Trac September 2011

trac@lists.einsteintoolkit.org

1 participants
184 discussions

[Einstein Toolkit] #545: Use ssh ProxyCommand instead of multiple ssh commands
by Einstein Toolkit 08 Nov '11

08 Nov '11

#545: Use ssh ProxyCommand instead of multiple ssh commands -------------------------+-------------------------------------------------- Reporter: eschnett | Owner: eschnett Type: enhancement | Status: new Priority: major | Milestone: Component: SimFactory | Version: Keywords: | -------------------------+-------------------------------------------------- This would simplify quoting, thus also simplify the screen output, and may speed up things on remote connections. Here is an example: ssh -Y -o ProxyCommand='ssh eschnetter(a)mars.perimeterinstitute.ca nc -w 1 curry.pi.local 22' eschnett(a)curry.pi.local -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/545> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 9

[Einstein Toolkit] #544: Enable optimisation in Fortran array index calculations
by Einstein Toolkit 27 Oct '11

27 Oct '11

#544: Enable optimisation in Fortran array index calculations -------------------------+-------------------------------------------------- Reporter: eschnett | Owner: Type: enhancement | Status: new Priority: major | Milestone: Component: Cactus | Version: Keywords: | -------------------------+-------------------------------------------------- In Fortran, Cactus passes all grid functions as arguments to routines, and the size of the grid functions are declared in DECLARE_CCTK_ARGUMENTS. In these declarations, Cactus assumes that the size of each grid variable group can be different, and the compiler thus does not see that all grid functions have (locally) the same array shape. The compiler can thus not simplify array index calculations. This is probably only relevant in short loops accessing grid functions from multiple groups (GRHydro?). The attached patch passes three additional integer arguments (cctk_lsh[123]) which contain the shape of grid functions, and uses these to declare grid functions. This patch is quite old, but was never applied. It comes from a time when g77 didn't support declaring array shapes via an integer array (since this is not allowed in Fortran 77). If this is now possible, one could use the array cctk_lsh instead of passing three extra integer arguments. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/544> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 7

[Einstein Toolkit] #235: Improve performance of Fortran index calculations
by Einstein Toolkit 27 Oct '11

27 Oct '11

#235: Improve performance of Fortran index calculations -------------------------+-------------------------------------------------- Reporter: eschnett | Owner: Type: enhancement | Status: new Priority: minor | Milestone: Component: Cactus | Version: Keywords: | -------------------------+-------------------------------------------------- In Fortran, Cactus currently declares grid functions e.g. as (this is the expansion of DECLARE_CCTK_ARGUMENTS) REAL*8 gxx (X0metric,X1metric,X2metric) where X0metric etc. are integers passed into the routine. Each grid function group has its own, independent size. This has two disadvantages: 1. The compiler does not know that all grid functions have the same size (namely cctk_lsh), and thus has to perform array index calculations separately for each group 2. The argument list is longer than neded The enclosed patch declares grid functions via cctk_lsh. Grid arrays are still declared independently. This reduces the code size of e.g. GRHydro/GRHydro_Tmunu.F90 from 6836 to 6241 bytes on my system. I have not attempted to measure a performance difference. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/235> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 6

[Einstein Toolkit] #458: Improve bboxset efficiency (e.g. for regridding)
by Einstein Toolkit 27 Oct '11

27 Oct '11

#458: Improve bboxset efficiency (e.g. for regridding) -------------------------+-------------------------------------------------- Reporter: eschnett | Owner: eschnett Type: enhancement | Status: new Priority: major | Milestone: Component: Carpet | Version: Keywords: | -------------------------+-------------------------------------------------- The attached patch modifies the algorithm used to insert a new bbox into an existing bboxset. It reduces the computational complexity of this operation from O(n^2) to O(n), where n is the number of elements in the bboxset. This has the potential to speed up regridding significantly. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/458> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 4

[Einstein Toolkit] #329: SimFactory/ranger-intel11.cfg: Are BLAS and LAPACK compilations on Ranger optimized?
by Einstein Toolkit 27 Oct '11

27 Oct '11

#329: SimFactory/ranger-intel11.cfg: Are BLAS and LAPACK compilations on Ranger optimized? -------------------------+-------------------------------------------------- Reporter: bmundim | Owner: mthomas Type: enhancement | Status: new Priority: minor | Milestone: Component: SimFactory | Version: Keywords: | -------------------------+-------------------------------------------------- ranger-intel11.cfg now sets BLAS and LAPACK to be built along with the Cactus and other external libraries. Usually the host system BLAS and LAPACK are fine tuned to take advantage of the particular memory hierarchy of the host machine. I was wondering if this fine tuning is performed when we set BLAS = BUILD or LAPACK = BUILD, and, in case it is not, if it would be interesting to define Cactus env variables to set the size of the several cache memories for example and use it for this necessary fine tuning. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/329> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 5

