Control Systems and Computers

DOI: https://doi.org/10.15407/usim.2016.01.056

Upr. sist. maš., 2016, Issue 1 (261), pp. 56-61.

UDC 004.27; 004.274; 004.315

Yakovlev Yu.S., Doctor (Eng.), Glushkov Institute of Cybernetics of NAS of Ukraine, Kyiv, Glushkov ave., 40, Kyiv, 03187

About the Estimation of Efficiency of Application the PLIS in Structure PIM-System

Introduction. General provisions of multi sequencing of the application with reference to PIM-system with application the PLIS in correspondence with the formulated rules are observed. Three alternatives of block diagrammes of the PIM-systems using the PLIS are offered:

a) processor units supplement the PLIS for PYa;
b) processor units the PLIS are used instead of VP and supplement PYa;
c) the PIM-system is completely realised on the PLIS.

Methods. For alternative b expressions are gained and numerical estimations of efficiency of application the PLIS are fulfilled. On these expressions schedules of dependences of a time of implementation of q^th block of the program Θ_q and all program Θ_пр from execution time of t_s s^th operation (COPs) q^th block of the program by means of the processor of type VP, total of blocks of the program m and parametre δ – total of VP commands which codes of commands coincide with codes of operations q^th the program block are builted. It is shown, that efficiency of application the PLIS (instead of VP) essentially increases at parametre increase m, and also at parametre increase δ. It leads to expediency of mutual optimisation of these parametres, taking into consideration parallelism of data.

Conclusion. Output is drawn on expediency of performance of all PIM-system on the PLIS, that in addition allows to reduce a time of implementation of parallel algorithm at the expense of delay exclusion at an information transfer through the interface of link the PLIS with PIM-system which are fulfilled on separate chips, and also to reduce signal passage hold time through the diagramme of connections which connects all PYa and appropriate processor units the PLIS.

Download full text! (In Russian).

Keywords: structure PIM-System, mutual optimisation,  PLIS, supplement PYa.

1. Yakovlev, Yu.S., Komukhayev, Ye.I., 2013. “Razvitiye elementnoy bazy i sootvetstvuyushchikh arkhitekturno-struktur­nykh resheniy dlya sistem vysokoproizvoditel’nykh vychisleniy”. Upravlausie sistemy i masiny, 5, pp. 38–45. (In Russian).
2. Opisaniye FPGA-superkomp’yutera Maxwell. http://parallel. ru/FPGA/maxwell.html (In Russian).
3. Yakovlev, Yu.S., Yeliseyeva, Ye.V., 2014. “Primeneniye PLIS dlya sozdaniya vysokoproizvoditel’nykh vychislitel’nykh sistem i ikh komponentov”. Matematichní mashini í sistemi, 1, pp. 22–35. (In Russian).
4. Levin, I.I., 2008. “Semeystvo vysokoproizvoditel’nykh rekonfiguriruyemykh vychislitel’nykh sistem”. Vestn. Tomskogo gos. un-ta: Upravleniye, vychislitel’naya tekhnika i informatika, 2 (3), pp. 77–93. (In Russian).
5. Yakovlev, Yu.S., Yeliseyeva, Ye.V., 2015. “Primeneniye PLIS dlya skhemnoy realizatsii moduley programmy raspre­deleniya prilozheniy dlya PIM-sistemy”. Matematichny mashini i sistemy, 2, pp. 28–39. (In Russian).
6. Gergel, B.P., Fursov,V.A., 2009. Lektsii po parallel’nym vychisleniyam. Samara: Izd-vo SGAU, 166 p. (In Russian).
7. Gergel, V.P., 2007. Teoriya i praktika parallel’nykh vychisleniy: Ucheb. posobiye. M.: Internet-Universi­tet Informatsionnykh tekhnologiy; BINOM. Laboratoriya znaniy, 423 p. (In Russian).
8. Yakovlev, Yu.S., 2009. Odnokristal’nyye komp’yuternyye sistemy vysokoy proizvoditel’nosti. Osobennosti arkhitekturno-strukturnoy organizatsii i vnutrennikh protsessov. Vinnitsa: VNTU, 294 p. (In Russian).
9. Yakovlev, Yu.S., Yeliseyeva, Ye.V., 2011. “O realizatsii raspredeleniya prilozheniya dlya parallel’nogo vypolneniya na PIM-sisteme”. Informatsiyni tekhnologii ta komp’yuterna inzheneriya, 3, pp. 53–59. (In Russian).

Recieved 09.04.2015