1. Electronics Lab, Physics Dept., Aristotle Univ. of Thessaloniki, Greece 17th IEEE International Conference on Electronics, Circuits, and Systems ICECS 2010 – Athens - Greece

2.  Motivation  Architecture Platform  Design Space Exploration  Results  Conclusions C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab 2

3.  Motivation  Architecture Platform  Design Space Exploration  Results  Conclusions C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab 3

4.  Modern multimedia applications › Increased need for computational power  High resolution/throughput imaging/digital signal processing › Need for larger memory space  Modern FPGA devices › Larger › More powerfull › Offer a variety of memory architectures › MPSoC capabilities  Formulate the complete Design Space Exploration problem 4 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

5.  Design space of MPSoC FPGA platforms taking into account: › The number of processors › Data/task level parallelism › Different interconnection strategies › Different memory architectures offered 5 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

6.  Motivation  Architecture Platform  Design Space Exploration  Results  Conclusions C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab 6

7.  MPSoC based on the Microblaze processor › Exploration of systems with one to four processors › Interconnection of the processors chosen  Fast Simplex Links (FSL)  FIFO based therefore can also serve as a data buffer  Different memory architectures used › External memory  DDR2 on the xupv5-lx110t (Virtex-5 board) › Local BRAM › Combination of both for all architectures 7 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

8.  Motivation  Architecture Platform  Design Space Exploration  Results  Conclusions C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab 8

9.  Algorithm selection › A widely used streaming multimedia application › Different types of parallelism  The Powerstone JPEG decoder › 4 stages  1-D DC prediction stage (DC)  Entropy Decoder (AC)  DeQuantization (DeQ)  2-D IDCT (IDCT) 9 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

10.  Single Microblaze implementations for profiling the application (execution cycles) › With extrernal DDR2 memory › With BRAM Single MB JPEG Decoding Stage DC prediction Entropy Decoding DeQuantization2D-IDCT DDR23,81%28,83%10,81%48,69% BRAM4,11%29,92%8,50%51,00% 10 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

11.  8 x 8 pixel blocks are used in every calculation stage  There is no data dependence between the blocks apart from the DC prediction stage  Architecture with only external memory  FSL depth = 4 only pointers are propagated  Architecture with use of BRAMs  FSL depth = 64 they are also used as data buffers 11 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

12. 12  One Microblaze serves as a master  FSL depth of 4  for synchronization purposes C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

13. 13  One Microblaze serves as a master  External memory  FSL depth of 4 (pointer propagation)  Internal memory  FSL depth of 64 (data propagation) C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

14. 14  Microblaze 0 serves as a master  Data equally divided between Microblaze 0 and 1 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

15.  Motivation  Architecture Platform  Design Space Exploration  Results  Conclusions C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab 15

16.  Calculations of speed up and efficiency with reference to the single Microblaze architectures  Introducing a new parameter hardware efficiency to associate the area increase of the design with the speed up 16 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

17. 17 C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

18. 18 External MemoryExternal Memory + Local BRAMLocal BRAM C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

19. 19  There is a limitation in the speed up when only external memory is used due to simultaneous memory requests of the processors  This is overcome by using both external and internal memory  The greatest speed up is achieved by the system with both data and task level parallelism, both external memory/local BRAM and 4MB (x3.27) C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

20. 20  The systems with the greatest hardware efficiency are the systems which use only internal BRAMs  The greatest HW_efficiency (3.27) is achieved by the system with 2MB and local BRAMs, followed the system with 3MB (2.8)  HW_efficiency demonstrates the revenue gained at a certain hardware cost (area) C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

21.  Motivation  Architecture Platform  Design Space Exploration  Results  Conclusions C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab 21

22. 22  A design space exploration for FPGA-based multiprocessing and memory architecture based on the JPEG algorithm  20 different system implementations with 3 different memory approaches and 4 different processor architectures were examined C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

23. 23  Higher speed ups were observed for the architectures which use both internal and external memories and have 4 processors  Higher hardware efficiencies are achieved for architectures that use only internal memories at the expense of total BRAM usage  Our goal is to formulate a methodology for an optimum MPSoC architecture selection based on the performance (speed up) and the cost- effectiveness (HW_efficiency) chosen by the user C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab

24. The research activities that led to these results, were co-financed by Hellenic Funds and by the European Regional Development Fund (ERDF) under the Hellenic National Strategic Reference Framework (ESPA) 2007- 2013, according to Contract no. MICRO2-49-project LoC. 24 Thank you very much for your attention! C.-L. Sotiropoulou – Design Space Exploration for FPGA-based Multiprocessing Systems – AUTH-eLab