@@ -24,13 +24,15 @@ See the corresponding article for a theoretical analysis of the proposed truncat
## Scalability
| processors p | mean maximum time (s) | std maximum time (s) |
| ------- | -------- | ------- |
| 1 | 24.940 | 0.020 |
| 4 | 26.448 | 0.274 |
| 6 | 28.812 | 0.254 |
| 9 | 24.698 | 0.367 |
| 16 | 28.054 | 0.097 |
| 25 | 37.666 | 0.318 |
| 36 | 37.664 | 0.337 |
| 64 | 38.096 | 0.177 |
The problem size $`n`$, and number of processors $`p`$, are taken such that the number of grid points per processing unit remains constant. Specifically: $` \frac{n^2}{p} \simeq 4096^2 `$. Also, the problem sizes are of the form: $` n = q \times 2^r, \: q \in \{1, 3\} `$. The reported mean and deviation are calculated over five tests for each pair of parameters $` (p,n) `$.
| processors p | problem size n | cluster nodes c | mean maximum time (s) | std maximum time (s) |
