Hi Paul,
Thanks for the quick reply. To clarify, you're saying the zedboard locally can't even keep up with 1.5MSPS? So I would expect dropouts at any data rate if I process locally? And the max limit on passing data to a remote host is just 6MSPS for sustained sampling?
As far as what sample rates I used, the answer is many. I initially was trying for 12MSPS when I noticed the dropouts. From there I experimented with buffer size, sample rate, and running locally or remotely. I did find that there was a non-linear relationship between effective data rate and specified data rate. There were improvements as I increased the sample rate from the minimum up to maybe 3 or 4 MSPS at which point the effective data rate hit a maximum value (so the effective rate was the same for 4 MSPS and 61.44 MSPS). What buffer size would you recommend to hit the 6MSPS when using remote host?
I have versions of the 3.17 and 3.18 kernel, so I'll try those out to see if they fix my Tx issues.
So it seems to me that the Zedboard hardware ought to be able to perform better than 1.5MSPS or even 6MSPS. Is the processing limitation in the IIO driver or the architecture or GNURadio? What I'm getting at is would I be able to get higher performance by ditching GNURadio and writing custom applications based on the IIO driver?
Thanks,
-Austin