Details | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 1806 | yogev_ezra | 1 | |
| 2 | -------------------------------- |
||
| 3 | |||
| 4 | |||
| 5 | the the best dirtyrect candidates for merging. While |
||
| 6 | searching, it looks out for perfect or sufficiently |
||
| 7 | good candidates. |
||
| 8 | |||
| 9 | |||
| 10 | The merged rectangle is of the same size |
||
| 11 | as the largest of the two input rectangles: |
||
| 12 | |||
| 13 | |||
| 14 | |||
| 15 | |||
| 16 | |||
| 17 | |||
| 18 | Not Perfect, but good enough to be treated |
||
| 19 | as such, considering the cost of going on |
||
| 20 | searching for a better candidate: |
||
| 21 | Amerged 100 / MAX(A1, A2) - 100 <= PIG_INSTANT_MERGE |
||
| 22 | |||
| 23 | |||
| 24 | no more than PIG_INSTANT_MERGE % bigger than |
||
| 25 | the area of the larger of the two input rects.) |
||
| 26 | |||
| 27 | |||
| 28 | that thereis* a better candidate. Assuming |
||
| 29 | that PIG_INSTANT_MERGE is set to a sensible |
||
| 30 | value, it is not very likely at all. There |
||
| 31 | would have to be another dirtyrect nearby, and |
||
| 32 | the chance of that being a better match is |
||
| 33 | rather low, since that would most likely have |
||
| 34 | caused it to be merged with the tested dirtyrect |
||
| 35 | long before our new rect came in. |
||
| 36 | |||
| 37 | |||
| 38 | The area of the merged rectangle is smaller |
||
| 39 | than the total area of the two input rectangles: |
||
| 40 | |||
| 41 | |||
| 42 | |||
| 43 | |||
| 44 | |||
| 45 | |||
| 46 | The area of the merged rectangle is larger |
||
| 47 | than the total of the two input rectangles, but |
||
| 48 | since there is some per-rectangle overhead, |
||
| 49 | merging is still a win: |
||
| 50 | |||
| 51 | |||
| 52 | |||
| 53 | |||
| 54 | rectangle is in the range of 300 pixels. One |
||
| 55 | should probably benchmark a few different systems |
||
| 56 | to see if that's reasonable. |
||
| 57 | |||
| 58 | |||
| 59 | The area of the merged rectangle is larger than |
||
| 60 | the total of the input rectangles to the degree |
||
| 61 | that merging is a definite loss: |
||
| 62 | |||
| 63 | |||
| 64 | |||
| 65 | |||
| 66 | solutions. If there are only Good and Acceptable |
||
| 67 | candidates, the best one (lowest number of wasted |
||
| 68 | pixels) is the solution. |
||
| 69 | |||
| 70 | |||
| 71 | sensible merger solution, so pig_dirty() will try to add |
||
| 72 | a new dirtyrect. |
||
| 73 | |||
| 74 | |||
| 75 | even though it would have been Unacceptable under normal |
||
| 76 | circumstances. |
||
| 77 | |||
| 78 | |||
| 79 | |||
| 80 | TODO: "don't merge"! For example, it might pay off to |
||
| 81 | TODO: detect overlapping and clip dirtyrects to avoid |
||
| 82 | TODO: it, when merging is not a viable option. |
||
| 83 |