./testingmlp
Initializing array made of 33554432 64-bit words (256.00 MiB).
Legend:
BandW: Implied bandwidth (assuming 64-byte cache line) in MB/s
% Eff: Effectiness of this lane count compared to the prior, as a % of ideal
Speedup: Speedup factor for this many lanes versus one lane
Applying Sattolo's algorithm... chain total: 33554432
Time to sum up the array (linear scan) 0.005 s (x 8 = 0.038 s), bandwidth = 54468.1 MB/s
Size: 33554432 (262144.00 KiB, 256.00 MiB)
---------------------------------------------------------------------
- # of lanes --- time (s) ---- BandW -- ns/hit -- % Eff -- Speedup --
---------------------------------------------------------------------
1 3.427202 598 102.1 0% 1.0
2 1.692744 1210 50.4 101% 2.0
3 1.149349 1782 34.3 96% 3.0
4 0.919902 2226 27.4 80% 3.7
5 0.791319 2588 23.6 70% 4.3
6 0.672633 3045 20.0 90% 5.1
7 0.542176 3777 16.2 136% 6.3
8 0.477443 4290 14.2 96% 7.2
9 0.423175 4840 12.6 102% 8.1
10 0.370436 5529 11.0 125% 9.3
11 0.341503 5997 10.2 86% 10.0
12 0.316246 6476 9.4 89% 10.8
13 0.294219 6961 8.8 91% 11.6
14 0.260368 7866 7.8 161% 13.2
15 0.246624 8304 7.3 79% 13.9
16 0.232484 8809 6.9 92% 14.7
17 0.220770 9277 6.6 86% 15.5
18 0.208558 9820 6.2 100% 16.4
19 0.199549 10263 5.9 82% 17.2
20 0.190742 10737 5.7 88% 18.0
21 0.181904 11259 5.4 97% 18.8
22 0.174541 11734 5.2 89% 19.6
23 0.168976 12120 5.0 73% 20.3
24 0.162494 12604 4.8 92% 21.1
25 0.156500 13086 4.7 92% 21.9
26 0.152832 13400 4.6 61% 22.4
27 0.146827 13948 4.4 106% 23.3
28 0.143676 14254 4.3 60% 23.9
29 0.147324 13901 4.4 -74% 23.3
30 0.143316 14290 4.3 82% 23.9
31 0.143506 14271 4.3 -4% 23.9
32 0.143489 14273 4.3 0% 23.9
33 0.136170 15040 4.1 168% 25.2
34 0.142730 14349 4.3 -164% 24.0
35 0.135370 15129 4.0 180% 25.3
36 0.132009 15514 3.9 89% 26.0
37 0.132157 15497 3.9 -4% 25.9
38 0.130479 15696 3.9 48% 26.3
39 0.127773 16028 3.8 81% 26.8
40 0.128949 15882 3.8 -37% 26.6
Maybe you have about 28 parallel paths?
