Does anyone know why Ginp may give you more not less speed? I have major crosstalk issues on eci cab so if ginp ever comes along I hope I will get a bit of speed back but as ginp is another channel bearer1 wouldn't that mean the bandwidth b1 takes up would cause a speed drop?
Loonylions post explains the crux of it. G.INP is a form of error Protection. Where as the traditional form of using Reed Soloman for Error Correction caused overheads because it carries redundant data that is used to re-assemble data if some of it is lost through noise bursts etc.
Reed Soloman (RS encoding) means that redundant data is carried 100% of the time, so if the line isnt erroring then that redundant data is being transmitted taking up some of the bandwidth and reducing the available sync.
G.INP stores data in a buffer and if data is corrupted, then and only then is it re-requested. Bearer 1 only retransmits data from the buffer when needed. However unlike Error Correction it cant re-assemble corrupt data, just attempt to request and if the re-request fails then this triggers traditional CRC type errors. It all depends on the type of noise bursts but g.inp is very good in particular for REIN type bursts which are very short.
The downsides of G.INP - it doesnt cope as well for different (longer) noise bursts, so some lines would be better served with traditional Error Correction. There's also some good/short lines that dont need any error protection or error correction which wont really see any benefit.
It comes to real benefit for lines which previously needed Error Correction for REIN type noise. Their increased sync comes from giving back speed which was previously taken via RS overheads. However as already mentioned by LL, whilst those with g.inp may think they have a better sync - at times of noise then their available bandwidth does drop whilst retransmission is occurring.