We found out that the root cause is a missing sample-offset in the AUv3 midi message send to the host. (That playbeat 3 doesn‘t use the host‘s midi connection is a different bug report) .
In the thread you find i depth description on how to check for that error (i offered an AUM and AudioBus session) and Corne Driesprong (dev of polybeat and cycle) offered a link to a git containg the source for correct midi sessaging including the sample-offset.
We messages your AudioModern moniker several times on that thread, but you didn‘t react - that‘s why i repost the problem over here on your own forum.
Someone in the AudioBus forum thread linked above discovered a drift problem which i could confirm using his AUM session. After 200 bars, Riffer was about 1/16 off. The drift only happened in AUM and not in apeMatrix or Audiobus.
The mysterious thing is that when setting up different sync tests with Riffer versus X0X in AUM, the timing seems to be 100% okay even at 200 bars. I need to investigate further.
That‘s why i removed the initial replay with details
Could you please send the AUM dev (Jonatan Liljedahl, info@kymatica.com) a promo code for Riffer.
He offered his help to investigate the difference between the two seemingly similar AUM sessions with/without riffer drifting that i posted in the AB forum thread.
Could it be that you are checking the latest Desktop version and not the iOS version?
Version numbers for the two different systems are not always the same, especially when there are minor fixes needed for one of the two platforms.