??? 02/23/11 17:46 Read: times |
#181264 - no Responding to: ???'s previous message |
No, I did not use bitbanding, but I don't think it would make any significant difference (not to mention incompatibility - there is no native support in the compiler so I would need to hand-code every such access).
Much of the work revolves around the following structure: typedef struct{ uint8_t x, y; union { struct { uint8_t t00 : 2; uint8_t t10 : 2; uint8_t t01 : 2; uint8_t t11 : 2; uint8_t t02 : 2; uint8_t t12 : 2; }; uint16_t tAll; }; } TTile;So, these are not individual bits. However, the same penalty when working with the 2-bit-of-N applies to the 8-bitters: they need to perform the shift and mask operations, too. Moreover, they don't have the barrel shift the ARM has and can use simultaneously with other operation (possibly the mask)! JW |
Topic | Author | Date |
Snake, as a dubious benchmark | 01/01/70 00:00 | |
especially when --- | 01/01/70 00:00 | |
mentioned in the "header" of the table | 01/01/70 00:00 | |
more data | 01/01/70 00:00 | |
Bit banding? | 01/01/70 00:00 | |
no | 01/01/70 00:00 | |
Barrel shifter | 01/01/70 00:00 |