Mars Saturn aspect in natal chart, transit and synastry - 未名空间MITBBS历史存档

国际科技财经博客移民网络热点娱乐民生时事公众号

Redian新闻

>未名空间

>astrology - 星座物语

Mars Saturn aspect in natal chart, transit and synastry

Mars Saturn aspect in natal chart, transit and synastry# astrology - 星座物语

f*r2014-09-01 07:09

1 楼

For example, suppose you have ~1 million text files, each containing ~1 mill
ion positive integers, one integer per line. How would you find the top 3 in
tegers that appear most frequently?
how many machines you will use, what algorithms to run on them, how they com
municate together, the running time and space complexity of your approach, e
tc.
有兴趣的同学们，大家看看如何solve吧:)

A*y2014-09-01 07:09

2 楼

i*g2014-09-01 07:09

3 楼

use HDFS as the storage, per the report from Yahoo, 10000 PC(linux), each Terabytes so, this is a minor case :) anyway use as much as HDisk to enhance the I/O performance and push the CPU, there could be a trade-off
using mapreduce to clac the frequency, map for the integer hashing(nothing to do), reduce from intermediate results of map shaping for the frequency, simply, just accumulate
Even, need pratice on different OS, CPU arch(generally X86 is OK)

A*y2014-09-01 07:09

4 楼

Done!

f*r2014-09-01 07:09

5 楼

非常感谢有如此深度的答复:)

Terabytes so, this is a minor case :) anyway use as much as HDisk to enhance
the I/O performance and push the CPU, there could be a trade-off
to do), reduce from intermediate results of map shaping for the frequency,
simply, just accumulate

【在 i*******g 的大作中提到】

: use HDFS as the storage, per the report from Yahoo, 10000 PC(linux), each Terabytes so, this is a minor case :) anyway use as much as HDisk to enhance the I/O performance and push the CPU, there could be a trade-off
: using mapreduce to clac the frequency, map for the integer hashing(nothing to do), reduce from intermediate results of map shaping for the frequency, simply, just accumulate
: Even, need pratice on different OS, CPU arch(generally X86 is OK)