MATLAB Central Newsreader - Seeking numerical package for quadruple precision...

Seeking numerical package for quadruple precision...

Vista — Tue, 03 Jul 2007 18:33:45 -0400

Hi all,

I have some computation and simulation which need quadruple precision. I
mean, the double precision is not enough, while high precision such as 1000
digits is not needed, and that's too slow. I guess quadruple precision
exactly fits my problem and should be much faster than double precision.

Could anybody tell me if MS Visual C++, or Intel C, Intel Fortran have
already got quadruple precision in them and they are fully supported on
Intel based cpus? I guess Intel Fortran has quadruple precision, but since I
prefer C/C++ so my first choices are within C/C++ languages.

Moreover, do GSL, Intel MKL, and IMSL etc. have got quadruple precision
support? I only need +, -, *, /, exp, and log.

I eventually have to do everything in C/C++ so I am primarily looking for
numerical libraries, instead of Matlab or Maple or Mathematica's symbolic
capability.

But if you know how to "simulate" quadruple precision in Matlab or Maple, or
Mathematica, in order to see if an algorithm will overslow when converting
into C/C++/Fortran, please let me know. I want to do the algorithm design in
Matlab, and test if it will overflow, before converting everything into
C/C++/Fortran.

If you know how to "simulate" quadruple precision in Matlab, Maple or
Mathematica even with the symbolic toolbox, please let me know too... this
is for algorithm design and testing...

Moreover, are there popular quadruple precision packages? Please recommend
the fastest one. I am really in huge need of speed.

Thank you very much!

Re: Seeking numerical package for quadruple precision...

Vista — Tue, 03 Jul 2007 18:53:39 -0400

I mean, I am looking for the fastest one, not just any one...

I know from Google there are quite a few quadruple packages out there.

But I need the fastest ones... Speed is really critical...

"Vista" <abc@gmai.com> wrote in message
news:f6etbs$2of$1@news.Stanford.EDU...
> Hi all,
>
> I have some computation and simulation which need quadruple precision. I
> mean, the double precision is not enough, while high precision such as
> 1000 digits is not needed, and that's too slow. I guess quadruple
> precision exactly fits my problem and should be much faster than double
> precision.
>
> Could anybody tell me if MS Visual C++, or Intel C, Intel Fortran have
> already got quadruple precision in them and they are fully supported on
> Intel based cpus? I guess Intel Fortran has quadruple precision, but since
> I prefer C/C++ so my first choices are within C/C++ languages.
>
> Moreover, do GSL, Intel MKL, and IMSL etc. have got quadruple precision
> support? I only need +, -, *, /, exp, and log.
>
> I eventually have to do everything in C/C++ so I am primarily looking for
> numerical libraries, instead of Matlab or Maple or Mathematica's symbolic
> capability.
>
> But if you know how to "simulate" quadruple precision in Matlab or Maple,
> or Mathematica, in order to see if an algorithm will overslow when
> converting into C/C++/Fortran, please let me know. I want to do the
> algorithm design in Matlab, and test if it will overflow, before
> converting everything into C/C++/Fortran.
>
> If you know how to "simulate" quadruple precision in Matlab, Maple or
> Mathematica even with the symbolic toolbox, please let me know too... this
> is for algorithm design and testing...
>
> Moreover, are there popular quadruple precision packages? Please recommend
> the fastest one. I am really in huge need of speed.
>
> Thank you very much!
>
>
>
>

Re: Seeking numerical package for quadruple precision...

Rune Allnor — Wed, 04 Jul 2007 08:10:35 -0400

On 4 Jul, 14:21, "Vista" <a...@gmai.com> wrote:

> I need 30 digits of precision digits.

Just out of curiousity - why?

Rune

Re: Seeking numerical package for quadruple precision...

dpb — Wed, 04 Jul 2007 11:16:33 -0400

Rune Allnor wrote:
> On 4 Jul, 14:21, "Vista" <a...@gmai.com> wrote:
>
>> I need 30 digits of precision digits.
>
> Just out of curiousity - why?

It's not clear "Vista" knows/understands the "why"...he's posted similar
topic in c.l.fortran and so far the answers have been anything but
enlightening.

I've been wondering where he's going to get the 30 digits of precision
for the initialization to start from myself... :)

--

Re: Seeking numerical package for quadruple precision...

Jerry Avins — Wed, 04 Jul 2007 12:02:33 -0400

Vladimir Vassilevsky wrote:
>
>
> Vista wrote:
>
>> I have some computation and simulation which need quadruple precision. I
>> mean, the double precision is not enough, while high precision such as
>> 1000 digits is not needed, and that's too slow. I guess quadruple
>> precision
>> exactly fits my problem and should be much faster than double precision.
>
> MIRACL
>
>> Moreover, are there popular quadruple precision packages? Please
>> recommend the fastest one. I am really in huge need of speed.
>
> And what is the problem to develop such a basic thing yourself?

