Date: Wed, 29 Dec 2010 15:49:44 +0900
--------------------------------------------------
[san-tech][02873] NVIDIA GPUDirect等々 (+Storage と GPU)
ちょっと気になったので、Erasure Codesと GPUで調べてみました:
前回紹介したと思いますが
"A GPU Accelerated Storage System,"
Abdullah Gharaibeh, Samer Al-Kiswany, Sathish Gopalakrishnan,
Matei Ripeanu, The University of British Columbia, CANADA
IEEE/ACM International Symposium on High Performance Distributed
Computing (HPDC 2010)
http://www.ece.ubc.ca/~abdullah/papers/gharaibeh-hpdc10.pdf
http://www.ece.ubc.ca/~abdullah/talks/gharaibeh-hpdc10.ppt
Paper (Page 3)から
"Similarly, erasure codes' encoding throughput is limited by their
computational complexity" ....
We aim to understand the viability of offloading these
data-processing intensive operations to a GPU to dramatically reduce
the load on the source CPU and enhance overall system performance.
In this context, we focus on the use of hashing-based primitives in
storage systems. This section highlights their computational overheads
(2.1), and presents a brief overview of NVIDIA GPU’s architecture
(2.2)"
StoreGPU: Exploiting Graphics Processing Units to Accelerate Distributed
Storage Systems (上記プロジェクト)
http://www.ece.ubc.ca/~samera/projects/StoreGPU/index.htm
StoreGPU
http://sourceforge.net/projects/storegpu/
"StoreGPU is a library that accelerates hashing functions (MD5 and
SHA1) by offloading the hashing computation to the GPU. StoreGPU
provides two hashing modes: standard hashing mode, and sliding window
hashing mode."
Version 1.0, 2010-05-30
Recent Publications
http://netsyslab.ece.ubc.ca/wiki/index.php/Publications
前回紹介した東工大 TUBAME 2.0でも
> "TSUBAME 2.0 Supercomputer"
> Satoshi Matsuoka, Titech, November 17, 2010
> http://www.nvidia.com/content/PDF/sc_2010/theater/Matsuoka_SC10.pdf
"Hybrid Diskless Checkpoint" (71枚目から) で少し触れてます。
関連発表
Scalable encoding algorithm and efficient group mapping
"Distributed Diskless Checkpoint for Large Scale Systems"
Leonardo Arturo Bautista Gomez, ..... , Satoshi Matsuoka
IEEE/ACM International Conference on Cluster, Cloud and Grid Computing,
2010 (CCGrid10)
http://www.computer.org/portal/web/csdl/doi/10.1109/CCGRID.2010.40
Fast encoding work on idle resource (CPU/GPU)
"Low-overhead diskless checkpoint for hybrid computing systems"
Leonardo Bautista, ..... , Satoshi Matsuoka
International Conference on High Performance Computing (HiPC 2010)
※オンライン化はまだです。ただし、Lecture Notes in Computer Science
後述しますが、同じメンバーでの講演資料が見つかりました。
Software Framework for GPGPU Memory FT
"A High-Performance Fault-Tolerant Software Framework for Memory on
Commodity GPUs"
Naoya Maruyama, Akira Nukada and Satoshi Matsuoka
IEEE International Parallel and Distributed Processing Symposium
(IPDPS'10)
http://matsu-www.is.titech.ac.jp/~naoya/publications/ipdps10.pdf
http://matsu-www.is.titech.ac.jp/~naoya/publications/ipdps10-talk.pdf
"Transparent low-overhead checkpoint for GPU-accelerated clusters"
Leonardo Bautista, ..... , Satoshi Matsuoka
Forth workshop of the Joint Laboratory for Petascale Computing
2010 (NCSA & INRIA)
https://wiki.ncsa.illinois.edu/download/attachments/17630761/INRIA-UIUC-WS4-lbautista.pdf
https://wiki.ncsa.illinois.edu/display/jointlab/Workshop+Program#WorkshopProgram-GomezA
"We propose a transparent low-overhead checkpointing technique for
GPU accelerated clusters that avoid the I/O bottleneck by using
erasure codes and SSDs on the compute nodes."
※講演資料とアブストラクトだけでペーパーは公開されていません。
Joint Laboratory for Petascale Computing
http://jointlab.ncsa.illinois.edu/
The fourth workshop of the Joint Laboratory for Petascale Computing
November 22-24, 2010
http://jointlab.ncsa.illinois.edu/events/workshop4/
Workshop Program
https://wiki.ncsa.illinois.edu/display/jointlab/Workshop+Program
※講演資料とアブストラクトが公開されています。
ちょっと関連:
"What Comes After RAID? Erasure Codes", December 16, 2010
http://www.networkcomputing.com/deduplication/what-comes-after-raid---erasure-codes.php
http://www.networkcomputing.com/deduplication/what-comes-after-raid---erasure-codes.php?p=2
"Erasure codes get really interesting, however, when we up the ante
beyond n+2 as several vendors have. NEC's HydraStor deduplicating
grid system uses erasure codes to spread each data chunk across
twelve disk drives in the grid."
※Cleversafeについても言及してます (これは面白い分散ストレージです)。
前回コメントを忘れましたが
> "Processing Petabytes per Second with the ATLAS experiment at the
> Large Hadron Collider at CERN"
> Philip Clark, Andrew Washbrook - University of Edinburgh
> http://www.nvidia.com/content/GTC-2010/pdfs/2135_GTC2010.pdf
これは、CERN加速器系のデータ処理量のすごさが解ります。
今回はストレージ系にはほとんど言及してませんが。
0 件のコメント:
コメントを投稿