FM Radio Benchmark Summary

This file details a summary of how the numbers were obtained for this benchmark.

Benchmark Description lines of code # of constructs in the program Number of filters in the expanded graph

filters pipelines splitjoins feedbackloops

Radio FM Radio with an equalizer 525 14 6 4 0 26

Benchmark 250 MHz RAW processor C on a 2.2 GHz Intel Pentium IV

StreamIt on 16 tiles C on a single tile

Utilization # of tiles used MFLOPS Throughput (per 10⁵ cycles) Throughput (per 10⁵ cycles) Throughput (per 10⁵ cycles)

Radio 73% 16 421 53.9 8.85 14.1

Benchmark	Description	lines of code	# of constructs in the program	Number of filters in the expanded graph
filters	pipelines	splitjoins	feedbackloops
Radio	FM Radio with an equalizer	525	14	6	4	0	26

Benchmark	250 MHz RAW processor	C on a 2.2 GHz Intel Pentium IV
StreamIt on 16 tiles	C on a single tile
Utilization	# of tiles used	MFLOPS	Throughput (per 10⁵ cycles)	Throughput (per 10⁵ cycles)	Throughput (per 10⁵ cycles)
Radio	73%	16	421	53.9	8.85	14.1

StreamIt Number Calculations

Each iteration of the steady state schedule for the StreamIt code produces 1 output. Utilization numbers reported were 116157 useful cycles / 160000 total cycles = 0.726.

flops reported by RAW's cycle-accurate simulator are 2970 per steady state iteration, which is (2970 flops/1858 cycles) * 250 million cylces/second = 399.62325 MFLOPS.

Each steady-state iteration takes 1858 cycles.

1 outputs every 1858 cycles, normalized to 10^5 cycles results in a throughput of 1*(100000/1858) = 53.8213 outputs every 10^5 cycles.

C Code Single Tile Raw Number Calculations

first iteration done at 0xdf95c (915804 cycles)
second iteration done at 0xe9813 (956435 cycles) (delta=40631)
third iteration done at 0xf3684 (996996 cycles) (delta=40561)
Based on these cycle counts, each iteration takes 40631 first/40561 second cycles (average = 40596).

1 outputs every 40596 cycles normalized to 10^5 cycles, 1*(100000/40596) = 2.4633 outputs / 10^5 cycles.

flops reported are 4752 flops, which is (4752 flops/40596 cycles) * 250 million cycles/second = 29.2639 MFLOPS.

Utilization numbers reported were 635369 useful cycles/ 650096 total cycles = 0.97734642

C Code Number Calculations

Performing 10 million iterations at 1 outputs/iteration.
Runtime for 10^7 iterations is 94.11 seconds.

Number of cycles per iteration: 10^7 iterations/ 94.11 second * 1 outputs / 1 iteration * 1 second / 2.2*10^9 cycles * 10^5 cycles = 4.82994 outputs / 10^5 cycles.