Quad precision?

marupio · May 17, 2011, 15:17

How much trouble would it be to try implementing quad precision in OpenFOAM?

- Easy!
- Tough!
- Why would you want to try? The schemes are only "second order accurate", etc..

deepsterblue · May 17, 2011, 15:27

Are you referring to numerical precision, or something like fourth-order accuracy?

marupio · May 17, 2011, 15:49

Need more digits! I mean using "long double", or __float128.

deepsterblue · May 17, 2011, 15:54

I've implemented an arbitrary-precision data-type as a wrap-around to the GNU MPFR library. You should be able to use this as a drop-in replacement to the scalar class:

http://www.ecs.umass.edu/%7Esmenon/t...iPrecision.tgz

However, this is still emulation in software, so don't expect blazing performance.

marupio · May 17, 2011, 16:36

Thanks a lot, Sandeep!

I've downloaded it, and am in the process of figuring out how to use it. I'm guessing to use it in finiteVolume calculations, I need to define a new fvMatrix template specialization for a multiPrecisionScalar... fvMultiPrecisionScalar (basically identical to fvScalarMatrix).

Do you have any sample code that uses your multiPrecisionScalar class?

-Dave

May 17, 2011, 15:17	Quad precision?	#1
marupio Senior Member David Gaden Join Date: Apr 2009 Location: Winnipeg, Canada Posts: 437 Rep Power: 22	How much trouble would it be to try implementing quad precision in OpenFOAM? - Easy! - Tough! - Why would you want to try? The schemes are only "second order accurate", etc..

May 17, 2011, 15:27		#2
deepsterblue Senior Member Sandeep Menon Join Date: Mar 2009 Location: Amherst, MA Posts: 403 Rep Power: 25	Are you referring to numerical precision, or something like fourth-order accuracy? __________________ Sandeep Menon University of Massachusetts Amherst https://github.com/smenon

May 17, 2011, 15:54		#4
deepsterblue Senior Member Sandeep Menon Join Date: Mar 2009 Location: Amherst, MA Posts: 403 Rep Power: 25	I've implemented an arbitrary-precision data-type as a wrap-around to the GNU MPFR library. You should be able to use this as a drop-in replacement to the scalar class: http://www.ecs.umass.edu/%7Esmenon/t...iPrecision.tgz However, this is still emulation in software, so don't expect blazing performance. __________________ Sandeep Menon University of Massachusetts Amherst https://github.com/smenon

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Switch from single precision to double precision	kk81	OpenFOAM Bugs	0	March 11, 2010 07:14
Number of elements x double precision	matheusguzella	ANSYS Meshing & Geometry	2	October 7, 2009 11:15
Parallel computing quad core	Prad	Main CFD Forum	13	February 9, 2009 15:28
what's wrong about my code for 2d burgers equation	morxio	Main CFD Forum	3	April 27, 2007 11:38
double precision assigning in FRTRAN	R.A.Khurram	Main CFD Forum	12	October 2, 2001 12:00

May 17, 2011, 15:49		#3
marupio Senior Member David Gaden Join Date: Apr 2009 Location: Winnipeg, Canada Posts: 437 Rep Power: 22	Need more digits! I mean using "long double", or __float128.

May 17, 2011, 16:36		#5
marupio Senior Member David Gaden Join Date: Apr 2009 Location: Winnipeg, Canada Posts: 437 Rep Power: 22	Thanks a lot, Sandeep! I've downloaded it, and am in the process of figuring out how to use it. I'm guessing to use it in finiteVolume calculations, I need to define a new fvMatrix template specialization for a multiPrecisionScalar... fvMultiPrecisionScalar (basically identical to fvScalarMatrix). Do you have any sample code that uses your multiPrecisionScalar class? -Dave