Changes between Version 1 and Version 2 of 2020WP/HPC-10_mcastril_HPDAonlineDiagGPU
- Timestamp:
- 2020-01-08T10:13:05+01:00 (4 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
2020WP/HPC-10_mcastril_HPDAonlineDiagGPU
v1 v2 11 11 == Summary 12 12 13 ||=Action || Name and subject of the action||14 ||=PI(S) || Names||15 ||=Digest || Brief description with motivations and main tasks||16 ||=Dependencies || If any||13 ||=Action || HPC-10_mcastril_HPDAonline DiagGPU || 14 ||=PI(S) || Miguel Castrillo || 15 ||=Digest || High Performance GPU Diagnostics Online - 2nd Phase. After having successfully ported the dia_hsb diagnostic into a toy model, achieving 50x speedup, this task will focus on implementing the rest of the diagnostics and improving the data transfer between CPU and GPU. || 16 ||=Dependencies || HPC-04_MCastrillo_HPDAonlineDiagGPU (completed) || 17 17 ||=Branch || source:/NEMO/branches/{YEAR}/dev_r{REV}_{ACTION_NAME} || 18 ||=Previewer(s) || Names||19 ||=Reviewer(s) || Names||18 ||=Previewer(s) || Italo Epicoco || 19 ||=Reviewer(s) || Italo Epicoco || 20 20 ||=Ticket || #XXXX || 21 21 22 22 === Description 23 23 24 {{{#!box width=25em help 25 Describe the goal of development and the methodology, \\ 26 add reference documents or publications if relevant. 27 }}} 24 High performance data analytics solutions aiming at tackling the online diagnostics of the NEMO model will be explored as complementary components in the model diagnostics software eco-system. Online techniques leveraging fast (low latency and real-time) data analytics approaches (e.g. on fat nodes) will be evaluated in real cluster environments. In particular, an interface of NEMO to the High Performance Data Analitics (HPDA) framework will be designed and implemented for online diagnostics. 28 25 29 ''...'' 26 The rationale of this activity is to improve the NEMO computational performance by executing the computations for diagnostics on GPU. 30 27 31 28 === Implementation 32 29 33 {{{#!box width=35em help 34 Describe flow chart of the changes in the code. \\ 35 List the Fortran modules and subroutines to be created/edited/deleted. \\ 36 Detailed list of new variables to be defined (including namelists), \\ 37 give for each the chosen name and description wrt coding rules. 38 }}} 30 As first step, the portability of NEMO diagnostic calculations to GPUs has been analyzed, exploring how to adapt these regions from the current MPI implementation to the CUDA paradigm. A toy model has been created to perform preliminary tests, that were done using the dia_hsb diagnostic. The code itself is executed 50x faster than in a single CPU but the data transfer to and from GPU is the main bottleneck. 39 31 40 ''...'' 32 We are working on the asynchronous strategy in order to hide all communications among GPU/CPU. We also plan to increase the efficiency of the overall solution, by mitigating the impact of the offloaded data and extending our approach to the rest of the diagnostics. 41 33 42 34 === Documentation updates 43 35 44 {{{#!box width=55em help45 Using previous parts, define the main changes to be done in the NEMO literature46 (manuals, guide, web pages, …).47 }}}48 49 ''...''50 36 51 37 == Preview