[Rd] Manipulating single-precision (float) arrays in .Callfunctions
mdowle at mdowle.plus.com
Tue Jul 19 16:53:46 CEST 2011
"Duncan Murdoch" <murdoch.duncan at gmail.com> wrote in message
news:4E259600.5070103 at gmail.com...
> On 11-07-19 7:48 AM, Matthew Dowle wrote:
>> "Prof Brian Ripley"<ripley at stats.ox.ac.uk> wrote in message
>> news:alpine.LFD.2.02.1107190640280.28269 at gannet.stats.ox.ac.uk...
>>> On Mon, 18 Jul 2011, Alireza Mahani wrote:
>>>> Thank you for elaborating on the limitations of R in handling float
>>>> types. I
>>>> think I'm pretty much there with you.
>>>> As for the insufficiency of single-precision math (and hence
>>>> GPU), my personal take so far has been that double-precision becomes
>>>> when some sort of error accumulation occurs. For example, in
>>>> equations where boundary values are integrated to arrive at interior
>>>> etc. On the other hand, in my personal line of work (Hierarchical
>>>> models for quantitative marketing), we have so much inherent
>>>> noise at so many levels in the problem (and no significant error
>>>> accumulation sources) that single vs double precision issue is often
>>>> inconsequential for us. So I think it really depends on the field as
>>>> the nature of the problem.
>>> The main reason to use only double precision in R was that on modern
>>> double precision calculations are as fast as single-precision ones, and
>>> with 64-bit CPUs they are a single access.
>>> So the extra precision comes more-or-less for free.
>> But, isn't it much more of the 'less free' when large data sets are
>> considered? If a double matrix takes 3GB, it's 1.5GB in single.
>> That might alleviate the dreaded out-of-memory error for some
>> users in some circumstances. On 64bit, 50GB reduces to 25GB
>> and that might make the difference between getting
>> something done, or not. If single were appropriate, of course.
>> For GPU too, i/o often dominates iiuc.
>> For space reasons, is there any possibility of R supporting single
>> precision (and single bit logical to reduce memory for logicals by
>> 32 times)? I guess there might be complaints from users using
>> single inappropriately (or worse, not realising we have an instable
>> result due to single).
> You can do any of this using external pointers now. That will remind you
> that every single function to operate on such objects needs to be
> It's a huge amount of work, benefiting very few people. I don't think
> anyone in R Core will do it.
Ok, thanks for the responses.
> Duncan Murdoch
>>> You also under-estimate the extent to which stability of commonly used
>>> algorithms relies on double precision. (There are stable
>>> versions, but they are no longer commonly used. And as Simon said, in
>>> some cases stability is ensured by using extra precision where
>>> I disagree slightly with Simon on GPUs: I am told by local experts that
>>> the double-precision on the latest GPUs (those from the last year or so)
>>> is perfectly usable. See the performance claims on
>>> http://en.wikipedia.org/wiki/Nvidia_Tesla of about 50% of the SP
>>> performance in DP.
>>>> View this message in context:
>>>> Sent from the R devel mailing list archive at Nabble.com.
>>>> R-devel at r-project.org mailing list
>>> Brian D. Ripley, ripley at stats.ox.ac.uk
>>> Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/
>>> University of Oxford, Tel: +44 1865 272861 (self)
>>> 1 South Parks Road, +44 1865 272866 (PA)
>>> Oxford OX1 3TG, UK Fax: +44 1865 272595
>> R-devel at r-project.org mailing list
More information about the R-devel