You are not logged in.
Currently we have two MIMO nodes each hooked up to their own computer via Ethernet. We are using the 2x2 MIMO code and streaming a video from one PC to the other using VLC per the instructions on the OFDM reference design page. However, sometimes the transmitting node just stops transmitting after a certain amount of time and will not begin to transmit unless the board is reprogrammed (hitting reset on the WARP board does not help). This can be noticed by (1) the receiver stops receiving and (2) the green LEDs on the transmitter stop blinking steadily. It is also important to note that the Ethernet LEDs on the WARP board don't blink either, which makes sense that it isnt sending any data since it is not getting any data on the Ethernet port. So I was not sure if the link between the computer and WARP board dies, and if so, if this is caused by the WARP node or by the computer. It is also worth noting that the computers which are hooked up to the WARP nodes are not great, and maybe there is a problem with computer buffering? I was just curious if you have ever experienced anything similar or have any insights on why the Ethernet link between a PC and WARP might suddenly die.
Offline
Which reference design are you using? I suspect it's not a hardware problem, but a well-hidden bad state. We've seen something similar, where a node gets stuck with an interrupt asserted but not serviced.
Offline
Thanks. We are using OFDM reference design v9...I just noticed that v10 is out. I did find another post (http://warp.rice.edu/forums/viewtopic.php?id=245) which seems like it may fix the problem. I implemented the solution, and it has seemed to work for the most part. At some point the transmitter stopped transmitting, but the Ethernet was still active; however, a push of the reset button fixed this problem, so it must have been stuck in a state. What was changed for the new v10 OFDM reference design?
Offline
Here's the reference design changelog. The big change was the migration to the 10.1.02 Xilinx tools. This required rearchitecting a lot of the design, since the PLB34/OPB were replaced by PLB46. We're close to posting ref design v11; this is a less drastic update that improves the PHY's performance (the new PHY is already committed to svn; we're just wrapping up the code that drives it).
Offline