Implementation and preliminary evaluation of collective communication using direct GPU-to-GPU communication with the Tightly Coupled Accelerators (TCA) architecture
Implementation and preliminary evaluation of collective communication using direct GPU-to-GPU communication with the Tightly Coupled Accelerators (TCA) architecture
松本 和也, 塙 敏博, 児玉 祐悦, 藤井 久史, 朴 泰祐, "Implementation and preliminary evaluation of collective communication using direct GPU-to-GPU communication with the Tightly Coupled Accelerators (TCA) architecture", IPSJ SIG Technical Report, Vol.2014-HPC-147 No.23, 2014. (in Japanese)
IPSJ SIG Technical Report
BiBTeX entry
@inproceedings{hpcs2014-3133,
title = {密結合並列演算加速機構 TCA を用いた GPU 間直接通信による Collective 通信の実装と予備評価},
booktitle = {情報処理学会研究報告},
vol = {2014-HPC-147 No.23},
year = {2014},
}