Timing problems with VA Synth

Hi there!

Great to be here on the forum now. Greetings to all you MPC-User's and sorry in advance for my clumsy english ...

I'm sitting here in front of my MPC 5000 playing around with the VA Synth engine. I made a few sequences with about 4 to 5 different VA synth patches and observed some timing problems when the 5k had to play about 15 to 20 of these VA voices at the same time. Adding one or two simple MIDI drum tracks revealed serious delays in the sequencing engine. If there are about 15 to 20 VA-voices receiving their note-on event right at the start of the sequence the 5k slows down for a moment. Looking at the CPU-Meter there was a CPU-Usage of about 80 to 90 percent and a voice-count of hardly 20. Besides the whole UI slowing down, which isn't great but ok for me, I was wondering if the MPC is loosing his(her?) rock-solid timing when doing the DSP for the VA synth.

Did you made similar experiences or am I just "abusing" my 5k ...

Thanks for your statements and observations!

P.S.: My 5k is fully equipped with OS 1.02, 192 MB and CD-Drive. I was using a MPC 2500 with JJOS 2 before.

The MPC 5000 has a 20-voice VA synth and 64-voice sampler. I suspect you will have problems going beyond that. If you are running out of voices, I would resample or use the hard disk audio tracks.

Hi Askia and thanks for your reply!

My point isn't the maximum polyphony but the timing of my 5k that's getting worse when using the synth engine extensivly. Timing gets worse on my 5k when the CPU is loaded more than about 80 to 90 percent. I get that CPU load when the 5k plays about 15 to 20 voices with the synth engine at the same time. In that case some midi-events of my drum tracks are not played in time when at the same time position the synth is playing a layered chord or whatever that is using nearly the maximum synth polyphony available. It's a little hard to describe, sorry ...

My (naive) guess is that the DSP of the synth is calculated on the same CPU as the sequencing engine. In order to avoid DSP drop outs the CPU prefers the synth calculation. Pushed to the limits of the synth polyphony the senquencing engine is running out of the necessary CPU time to provide a stable timing.

Maybe anyone can reproduce that behaviour?