1. Design of a Software Correlator for the Phase I SKA Jongsoo Kim Cavendish Lab., Univ. of Cambridge & Korea Astronomy and Space Science Institute Collaborators: Paul Alexander (Univ. of Cambridge) Andrew Faulkner (Univ. of Cambridge) Bong Won Sohn (KASI, Korea) Minho Choi (KASI, Korea) Dongsu Ryu (Chungnam Nat’l Univ., Korea)

2. PrepSKA 7 th EU Framework Programme 01/04/2008 – 31/03/2011(?) 20+11 institutions 09/2009 Manchester WP2 meeting –AA, FPAA, WBSF –Power Issue –Correlator

3. Work Packages

4. Participating institutions

5. Correlators for Radio Interferometry ASIC (Application-Specific Integrated Circuit) FPGR (Field-Programmable Gate Arrays) Software (high level-languages, e.g., C/C++) –Rapid development –Expandability –…

6. Current Status of SC LBA (Australian Long Baseline Array) –8 antennas (Parkes, … 22-64m, 1.4-22GHz) –DiFX software correlator (2006; Deller et al. 2007) VLBA (Very Long Baseline Array) –10 antennas (25m, 330MHz - 86GHz) –DiFX (2009) GMRT (Giant Metrewave Radio Telescope) –30 antennas (45m, 50MHz-1.5GHz), 32MHz –ASIC  software correlator (Roy et al. 2010)

7. Current Status of SC (cont.) LOFAR (Low Frequency Array) –LBA (Low Band Antennae) 10-90MHz –HBA (High Band Antennae) 110 – 250MHz –IBM BlueGene/P: software correlation

8. SKA Timeline

9. Phase I SKA (2013-2018) 10% of the final collecting area ~300 dishes software correlator

10. Correlation Theorem, FX-correlator F-step (FT): ~log 2 (N c ) operations per sample X-step (CMAC): ~ N operations per sample

11. FLOPS of the X-step in FX correlator 4 is from 8 is from 4 multiplications and 4 additions: N(N+1)/2 is the number of auto- and cross- correlations with antenna N if B (bandwidth) = 4 GHz N=2  96x4 GFLOPS = 384 GFLOPS N=100  16x4x10 4 GFLOPS = 640 TFLOPS N=300  16x9x4x10 4 GFLOPS = 5.76 PFLOPS N=3000  16x9x4x10 6 GFLOPS = 576 PFLOPS

12. Top500 supercomputer

13. Fermi Streaming Multiprocessor Fermi features several major innovations: 512 CUDA cores NVIDIA Parallel DataCache technology NVIDIA GigaThread™ engine ECC support –Performance: > 2 Tflops

14. Design goals Connect antennas and computer nodes with simple network topology Use future technology development of HPC clusters Simplify programming

15. CPUs+(GPUs) Required BW>512Gb/s =64GB/s 100 Gb/s Ethernet Software Correlator for Phase I SKA (‘2018) 2x4x2x4GHz =64Gb/s 2 pols, 4bit sampling, Nyquist, BW 300 nodes >20 TFLOPS

16. Communication between Computer Nodes I Node 1 FFT NcNc 1 2 after FFT NcNc 2x4x2xB 2x32x2xB Node 2 FFT

17. Communication between Computer Nodes II Node 1 FFT 1 2 after communication N c /2 1 2 1 2 2x4x2xB 2x32x2xB Node 2 FFT N c /2

18. All-to-All Communication between Computer Nodes Node 1 FFT 1 2 after communication N c /4 1 2 2x4x2xB 2x32x2xB Node 2 FFT N c /4 Node 2 FFT Node 2 FFT 3 4 3 4 1 2 3 4 N>>1 BW(interconnect) =8 x 2x4x2xB =128B

19. Data Flows GPU, CMAC memory CPU, FFT 4  32bits bit/sample GPU Memory BW: 102GB/s (C1060) threads PCI-e BW: 8GB/s (gen2 16x) GPU, CMAC memory CPU, FFT 4  32bits bit/sample threads Node Interconnect BW: 40Gb/s (Infiniband)

20. AI (Arithmetic Intensity) Definition: number of operations (flops) per byte AI = 8flops/16bytes (R i,R j ) = 0.5 AI = 32 flops/32bytes (R i, L i, R j, L j ) = 1.0 for 1x1 tile AI = 2.4 for 3x2 tiles Since AIs are small numbers, correlation calculations are bounded by the memory bandwidth. Performance: AI x memory BW (=102GB/s) Wayth+ (2009), van Nieuwpoort+ (2009)

21. Measured bandwidth of host-to-device (Tesla C1060) ~70% of PCI -e2 bandwidth

22. Performance of Tesla C1060 as a function of AI Performance is, indeed, memory-bounded. Maximum performance is about 1/3 of the peak performance.

23. AI for host-device and host-host AI = (N+1) FLOP/byte N x 16 bytes (R,L) = 16 N byte 4x8xN(N+1)/2 FLOP PCI bus bandwidth PCI-e 2.0: 8.0GB/s (15 Jan. 2007) PCI-e 3.0: 16.0GB/s (2Q 2010, 2011) PCI-e 4.0: 32.0GB/s (201?) PCI-e 5.0: 64.0GB/s (201?) Performance [GFLOPS] = PCI BW [GB/s] x AI [FLOP/B]

24. Benchmark: Performance 10GF 100GF 40Gb/s IB

25. Expected performance bounded by the BWs of PCI bus and interconnect

26. Power usage and Costing Computer nodes –1.4 KW, 4 K Euro for each server including 2x0.236 KW (2 GPUs) –0.4 MW, 1.2 M Euro for 300 servers Network Switches –3.8 KW for IB (40Gb) 328 ports

27. Technology Development in 2010 2Q 2010, (2011): PCI-e 3 rd Generation 2Q (April) 2010: Nvida Fermi (512 cores, L1,L2 cache) (March 29) 2010: AMD 12 core Opteron 6100 processor (March 30) 2010: Intel 8 core Nehaem-EX Xeon processor

28. Infiniband Roadmap (IBTA)

29. CPUs+(GPUs) Required BW>512Gb/s =64GB/s 100 Gb/s Ethernet Software Correlator for Phase I SKA (‘2018) 2x4x2x4GHz =64Gb/s 2 pols, 4bit sampling, Nyquist, BW 300 nodes >20 TFLOPS