[Einstein Toolkit] #271: python simfactory submit creates only the SubmitScript
by Einstein Toolkit 27 Oct '11

27 Oct '11

#271: python simfactory submit creates only the SubmitScript ----------------------------------------------+----------------------------- Reporter: alexander.beck-ratzka@… | Type: defect Status: new | Priority: major Milestone: | Component: Other Version: | Keywords: simfactory submit ----------------------------------------------+----------------------------- The python version of simfactory submit creates only the submitScript called PreparedSubmitScript. The RunScript is created at the moment of the simfactory run call at the end of the SubmitScript. So a user has no possibilty to check whether his settings e.g. for mpirun have been correct or not for all the time where the job is only queued. I feel this is not a user friendly solution. The RunScript should be created together with the SubmitScript. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/271> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 7

[Einstein Toolkit] #326: Update gsl path in Ranger option file to gsl 1.13
by Einstein Toolkit 27 Oct '11

27 Oct '11

#326: Update gsl path in Ranger option file to gsl 1.13 ---------------------------------------------+------------------------------ Reporter: bmundim | Owner: mthomas Type: task | Status: new Priority: minor | Milestone: Component: SimFactory | Version: Keywords: Ranger option configuration gsl | ---------------------------------------------+------------------------------ Ranger config file has been updated recently, so I thought to bring to your attention that gsl library on Ranger now defaults to version 1.13. Cheers, Bruno. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/326> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 7

[Einstein Toolkit] #327: Simfactory: ranger-intel11.run question.
by Einstein Toolkit 27 Oct '11

27 Oct '11

#327: Simfactory: ranger-intel11.run question. ------------------------+--------------------------------------------------- Reporter: bmundim | Owner: mthomas Type: defect | Status: new Priority: minor | Milestone: Component: SimFactory | Version: Keywords: | ------------------------+--------------------------------------------------- Shouldn't we load intel/11.1 there, since the default on Ranger is intel/10.1? Also unless "module load intel/11.1" is in your .login_user, it will fail to "module load mvapich" since no compiler has been selected. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/327> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 2

[Einstein Toolkit] #472: Correct scheduling of TmunuBase's SetTmunu
by Einstein Toolkit 27 Oct '11

27 Oct '11

#472: Correct scheduling of TmunuBase's SetTmunu ----------------------+----------------------------------------------------- Reporter: eschnett | Owner: Type: defect | Status: new Priority: minor | Milestone: Component: Other | Version: Keywords: | ----------------------+----------------------------------------------------- I notice that TmunuBase doesn't schedule the group SetTmunu often enough. In particular, it is not scheduled after recovery, and it's scheduling at evol is also questionable. I also notice that GRHydro "fixes" this by scheduling SetTmunu by itself. This needs to be cleaned up. -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/472> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 2

[Einstein Toolkit] #480: Problem with determining host name in SimFactory 1
by Einstein Toolkit 27 Oct '11

27 Oct '11

#480: Problem with determining host name in SimFactory 1 ------------------------+--------------------------------------------------- Reporter: hinder | Owner: eschnett Type: defect | Status: new Priority: major | Milestone: Component: SimFactory | Version: Keywords: | ------------------------+--------------------------------------------------- On the new Scientific Linux 6 installation at the AEI, SimFactory 1 thinks that the local machine is called localhost.localdomain: $ simfactory/sim build --thornlist WaveToy.th --optionlist damiana_sl6.cfg Simulation Factory: Configuration name(s) not specified -- using default configuration "sim" Uncaught exception from user code: Unknown machine name "localhost.localdomain" at simfactory/sim line 5061. at simfactory/sim line 5061 main::get_machine('') called at simfactory/sim line 1498 main::command_build() called at simfactory/sim line 454 Looking at the code for get_hostnamealias, it appears to look for all the host name aliases of the current machine and select the first one in the returned list which contains a dot. On the workstation in question, I get: [ianhin@sl-18 ~]$ hostname -a localhost.localdomain localhost localhost6.localdomain6 localhost6 sl-18 sl-18.damiana.admin which explains the behaviour. My first reaction would be to filter the list to remove all the "local" aliases. This might break peoples' existing setup. I have seen mailing list posts which indicate that people use localhost as their machine name in simfactory. Since SimFactory 1 is approaching end-of-life, it might just be better to say that you need to use the --hostname option directly if you run into this problem, rather than breaking peoples' existing workflow. Thoughts? -- Ticket URL: <https://trac.einsteintoolkit.org/ticket/480> Einstein Toolkit <http://einsteintoolkit.org> The Einstein Toolkit

1 4

← Newer
1
...
8
9
10
11
12
13
14
...
19
Older →

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

Trac September 2011