Changeset 1967

Show
Ignore:
Timestamp:
03/21/11 05:43:06 (1 year ago)
Author:
jwoithe
Message:

RME: more streaming stability fixes. Further extended testing is needed across all sample rates, but initial indications are that the streaming system is finally approaching a state of stability.

Files:

Legend:

Unmodified
Added
Removed
Modified
Copied
Moved

  • trunk/libffado/config.h.in

    r1962 r1967  
    234234// given by RME_TRANSMIT_TRANSFER_DELAY (in ticks), but we can send 
    235235// packets early if we want to. 
    236 #define RME_MAX_CYCLES_TO_TRANSMIT_EARLY                  2 
     236#define RME_MAX_CYCLES_TO_TRANSMIT_EARLY                  1 
    237237 
    238238// ensure that the RME tx SP clips all float values to [-1.0..1.0] 

  • trunk/libffado/src/libstreaming/rme/RmeReceiveStreamProcessor.cpp

    r1966 r1967  
    120120    rxdll_e2 = (TICKS_PER_SECOND*1.0) / ((float)m_Parent.getDeviceManager().getStreamProcessorManager().getNominalRate()); 
    121121//w = (2*M_PI*0.004); 
    122 //w = (2*M_PI*0.00225); 
    123 w = (2*M_PI*0.002); 
     122w = (2*M_PI*0.0025); 
     123 
    124124    rxdll_B = (sqrt(2.0)*w); 
    125125    rxdll_C = (w*w); 
    …… 
    135135// therefore avoid drifting too far away from the "smoothing" DLL.  Again, 
    136136// this value has been determined experimentally. 
    137 m_dll_bandwidth_hz = 10.0; 
     137//m_dll_bandwidth_hz = 10.0; 
     138m_dll_bandwidth_hz = 3.0; 
    138139    return true; 
    139140} 
    …… 
    184185        if (e < -64LL*TICKS_PER_SECOND) 
    185186          e += 128LL*TICKS_PER_SECOND; 
    186 //        if (e < 0) 
    187 //          e += 128LL*TICKS_PER_SECOND; 
    188187 
    189188// Very large e values indicate a discontinuity in processing, possibly due