Speed comparison

Cura is faster then Skeinforge. Really. It is.

However, Slic3r is said to be even faster. So lets test this.

I've put Slic3r 0.7.1 against Cura and Skeinforge. All with the same settings. The raw results are:

| Filename                     |Slic3r |  Cura  |Skeinforge|
| 18218/OwlReDo1_fixed_sc.stl  | 16:28 |  24:59 |  2:05:28 |
| 2577/10.stl                  |  2:26 |  13:46 |  1:06:34 |
| 2577/11.stl                  |  3:02 |   7:59 |    36:49 |
| 2577/12.stl                  |  2:12 |  11:09 |    49:57 |
| 2577/13.stl                  |  2:26 |  10:00 |    47:59 |
| 2577/14.stl                  |  2:33 |   7:41 |    39:08 |
| 2577/1.stl                   |  1:45 |  11:19 |    48:18 |
| 2577/2.stl                   |  1:46 |  10:36 |    48:45 |
| 2577/3.stl                   |  2:31 |  17:46 |  1:15:06 |
| 2577/4.stl                   |  2:37 |  14:55 |  1:15:58 |
| 2577/6.stl                   |  0:43 |   3:45 |    15:37 |
| 2577/7.stl                   |  2:12 |  19:42 |  1:21:52 |
| 2577/8.stl                   |  2:50 |  19:30 |  1:44:41 |
| 2577/9.stl                   |  2:19 |  13:36 |  1:10:22 |
| 5573/20mm-box.stl            |  0:03 |   0:44 |     1:26 |
| 5573/20mm-hollow-box.stl     |  0:03 |   0:49 |     2:12 |
| 5573/50mm-tower.stl          |  0:07 |   1:20 |     3:13 |
| 5573/BridgeTestPart.stl      |  0:09 |   0:23 |     0:20 |
| 5573/oozebane-test.stl       |  0:04 |   0:37 |     0:47 |
| 5573/overhang-test.stl       |  0:04 |   2:34 |     7:20 |
| 5573/perimeter-wt.stl        |  0:03 |   1:34 |     3:41 |
| 5573/precision-block.stl     |  0:27 |   1:19 |     3:05 |

These files are from thingiverse. The number before the filename is the thingID. Now how does this compare, and what does this tell us?

Changing the times into runtime compared to Cura:

18218/OwlReDo1_fixed_sc.stl    | 0.66 | 1.00 | 5.02
2577/10.stl                    | 0.18 | 1.00 | 4.84
2577/11.stl                    | 0.38 | 1.00 | 4.61
2577/12.stl                    | 0.20 | 1.00 | 4.48
2577/13.stl                    | 0.24 | 1.00 | 4.80
2577/14.stl                    | 0.33 | 1.00 | 5.09
2577/1.stl                     | 0.15 | 1.00 | 4.27
2577/2.stl                     | 0.17 | 1.00 | 4.60
2577/3.stl                     | 0.14 | 1.00 | 4.23
2577/4.stl                     | 0.18 | 1.00 | 5.09
2577/5.stl                     | 0.12 | 1.00 | 3.94
2577/6.stl                     | 0.19 | 1.00 | 4.16
2577/7.stl                     | 0.11 | 1.00 | 4.16
2577/8.stl                     | 0.15 | 1.00 | 5.37
2577/9.stl                     | 0.17 | 1.00 | 5.17
5573/20mm-box.stl              | 0.07 | 1.00 | 1.95
5573/20mm-hollow-box.stl       | 0.06 | 1.00 | 2.69
5573/50mm-tower.stl            | 0.09 | 1.00 | 2.41
5573/BridgeTestPart.stl        | 0.39 | 1.00 | 0.87
5573/oozebane-test.stl         | 0.11 | 1.00 | 1.27
5573/overhang-test.stl         | 0.03 | 1.00 | 2.86
5573/perimeter-wt.stl          | 0.03 | 1.00 | 2.35
5573/precision-block.stl       | 0.34 | 1.00 | 2.34

This shows that Cura is pretty consistently 4-5x faster then Skeinforge. Except with really small objects.

It also shows that in the worst case, Slic3r is still 30% faster then Cura. And in a best case it can be 10x faster. Why this huge difference? Well. Skeinforge (and thus Cura) gets slower if the object gets larger, EVEN if the amount of triangles remains the same. So a 20x20x20 box in Slic3r will be much faster then in Cura. The mini-mendel plates are mostly rectangle boxes with not that many polygons. And thus slice a lot faster in Slic3r. For an organic model, like the Owl. The difference is much less noticeable.