4 - Priority scale: High, Medium and Low
6 - Complexity scale: C1, C2, C4 and C8. The complexity scale is exponential,
7 with complexity 1 being the lowest complexity. Complexity is a function
8 of both task 'complexity' and task 'scope'.
10 The general rule of thumb is that a complexity 1 task should take 1-2 weeks
11 for a person very familiar with oFono codebase. Higher complexity tasks
12 require more time and have higher uncertainty.
14 Higher complexity tasks should be refined into several lower complexity tasks
15 once the task is better understood.
20 - Currently, only the decoder is optimized to take advantage of advanced
21 processor instruction sets. In use cases like HFP 1.6, optimizing the
22 encoder will bring a significant latency, power and performance advantage.
23 For example, the MMX encoder is 3 to 6 time faster than the SIMD encoder.
28 - Use a log2 table for byte integer scale factors calculation (sum log2
29 results for high and low bytes) fill bitpool by 16 bits instead of one
30 at a time in bits allocation/bitpool generation port to the dsp
35 - The neon optimization code is split in two parts. Sample reordering and
36 blocks encoding. There is a neon optimization for encoding SBC. But mSBC
37 is not supported by this optimizer because the reordering has been
38 specifically for mSBC.
46 - The decoder is optimized to take advantage of advanced processor
47 instruction sets. Currently implemented are MMX, arm neon, arm v6
48 and iwmmxt. SSE3 is available since almost 10 years now, on a large
49 range of Intel processors. It should be interesting to implement it
50 and to compare with MMX implementation on Intel processors.