Dev

dev@dynare.org

3402 discussions

An update on parallelization on GPU
by Sébastien Villemot 15 Jul '11

15 Jul '11

Hi, This morning I have documented myself a little bit on the subject of exploiting parallelization on GPUs (Graphical Processing Units). George had made a presentation on this topic a few years ago, and things have evolved a little bit since. There seem to be now two major competing programming frameworks for parallelization on GPUs (and possibly CPUs also): - CUDA, developed by Nvidia [1] - OpenCL, presented by a consortium (including Apple, AMD, ATI, Nvidia) [2] CUDA was initially only available on Nvidia graphics cards (see the packages "nvidia-cuda-*" in Debian, which contain the non-free implementation of CUDA for Nvidia cards). There is now a free implementation of CUDA, called GPU Ocelot [3], which looks promising. It is able to run CUDA programs on an heterogenous cluster of GPUs (Nvidia, AMD, Intel) and CPUs (multi-core x86). It is not yet packaged in Debian, but it seems very interesting and we should give it a try. On the long term, OpenCL looks more promising, but there is currently no free implementation of it. There is a Debian package called "opencl-headers", but it contains only the headers, since the binary implementation is non-free. There are also Debian packages containing the Nvidia non-free implementation of OpenCL (see "nvidia-opencl-*"). I am not aware of any ongoing project to build a free implementation of OpenCL. To summarize: whoever has the interest and time in our team should give a try to GPU Ocelot to see if we can take advantage of it in Dynare. Don't hesitate to react to this mail if you know things on this topic of which I am not aware. Best, [1] http://en.wikipedia.org/wiki/CUDA [2] http://en.wikipedia.org/wiki/OpenCL [3] http://code.google.com/p/gpuocelot/ -- Sébastien Villemot CEPREMAP - Paris School of Economics Homepage: http://www.dynare.org/sebastien Landline phone: +33 1 40 77 49 90 SIP phone: sebastien.villemot(a)ekiga.net PGP Key: 2069 1DFC C2C9 8C47 9529 84EE 0001 8C22 381A 7594

2 1

Fwd: Re: Suggestion
by Michel Juillard 14 Jul '11

14 Jul '11

... Thanks Johannes. This is an excellent suggestion. I pushed the modification to GIT. All the best Michel On 07/14/2011 12:01 PM, Johannes Pfeifer wrote: > /Please check i) if all parameters occurring in these equations are > defined and ii) that no division by an endogenous variable initialized > to 0 occurs. /

1 0

[Trac] #189: identification options
by Trac Dynare 12 Jul '11

12 Jul '11

#189: identification options --------------------------+------------------------------------------------- Reporter: houtanb | Owner: houtanb Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Preprocessor | Version: Keywords: | --------------------------+------------------------------------------------- Add lik_init and/or diffuse_filter -- Ticket URL: <https://www.dynare.org/trac/ticket/189> Dynare <http://www.dynare.org> The Dynare project

1 3

[Trac] #187: macrovariable passed on the command line (-D option)
by Trac Dynare 12 Jul '11

12 Jul '11

#187: macrovariable passed on the command line (-D option) --------------------------+------------------------------------------------- Reporter: michel | Owner: sebastien Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Preprocessor | Version: Keywords: | --------------------------+------------------------------------------------- It works for numerical values but not for strings. In the test.mod file: @#if var1 == "a" @#echo "OK" @#else @#echo "NO" @#endif Calling dynare test -Dvar1=a will fail with with unknown variable a (without any Dynare instruction, it is normal that test.mod fails at the end, but with an error message about unexpected $end) -- Ticket URL: <https://www.dynare.org/trac/ticket/187> Dynare <http://www.dynare.org> The Dynare project

1 4

[Trac] #193: add new sensitivity options
by Trac Dynare 12 Jul '11

12 Jul '11

#193: add new sensitivity options --------------------------+------------------------------------------------- Reporter: houtanb | Owner: houtanb Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Core M-files | Version: Keywords: | --------------------------+------------------------------------------------- pvalue_ks (default 0.001) threshold pvalue for significant Kolmogorov Smirnov test (i.e. plot parameters with pvalue<pvalue_ks) pvalue_corr (default 0.001) threshold pvalue for significant correlation in filtered samples (i.e. plot bivariate samples when pvalue<pvalue_corr) -- Ticket URL: <https://www.dynare.org/trac/ticket/193> Dynare <http://www.dynare.org> The Dynare project

1 3

[Trac] #192: Bug in irf_shocks (reported by Johannes Pfeifer)
by Trac Dynare 12 Jul '11

