= [=#wpm3]Work package M3: Benchmarking and internal validation = \\ == Contact information == '''Principal investigators:''' [[http://www.muk.uni-hannover.de/215.html?&no_cache=1&tx_tkinstpersonen_pi1%5Balias%5D=maronga|Dr. Björn Maronga]] [[Image(htdocs:other_images/mail.gif,title=Send email, link=mailto:maronga@muk.uni-hannover.de)]], [[http://www.muk.uni-hannover.de/215.html?&no_cache=1&tx_tkinstpersonen_pi1%5Balias%5D=raasch|Prof. Dr. Siegfried Raasch]] [[Image(htdocs:other_images/mail.gif,title=Send email, link=mailto:raasch@muk.uni-hannover.de)]] '''Staff:''' [[https://www.muk.uni-hannover.de/215.html?&no_cache=1&tx_tkinstpersonen_pi1%5Balias%5D=kanani|Dr. Farah Kanani-Sühring]] [[Image(htdocs:other_images/mail.gif,title=Send email, link=mailto:kanani@muk.uni-hannover.de)]] '''Institution:''' Institute of Meteorology and Climatology (http://www.muk.uni-hannover.de/), Leibniz Universität Hannover (https://www.uni-hannover.de/) \\\\ == Goals of WP-M3 == {{{#!table align=left style="border: none; text-align:left" {{{#!tr {{{#!td style="border: none; width:40%" The main goal of WP-M3 will be the internal validation of PALM-4U during its entire development process, involving several steps of a well-defined testing procedure, in close discussion with the coordinator (WP-C). These include code revision as well as testing of functionality of the different PALM-4U components and their interactions. Further, continuous performance checks (benchmarking) of the new components will be performed on supercomputer systems, optimizations PALM-4U with respect to computing time and scalability of the code on large numbers of processor cores. Feedback of the testing process will be returned to the PALM-4U developers, who will incorporate the revisions during the further code development. The workflow of this entire process is visualized in Fig. 1.\\ An essential goal of MOSAIK is to demonstrate that PALM-4U is able to run with limited computer resources – as it is the case e.g. in city-planning applications. These tests will be done on a demonstration PC with a reasonable number of processor cores. Besides these more technical tests, a verification of PALM-4U against the VDI requirements and a comparison with results from existing urban climate models will be part of WP-M3. The testing procedure requires a pre-definition of specific simulation set-ups, covering the main application scenarios. A documentation of the set-ups will be provided on the Trac server, and selected PALM-4U results will be added there at the end of the project. }}} {{{#!td align=left style="border: none; vertical-align:top; width: 60%" [[Image(WPM3_fig1_en.png, 100%, left, border=1, color=black)]]\\ Fig. 1: Workflow of development, checking and approval for PALM-4U. }}} }}} }}} \\\\\\ == Work program == The project starts with defining the setups and the testing procedure, and improving the installation- and run-scripts for PALM-4U. Furthermore, the demonstration PC will be configured. As soon as new PALM-4U components are available, they will be passed through the testing procedure. \\\\ ''WP-M3.1: Definition of test- and benchmark-setups, improvement of PALM-4U installation/run scripts, setup of standard PC'' The test- and benchmark setups will be developed in close coordination with all other work packages (especially WP-C and WP-D3). PALM-4U installation- and run-scripts will be developed and improved, based on existing PALM scripts. The project standard PC will be configured, which includes installation of compilers, libraries, and further tools (MPI, FFTW, netCDF4/HDF, NCL, etc.). PALM will be implemented on the standard PC and work packages will be given access to the system. A one-week seminar “Introduction to the PArallelized LES Model PALM” will be given for all project partners of module A, and interested partners of modules B/C, at project start, in order to make them familiar with the PALM-code as the basis of the new PALM-4U-code. ''WP-M3.2: Installation and application of existing urban climate models'' Existing urban climate models will be installed and run for the test-setups (as far as possible). A focus will be on codes currently used in Germany, like MUKLIMO_3, ASMUS, ENVIMET, etc.. Simulation results will later be used for comparison with results from PALM-4U. ''WP-M3.3: Test and benchmark runs'' New parts of PALM-4U code that create performance bottlenecks will be identified and scheduled for further optimization, including porting for accelerator boards (to be partly done by a sub-contract partner). Feedback will be given to work packages in order to improve the respective code. An optimized PALM-4U version will be availabe after 24 months. ''WP-M3.4: Final documentation of test- / benchmark-setups including results'' Besides the final documentation, first versions of test setup descriptions and results will be made available on the PALM-4U Trac server as soon as the setups have been defined.