW - throughput
UR - http://www.scopus.com/inward/record.url?scp=85006829856&partnerID=8YFLogxK
U2 - 10.1109/TPDS.2016.2549523
DO - 10.1109/TPDS.2016.2549523
M3 - Journal article
AN - SCOPUS:85006829856
SN - 1045-9219
VL - 28
SP - 72
EP - 86
JO - IEEE Transactions on Parallel and Distributed Systems
JF - IEEE Transactions on Parallel and Distributed Systems
IS - 1
M1 - 7445236
ER -