You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It seems that using ~1.5Mbaud, with high Rx data rates can cause the applet FIFOs to overflow, and silently drop data.
1.5Mbaud is worst-case of ~150KB/s... this overflow is likely due to the relatively inefficient USB utilization that is incurred with the default auto_flush=True.
Suggestions:
Set auto_flush=False, and implement a timer to flush buffers after a number of idle bit/word-times
Increase FPGA-side FIFO size, past the default of depth=512
Add a register to indicate overflow / dropped frames
I think that this should be a non-issue for Tx data, as the buffers will back up, eventually into Python.
This was discovered by @marcan while working on Asahi Linux.
The text was updated successfully, but these errors were encountered:
It seems that using ~1.5Mbaud, with high Rx data rates can cause the applet FIFOs to overflow, and silently drop data.
1.5Mbaud is worst-case of ~150KB/s... this overflow is likely due to the relatively inefficient USB utilization that is incurred with the default
auto_flush=True
.Suggestions:
auto_flush=False
, and implement a timer to flush buffers after a number of idle bit/word-timesdepth=512
I think that this should be a non-issue for Tx data, as the buffers will back up, eventually into Python.
This was discovered by @marcan while working on Asahi Linux.
The text was updated successfully, but these errors were encountered: