Added optimizations for several ADD fast paths:
- add_8_8
- add_8_8_8
- add_n_8_8
- add_n_8_8888
- add_8888_8888
- add_8888_8888_8888
- add_0565_8_0565
- add_8888_8_8888
- add_8888_n_8888
Benchmark results (lowlevel-blt-bench) on Malta board (@1Ghz) are
included in the log messages.
Any comments to these patches are welcome.