Affiliation:
1. National University of Science and Technology «MISIS»
2. PJSC «UEC-Kuznetsov»
Abstract
The utilization of computer simulation software for casting process simulation is becoming essential in the advancement of casting technology in aviation and other high-tech engineering fields. With the increase in the number of computational cores in modern CPUs, the use of multi-threaded computations is becoming increasingly relevant. In this study, the efficiency of multi-threaded computations in modeling casting processes was evaluated using finite element method casting simulation software ProCast and PoligonSoft, which utilize parallel computing architectures with distributed (DMP) and shared (SMP) memory, respectively. Computations were performed on Intel and AMD-based computers, varying the number of computational threads from 4 to 32. The calculation efficiency was evaluated by measuring the calculation speed increase in the filling and solidification of GP25 castings made of ML10 alloy, as well as the complex task of filling and solidification modeling nickel superalloy casing castings with radiation heat transfer simulation. The results indicate that the minimum computation time in ProCast software is observed when using 16 computational threads. This pattern holds true for both computing systems (Intel and AMD processors), and increasing the number of threads beyond this point does not make a practical difference. The performance decrease in this scenario can be attributed to the low-performance energy-efficient cores in systems based on Intel processors or the decrease in core frequency and full loading of physical cores in systems based on AMD processors. Multi-threading the modeling task in PoligonSoft software is less efficient than in ProCast, which is a result of the shared-memory architecture used in PoligonSoft. Despite the significant difference in parallel efficiency, the task of GP25 casting solidification in both PoligonSoft and ProCast is solved in a time close enough to be considered sufficient.
Publisher
National University of Science and Technology MISiS
Reference28 articles.
1. Schwiegelshohn U., Badia R.M., Bubak M., Danelutto M., Dustdar S., Gagliardi F., Geiger A., Hluchy L., Kranzlmüller D., Laure E., Priol T., Reinefeld A., Resch M., Reuter A., Rienhoff O., Rüter T., Sloot P., Talia D., Ullmann K., Yahyapour R., von Voigt G. Perspectives on grid computing. Future Generation Computer Systems. 2010;26:1104—1115. https://doi.org/10.1016/j.future.2010.05.010
2. Konopka K., Miłkowska-Piszczek K., Trebacz L., Falkus J. Improving efficiency of ccs numerical simulations through use of parallel processing. Archives of Metallurgy and Materials. 2015;60(1):235—238. https://doi.org/10.1515/amm-2015-0037
3. Istomin V.A., Shvarts D.R., Ishkhanov E.A., Tikhomirov M.D. Using the meshfree smoothed-particles method in the simulation of hydrodynamics in the casting simulation software “PoligonSoft”. Liteinoe Proizvodstvo. 2012;(8):20—22 (In Russ.).
4. Cai Y., Li G., Wang H., Zheng G., Lin S. Development of parallel explicit finite element sheet forming simulation system based on GPU architecture. Advances in Engineering Software. 2012;45:370—379. https://doi.org/10.1016/j.advengsoft.2011.10.014
5. Posey S., Kodiyalam S. Performance benefits of NVIDIA GPUs for LS-DYNA®. In: 8th European LS-DYNA Users Conference (May 2011). Strasbourg, 2011. P. 1—6.