HPL 相關

How to calculate peak theoretical performance of a CPU-based HPC system

FLOPS per cycle for sandy-bridge and haswell SSE2/AVX/AVX2

System Performance and Ranking Estimation

http://hpl-calculator.sourceforge.net/HPL-HowTo.pdf

Intel Xeon serial info

Xeon 7000 multi procesor configuration on MB
Xeon 5000 dual processoer configutation on MB
Xeon 3000 single processor

Letter Prefix:
X = Performance
E = Mainstream (rack mount)
L = Power Optimized

HPCC notes

修改CCFLAGS，如果用mpich，是$(HPL_DEFS) -fomit-frame-pointer -03-funroll-loops -W –Wall，

用openmpi就要修改成 $(HPL_DEFS) -03


HPL 產生器
http://www.advancedclustering.com/faq/how-do-i-tune-my-hpldat-file.html


HPCC inf 說明
http://onetfbao.spaces.live.com/blog/cns!393E9866CAB6E396!367.entry 
http://www.orientsky.com.cn/_new/Printpage.asp?ID=1496
http://onemansjourneyintolinux.blogspot.com/2009/01/hpl-tuning.html

如何做Linpack测试及性能优化
曹振南
http://www.netlib.org/benchmark/hpl/faqs.html

reference: Tuning considerations for HPC Challenge

http://www.ibm.com/developerworks/wikis/display/LinuxP/Tuning+considerations+for+HPC+Challenge

What is MPICH

What is MPICH

MPI(Message Passing Interface) 為一種訊息傳遞標準, 提供一個可靠的傳遞介面, 讓各個node 作資料交換, 主要應用在異質網路叢集電腦系統上。 MPICH 是MPI-1 標準的實作, 包含了MPI 函式庫及編譯環境。

編譯及安裝

目前最新版本為 1.2.7p1, 可以在下方的網頁上找到
http://www-unix.mcs.anl.gov/mpi/mpich1/

作業系統以Linux 為主, MPICH 也有Windows 版本可以在下載網頁上找到。而compiler 以Linux 環境下常見的gcc 為主。

tar zxf mpich.tar.gz
cd mpich-1.2.7p1
./configure --prefix=/opt/mpich/gnu -rsh=ssh
make && make install

Compile MPI program
/opt/mpich/gnu/bin/mpicc -o hello hello.c

Google: Cluster Computing and MapReduce Lecture

從Google codes 上看到, 是講有關Cluster Computing

Lecture 1 - Introduction


slides

Lecture 2 - MapReduce
slides

Lecture 3 - Distributed File Systems
slides

Lecture 4 - Clustering Algorithms
slides

Lecture 5 - Graph Algorithms
slides

原文網址
http://code.google.com/edu/content/submissions/mapreduce-minilecture/listing.html