I’m running GP on a headless desktop : i7-6700k, 12gb ram, RME Firerface 400, and tiny WiFi router for rdp control from laptop/tablet/smartphone.
I’m on win 10 LTSC with ultralite and optimized install.
All is working like a charm with 2 instances (one for each keyboard) running b3x, pianoteq, arturia synths… I’m at about 20-30% on two of the four cores, with 128 samples buffer @ 48khz. Nothing higher than 400us on latencymon when GP is running. No cracks…
No cracks except when I launch my gig file when GP is started. When plugins are loading there are some rattles. Were it possible to start the sound engine once all the plugins are loaded ? Do it automatically ?
I notices that there is almost no rattling when I connect RDP from Android (Microsoft rdp client from playstore) and Linux (remmina, freerdp) devices… Seems more frequent with the win RDP app from my win 10 laptop.
Any idea ?
Another little thing : I use some pad presets with long release times. When I switch back and forth rackspaces, I can hear the tail of the pad sound. Is it possible to ‘‘cut the sound tail’’ when I switch to another rackspace and not hearing it when I switch back ?
I’ve disabled the ‘‘audio recording redirection’’ of local group policiers (on the headless pc, not the laptop). Changes nothing, still 2 or 3 loud cracks during the plugins loading.
I also tried with my Android smartphone. Actually I noticed it happen with the phone too, but it is only when I start GP and load my gig for the first time after reboot. Once it has been launched one time, no more cracks on next loadings. Same behavior on win10 and Android.
Concerning the audio tails I tried to fiddle with input muting and output fading with no really success. When I play a pad chord and switch quickly to another rackspace, the sounds fades out, ok, but even after 10-15sec although it’s dead quiet, when I switch back to the pad rackspace, I clearly hear again the pad tail.
Ok I’ve fiddled with audio params of the Firerface (unchecked all useless input/output, restricted fw bandwidth to analog signals). I’ve rebooted the pc a lot of time, checked with win10 rdp, Android rdp…
I’ve also restricted GP to 2 of the 4 cores with Process Lasso, optimized PSU performance…
Glitches at the first launch are gone !
Now the remaining concern : audio tails!
In fact the phenomenon happens when I use a bypass widget to turn off a plugin.
For instance, I’ve a rackspace with piano and strings layered. Two widjets buttons are bypassing bith plugins. When I bypass the string as sound is still in the release phase, it is cut. But when I reactivate the plugins, I can hear the remaining sound tail that was cut at bypass. Even if it’s 2 minutes later !
Is there a trick to mute and de-mute audio when bypassing/reactivating plugin ?
Still on my audio glitches problem at GP startup.
I’ve still sometimes glitches when I launch a gig file. When it happens (only while loading the project), I can play rackspaces without a flaw. But as soon as I open a second instance, I’ve loud cracks staying all the time (it’s like 20hz continuons pulses).
I have to close both instances and when I launch GP again, all is always working nice.
These glitches only happen at the first time I launch GP after PC booting. Sometimes it happens, Sometimes it doesn’t… But once GP has been launched and closed one time, it never appears again.
CPU is 6700k with 12GB RAM (8+4)
My soundcard is an RME FF400. It works well at 48k with a 128 buffer (3 instances without any glitch).
I’ve tried many things without consistent results :
Windows sound parameters, untick all shitty options (effects, exclusive mode…)
Reducing FW bandwidth (enabling only analog channels) in RME control panel
Enabling MMCSS in RME CP
Enabling/disabling gig autolaunch in GP
Enabling only used analog channels on GP
The last thing could be the RAM… I’ve read that it’s not recommended to mix different size of RAM.
I’ve only two slots of RAM on the mb (HP ProDesk 400 G3)
Any advice ?
But overall, appart from these weird glitches, GP is very stable and playable.
I’d move to 44.1k and 192 samples. You cant tell the difference between 44.1k and 48k during live performance, and you’ll save yourself some CPU cycles. 192 samples shouldn’t affect your realtime playing either.