This page says about fast quadruple precision:

'...this allows more efficient usage of hardware capabilities of modern CPUs, even exploiting some of the advanced instructions, not to mention possibilities of harnessing power of the GPU!'

I've been trying to use mp to solve the example singular value decomposition problem (illustrated on the same page) on an NVIDIA Quadro 2000 GPU that I've successfully used MATLAB's gpu functions on. I try the following code:

mp.Digits(34); mp.GuardDigits(0); format longG
A = gpuArray.rand(1000);
X = mp(A);
tic; [U,S,V] = svd(X); toc;

But on encountering line 3 MATLAB says:

Error using mp: unsupported argument type

This is probably because the argument I'm passing to mp is a GPU array that it doesn't know how to handle.

How can I get mp to do this?

Hello! is there anyway to accelerate the advanpix code , it takes too long time to run, any suggestions ?!

Advanpix already allows the declaration of quadruple precision as in the  IEEE 754 standard. 

It would be great to do the same with octuple precision since it is defined by this standard. 

My point here is not speed, but to test algorithms with this peculiar precision in a quite portable way.

Hi Pavel,

I hope you are doing fine?! I'm running Advanpix Version 3.9.4 Build 10481 on Windows and ORDEIG crashes MATLAB 2015B with a "FatalException" on the following problem:

KK = [ mp('1.2856486930664507983834710103110410e+00'), mp('0.0000000000000000000000000000000000e+00') ;

ee = ordeig(HH,KK);

When calling ordeig(double(HH),double(KK)), everything works fine. Please see below for my mp.Info output and the crash report.

Cheers, Stefan

>> mp.Info

This error was detected while a MEX-file was running. If the MEX-file

(I CAN SEND YOU THE CRASH DUMP IF NECESSARY.)

Hello,
I need to implement MATLAB built-in function accumarray(), which is not supported by the Advanpix toolbox (yet?). In order to use the same code for single/double/mp precision, I tried to overload accumarray by creating a new function in the mp class. Whenever I want to pass a vector of mp values to that function, it gets "messed up".

I have put this piece of code into the methods section of mp class:
function result = accumarray(subs, val, sz)
size(val)
val(1)
val(2)
keyboard
...
end

The code...
>> accumarray([1; 2; 2; 1; 1], mp([1;2;3;4;5]))
...returns...
ans =
5 1

ans =
1
2
3
4
5

Index exceeds matrix dimensions.

Error in mp/accumarray (line 5038)
val(2)

As you can see, the second argument has still its size (and numel) but all the content was somehow shrinked into the first element. Could you, please, help me with implementing of my own methods into the toolbox?

Best regards,
Vladimir

Hello,
I think I have found bug/different behavior in bsxfun():
a = pi;
a(1) = [];
bsxfun(@plus, a, ones(4,1))
...returns...
ans =
Empty matrix: 4-by-0

but

a = mp('pi');
a(1) = [];
bsxfun(@plus, a, ones(4,1))
...returns...
Error using mp/subsref (line 990)
Index exceeds matrix dimensions.

Error in mpbsxfun (line 71)

Error in mp/bsxfun (line 1132)
[varargout{1:nargout}] = mpbsxfun(varargin{:});

I am running:
MATLAB Version: 8.5.0.197613 (R2015a)

Multiprecision Computing Toolbox, (c) 2008-2017 Advanpix LLC.
Version : 4.3.2 Build 12170
Platform: Win64
Release : 2017-01-16

Trial version. Valid until 2017-01-30

Best regards,
Vladimir

Hello Pavel.

I got the following problem:

mp.Digits(349);
array=mp('2')*mp('10')^mp('-4'):mp('2')*mp('10')^mp('-4'):mp('1');
array(1,100)==mp('0.02')

ans =
logical
1

mp.Digits(350);
array=mp('2')*mp('10')^mp('-4'):mp('2')*mp('10')^mp('-4'):mp('1');
array(1,100)==mp('0.02')

ans =
logical
0

mp.Digits(351);
array=mp('2')*mp('10')^mp('-4'):mp('2')*mp('10')^mp('-4'):mp('1');
array(1,100)==mp('0.02')
ans =
logical
1

This apperas quite randomly, deppending on the used precision. You can try for example mp.Digits(352) to mp.Digits(355).

Thank you for your help.

I have the following question:

In one of my examples I need to compute the Singular Value Decomposition (SVD) of a negative scalar, specifically -0.7276.

In double precision, Matlab returns:

>> [U S V]=svd(-0.7276)

In MP, I obtain:

s=mp('-0,7276');

>> [U,S,V]=svd(s)

The definition of SVD states that the matrix S should have nonnegative diagonal entries. In the above example, the first two elements seem to be multiplied by -1 compared to Matlab output. While the product doesn't change, the unexpected negative sign of the singular value causes problems in subsequent code.

Is this a bug or is there an option to make it return the same output as Matlab's svd?

The 4.2.3.11967 release of the toolbox was used.

Thank you,

Denis