Block-Throttled Pipes and FP-3 Feedback

Hi folks-

FrontPanel-3 has been out for a few months now and many of you have downloaded and upgraded your boards.

Aside from a handful of support questions, mostly dealing with the migration from FP-1 (v1.4) and FP-3, we have had little feedback from customers.

We’re particularly interested in implementations involving the block-throttled pipes, but would also like to hear any performance-related comments.

I’m grabbing 16-bit samples from an A/D converter and running them into a 16x16384 FIFO. The FIFO triggers a BTPipeOUT via a fullness indicator tied to EP_READY. The fullness indicator is set for 512 words (1024 bytes).

On the host side the okBTPipeOUT is set for block=1024 bytes, length=32768 bytes. I can get a pretty much error free throughput of 1.25 MB/sec, and have about 25 ms of time left in the loop before I have to read the pipe again.

Originally I had the fullness set to 8192 words (half the FIFO), then on the host side block=1024 bytes, length=16384 bytes (half the FIFO). This resultsed in 625 kB/sec.

Should the fullness indictator always be the same as block size?

Also setting BTPipePollingInterval does not seems to change anything. I can set it from 1 ms to 100 ms and nothing changes.