ReaNinjam clicking Reavisited

[the all new rob]

Not quite a year ago, I opened this thread in which I described a bug in ReaNinjam. The result of the bug is an audio glitch once per interval, which also manifests if you run DPCLAT.EXE as a test interval or two consecutive test intervals containing long DPC times.

In that thread a "resolution" was achieved by disabling hyperthreading, and at least one other frequent ninjammer confirmed the issue and fix.

So, fast forward to today....

I have a new quad-core machine. It runs all Reaper functions with alacrity, except that the ReaNinjam plugin will completely reliably cause the exact same symptoms described in my earlier thread, which were worked around on my old rig by disabling hyperthreading. Basically, an interruption to all the audio output that coincides with the start of every interval, and shows up on DPCLAT.EXE as a red bar.

The obligatory system description:
Reaper 3.13 rev 9301
Windows XP SP3, nLited
Q9550 with stock cooler
Gigabyte GA-EP45T-UD3LR motherboard
4GB OCZ Reaper DDR3 RAM
Edirol UA-25 (essentially USB1, 2 channel)

This is a new machine that I built to be a DAW, so it contains nothing unnecessary for that purpose. I updated each and every driver before returning here to reopen this bug report.

I am completely convinced that this behavior is related to ReaNinjam itself, and affects only multiprocessor machines (hence was masked by disabling hyperthreading on my prior machine). There may be other conditions necessary for it to manifest, I'm sure many ninjammers use multicore machines successfully.

I can get the Ninjam client to work fine with ReaRoute inputs from multiple Reaper tracks and ReaRoute outputs back to Reaper. I played about an hour on that configuration on this system last night with no issues, so I am comfortable that a) my system is not at issue, and b) whatever this is affects ReaNinjam but not Ninjam client.

Also as reported in the prior thread, I can play audio just fine as long as I leave the xmit box unchecked in ReaNinjam. I tested by playing along without transmitting for a considerable time and did not trigger this issue.

I'm obviously making no more than an educated guess, but it makes sense given that I can mask this by unchecking xmit that the code that wraps up the OGG file or the code that sends it to the server is the first place to look for the issue. I can make the glitches occur twice as frequently by setting the server bpi to half.

It may not affect others unless they are using a USB interface, or some other condition, but for me on this system (and the prior one with hyperthreading) it's a completely reproducible issue. I hope this description of my troubleshooting helps get it solved.

Thanks devs and community for your attention to this. I have logged well in excess of 1000 hours on ninjam and intend to continue doing so. I remain a quite contented Reaper+Ninjam user.