I drove someone's assertive grandmother on an errand, and she screamed
when I turned left off the road, "There's a car coming!" I answered that
the car would pass well behind us (which it did) and she asked
indignantly how I could know that. I told her that knowing it was part
of being a competent driver. She answered, again indignantly, "Well, I
didn't know it" to which I answered, "That's an easy syllogism. Draw
your own conclusion." Vista seems to believe that quad precision is
"much faster" than double precision. What conclusion are we to draw?

Jerry
--
Engineering is the art of making what you want from things you can get.
Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

Re: Seeking numerical package for quadruple precision...

Vladimir Vassilevsky — Wed, 04 Jul 2007 15:09:47 -0400

Vista wrote:

> I have some computation and simulation which need quadruple precision. I
> mean, the double precision is not enough, while high precision such as 1000
> digits is not needed, and that's too slow. I guess quadruple precision
> exactly fits my problem and should be much faster than double precision.

MIRACL

> Moreover, are there popular quadruple precision packages? Please recommend
> the fastest one. I am really in huge need of speed.

And what is the problem to develop such a basic thing yourself?

Vladimir Vassilevsky

DSP and Mixed Signal Design Consultant

http://www.abvolt.com

Re: Seeking numerical package for quadruple precision...

Le Chaud Lapin — Wed, 04 Jul 2007 19:08:13 -0400

On Jul 4, 7:21 am, "Vista" <a...@gmai.com> wrote:
> "Le Chaud Lapin" <jaibudu...@gmail.com> wrote in messagenews:1183534955.773488.278220@k79g2000hse.googlegroups.com...
>
>
>
> > On Jul 3, 8:53 pm, "Vista" <a...@gmai.com> wrote:
> >> I mean, I am looking for the fastest one, not just any one...
>
> >> I know from Google there are quite a few quadruple packages out there.
>
> >> But I need the fastest ones... Speed is really critical...
>
> >> "Vista" <a...@gmai.com> wrote in message
>
> >>news:f6etbs$2of$1@news.Stanford.EDU...
>
> >> > Hi all,
>
> >> > I have some computation and simulation which need quadruple precision.
> >> > I
> >> > mean, the double precision is not enough, while high precision such as
> >> > 1000 digits is not needed, and that's too slow. I guess quadruple
> >> > precision exactly fits my problem and should be much faster than double
> >> > precision.
>
> >> > Could anybody tell me if MS Visual C++, or Intel C, Intel Fortran have
> >> > already got quadruple precision in them and they are fully supported on
> >> > Intel based cpus? I guess Intel Fortran has quadruple precision, but
> >> > since
> >> > I prefer C/C++ so my first choices are within C/C++ languages.
>
> >> > Moreover, do GSL, Intel MKL, and IMSL etc. have got quadruple precision
> >> > support? I only need +, -, *, /, exp, and log.
>
> >> > I eventually have to do everything in C/C++ so I am primarily looking
> >> > for
> >> > numerical libraries, instead of Matlab or Maple or Mathematica's
> >> > symbolic
> >> > capability.
>
> >> > But if you know how to "simulate" quadruple precision in Matlab or
> >> > Maple,
> >> > or Mathematica, in order to see if an algorithm will overslow when
> >> > converting into C/C++/Fortran, please let me know. I want to do the
> >> > algorithm design in Matlab, and test if it will overflow, before
> >> > converting everything into C/C++/Fortran.
>
> >> > If you know how to "simulate" quadruple precision in Matlab, Maple or
> >> > Mathematica even with the symbolic toolbox, please let me know too...
> >> > this
> >> > is for algorithm design and testing...
>
> >> > Moreover, are there popular quadruple precision packages? Please
> >> > recommend
> >> > the fastest one. I am really in huge need of speed.
>
> >> > Thank you very much!
>
> > In C++:
>
> > long double x;
>
> > You can tweak the compiler options for optimum speed on your
> > particular CPU.
>
> > -Le Chaud Lapin-
>
> I need 30 digits of precision digits. Will "long double" give me that?
>
> Thanks!

I think you mean to ask for "multiple precision", not "quadruple"
precision.

In C++, one of the top dogs for speed is GMP.

http://gmplib.org/

To a C++ programmer, the naming using in the code might be
surprising. For example, if you are looking for the "big integer"
class, don't expect it to be named "Integer" or anything like that.
Same for floating-point.

-Le Chaud Lapin-

Re: Seeking numerical package for quadruple precision...

carlos lopez — Thu, 09 Aug 2007 20:07:56 -0400

"Vista" <abc@gmai.com> wrote in message
<f6etbs$2of$1@news.Stanford.EDU>...
> Hi all,
>
> I have some computation and simulation which need
quadruple precision. I
> mean, the double precision is not enough, while high
precision such as 1000
> digits is not needed, and that's too slow.
I agree with this
> I guess quadruple precision exactly fits my problem and
should be much faster than double precision.
I disagree with this.
However, you might want to consider other alternatives aside
from "quadruple precision". From
http://crd.lbl.gov/~dhbailey/mpdist/ you can look at
double-double precision, quad-double precision, etc. They
manage to represent higher precision numbers with two or
four ordinary double precision numbers.
I knew of some ports of this stuff to the matlab environment.
Regards
Carlos