home *** CD-ROM | disk | FTP | other *** search
Text File | 1997-04-03 | 1.3 KB | 40 lines | [TEXT/ttxt] |
- My test results:
-
- Small = 50x50 pixel sprite.
- Medium = 100x100 pixel sprite.
- Large = 300x300 pixel sprite.
-
- Numbers are frames/second.
-
- 6100/60 (60 MHz PPC 601), native:
-
- Method Small Medium Large
- Custom "deep" blitter 948 238 34
- CopyMask with 8-bit mask 64 18 2
- CopyMask with 1-bit mask 460 191 40
- CopyBits with region 444 212 46
- CopyBits with rect only 577 234 50
-
- Performa 630 (33 MHz 68LC040):
-
- Method Small Medium Large
- Custom "deep" blitter 548 152 20 (CodeWarrior)
- Custom "deep" blitter 698 193 24 (Think)
- CopyMask with 8-bit mask 55 15 1
- CopyMask with 1-bit mask 292 104 15
- CopyBits with region 478 203 35
- CopyBits with rect only 619 234 38
-
- The conclusion is the usual: You don't easily beat CopyBits
- on large areas. You can do it, by a little, but the only case
- I know where the difference is worth the trouble is 68040-only
- code using the MOVE16 assembly instruction. For small areas,
- however, you can gain a lot with custom blitters!
-
- On the 68k side, I tested with both the Think and CW versions.
- For the custom blitter, there was a difference, while they gave
- the same speed for all the QuickDraw-based methods. This is one
- more reason why Think remains my 68k compiler of choice.
-
- /Ingemar
-