12 Jul '11

#192: Bug in irf_shocks (reported by Johannes Pfeifer) --------------------------+------------------------------------------------- Reporter: houtanb | Owner: houtanb Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Core M-files | Version: Keywords: | --------------------------+------------------------------------------------- * when specifying options_.irf_shocks in stoch_simul, Dynare crashes * also add irf_shocks to estimation command -- Ticket URL: <https://www.dynare.org/trac/ticket/192> Dynare <http://www.dynare.org> The Dynare project

1 2

timings
by Stéphane Adjemian 25 Jun '11

25 Jun '11

Salut Ferhat, J'étais un peu étonné par les gains de temps (pour l'estimation de sw) que tu annonces dans tes slides. Voilà les résultats que j'obtiens sur mon portable : block+bytecode ........ 256 sec. (comme toi) linear ................ 215 sec. linear+use_dll ........ 209 sec. no options ............ 280 sec. (largement < 9 minutes annoncées). Clairement c'est mieux d'utiliser block+bytecode que rien du tout, mais la différence est marginale. La meilleure solution c'est quand même linear+use_dll (en fait c'est linear qui fait la différence). Bref je crois qu'il y a un problème avec tes comparaisons. Je vais essayer sur un SW non linéaire... à+ Stéphane.

2 1

[Trac] #185: parallel computing: default missing for option NodeWeight
by Trac Dynare 14 Jun '11

14 Jun '11

#185: parallel computing: default missing for option NodeWeight --------------------------+------------------------------------------------- Reporter: michel | Owner: Type: bug | Status: new Priority: major | Milestone: 4.3 Component: Core M-files | Version: Keywords: | --------------------------+------------------------------------------------- ./matlab/parallel/InitializeComputationalEnvironment.m crashes when no value is indicated for node weights in Dynare configuration file -- Ticket URL: <https://www.dynare.org/trac/ticket/185> Dynare <http://www.dynare.org> The Dynare project

3 5

[Trac] #186: normalization is buggy in preprocessor for expressions of the form a^x
by Trac Dynare 10 Jun '11

10 Jun '11

#186: normalization is buggy in preprocessor for expressions of the form a^x --------------------------+------------------------------------------------- Reporter: sebastien | Owner: ferhat Type: bug | Status: new Priority: major | Milestone: Component: Core M-files | Version: 4.2.1 Keywords: | --------------------------+------------------------------------------------- In ExprNode.cc, in the function BinaryOpNode::normalizeEquation, the case of an expression of the form a^x (power expression with endogenous in the exponent but not under the exponent) is not handled. This make the preprocessor exit silently with an error. -- Ticket URL: <https://www.dynare.org/trac/ticket/186> Dynare <http://www.dynare.org> The Dynare project

1 2

Release of Dynare 4.2.1
by Sébastien Villemot 24 May '11

24 May '11

Dear Dynare users, We are pleased to announce the release of Dynare 4.2.1. Many bugs have been fixed since the previous release. The reference manual has also been improved: new contents has been added at various places, the structure has been improved, an index of functions and variables has been added, the PDF/HTML rendering has been improved. The Windows package is already available for download at the official Dynare website [1]. The Mac and Linux packages should follow soon. All users are strongly encouraged to upgrade. This release is compatible with MATLAB versions ranging from 7.0 (R14) to 7.12 (R2011a) and with GNU Octave versions ranging from 3.0 to 3.4. Here is a list of the main bugfixes since version 4.2.0: * The `STEADY_STATE' operator has been fixed * Problems with MATLAB 7.3 (R2006b) and older have been fixed * The `partial_information' option of `stoch_simul' has been fixed * Option `conditional_variance_decomposition' of `stoch_simul' and `estimation' has been fixed * Automatic detrending now works in conjunction with the `EXPECTATION' operator * Percentage signs inside strings in MATLAB statements (like disp('% This is not a comment %')) now work * Beta prior with a very small standard deviation now work even if you do not have the MATLAB Statistical toolbox * External functions can now been used in assignment of model local variables * `identification' command has been fixed * Option `cova_compute' of `estimation' command has been fixed * Random crashes with 3rd order approximation without `use_dll' option have been eliminated [1] http://www.dynare.org On behalf of the Dynare team, -- Sébastien Villemot CEPREMAP - Paris School of Economics Homepage: http://www.dynare.org/sebastien Landline phone: +33 1 40 77 49 90 SIP phone: sebastien.villemot(a)ekiga.net PGP Key: 0xA6C029B9D06B2913D71C105EBE37E801FB6EFF8B (http://pgp.mit.edu/)

1 0

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

Dev