[R] Questions about 'bigmemory'

[Zhenfei Yuan]

Dear R users,

I'm now dealing with some big data set as big as 10GB, and my RAM got
16GB. Every time I run some models to this dataset using R, I have to
first load it into RAM via read.csv, which spends lots time.

I find the bigmemory package in the high performance task view of R, and
tried this package, found that it is really cool, especially for access
big matrix in shared memory through different R sessions.

My Question is that, when I call read.big.matrix on the .csv file (say
it aa.csv) the first time in R session 'R1', I would got one binary file
aa.bin and descriptor file aa.desc under specified backing path. I tried
to access the so called shared memory  via 'attach.big.matrix' in a
different R session 'R2', it works as I thought; however, when I quit
'R1', I could also access to the big matrix object via
attach.big.matrix('/path/to/aa.desc'), just like 'R1' still exists. I
copied the files 'aa.bin' and 'aa.desc' to another computer,
'attach.big.matrix' also worked on that one.


So, I don't know where the 'Shared Memory' is after turning 'R1' down;
is there differnece between existance and non-existance of 'R1'?

Best regards to you all,
Zhenfei