Massively LDPC Decoding on Multicore Architectures

1. Massively LDPC Decoding onMulticore Architectures Present by : fakewen

2. Authors • Gabriel Falcao • Leonel Sousa • Vitor Silva

3. Outline • Introduction • BELIEF PROPAGATION • DATA STRUCTURES AND PARALLEL COMPUTING MODELS • PARALLELIZING THE KERNELS EXECUTION • EXPERIMENTAL RESULTS

4. Outline • Introduction • BELIEF PROPAGATION • DATA STRUCTURES AND PARALLEL COMPUTING MODELS • PARALLELIZING THE KERNELS EXECUTION • EXPERIMENTAL RESULTS

5. Introduction • LDPC decoding on multicore architectures • LDPC decoders were developed on recent multicores, such as off-the-shelf general-purpose x86 processors, Graphics Processing Units (GPUs), and the CELL Broadband Engine (CELL/B.E.).

6. Outline • Introduction • BELIEF PROPAGATION • DATA STRUCTURES AND PARALLEL COMPUTING MODELS • PARALLELIZING THE KERNELS EXECUTION • EXPERIMENTAL RESULTS

7. BELIEF PROPAGATION • Belief propagation, also known as the SPA, is an iterative algorithm for the computation of joint probabilities

8. LDPC Decoding • exploit probabilistic relationships between nodes imposed by parity-check conditions that allow inferring the most likely transmitted codeword.

9. LDPC Decoding(cont.) White Gaussian noise

10. LDPC Decoding(cont.)

13. Outline • Introduction • BELIEF PROPAGATION • DATA STRUCTURES AND PARALLEL COMPUTING MODELS • PARALLELIZING THE KERNELS EXECUTION • EXPERIMENTAL RESULTS

14. DATA STRUCTURES AND PARALLEL COMPUTING MODELS • compact data structures to representthe H matrix

15. Data Structures • separately code the information about H in two independentdata streams, and

16. remind • rmn :是CNm->BNn • qnm :是BNn->CNm

17. Parallel Computational Models • Parallel Features of the General-PurposeMulticores • Parallel Features of the GPU • Parallel Features of the CELL/B.E.

18. Parallel Features of the General-PurposeMulticores • #pragma omp parallel for

19. Parallel Features of the GPU

20. Throughput

21. Parallel Features of the CELL/B.E.

22. Throughput

23. Outline • Introduction • BELIEF PROPAGATION • DATA STRUCTURES AND PARALLEL COMPUTING MODELS • PARALLELIZING THE KERNELS EXECUTION • EXPERIMENTAL RESULTS

24. PARALLELIZING THE KERNELS EXECUTION • The Multicores Using OpenMP • The GPU Using CUDA • The CELL/B.E.

25. The Multicores Using OpenMP

26. The GPU Using CUDA • Programming the Grid Using a Thread per Node Approach

27. The GPU Using CUDA(cont.) • Coalesced Memory Accesses

29. The CELL/B.E. • Small Single-SPE Model(AB C) • Large Single-SPE Model

30. Why Single-SPE Model • In the single-SPE model, the number of communications between PPE and SPEs is minimum and the PPE is relieved from the costly task of reorganizing data (sorting procedure in Algorithm 4) between data transfers to the SPE.

31. Outline • Introduction • BELIEF PROPAGATION • DATA STRUCTURES AND PARALLEL COMPUTING MODELS • PARALLELIZING THE KERNELS EXECUTION • EXPERIMENTAL RESULTS

33. EXPERIMENTAL RESULTS • LDPC Decoding on the General-Purpose x86 Multicores Using OpenMP • LDPC Decoding on the CELL/B.E. • Small Single-SPE Model • Large Single-SPE Model • LDPC Decoding on the GPU Using CUDA

34. LDPC Decoding on the General-Purpose x86 Multicores Using OpenMP

35. LDPC Decoding on the CELL/B.E.

36. LDPC Decoding on the CELL/B.E.(cont.)

37. LDPC Decoding on the CELL/B.E.(cont.)

38. LDPC Decoding on the GPU Using CUDA

39. The end Thank you~