Don't run CPU past 76%...

You'd think the FM3 would just start disabling effects to free up CPU instead of "blue screening". That's what my old FX8 did IIRC.
It behaves like the FX3 and FM9. When the situation is way out of control then threads starve and the system can't do anything.
 
How many controllers do you have in your preset?

I have a few presets which are guaranteed to freeze the FM3 with CPU not hitting 80%. I've reported/shared them to Fractal, but apparently those are not fixable due to many controllers used. I use Pitch followers and Envelope followers quite extensively. Some of my presets are below 80% but 100% freeze a device after an attempt to switch a scene, etc.
It is also affecting scene switching speed - when such preset is processing the sound (i.e. guitar volume knob is not 0), the scenes will not switch until I'm muting the guitar - I believe this is due Pitch followers producing a high load on CPU blocking other actions to keep the sound flowing.

Also, FM3 freezes quite often when uploading an IR from Cab-Lab 4 to such preset - no issues with presets which are not close to 80% and have less controllers. So I work on IRs using simplified presets and them transfer them to the main presets. Keep this in mind just in case.

Try removing a few controllers like Pitch followers if you use them and see if stability improves. This works for me, but I keep running the device on the edge anyway, as I'm not gigging a lot these days and can wait for the reboot :) The only bad thing is that my device sends a loud pop to the speakers when switching off and on, so when I'm plugged in directly into the speaker without volume control such as Genelec monitors, it brings some discomfort.

I'm rarely have any freeze issues with the presets over 80% if the controllers are not used.
Interesting and a bummer at the same
Time. So the controllers are an invisible culprit… I thought they were supposed to save cpu too! Off the top of my head these are features i use:

SC 1 on comp mix in different scenes
SC 2 on amp gain on different scenes
CS on out 1 level For +3dB
(CS 2 sometimes for tremolo in amp block)

So not a lot but who knows maybe too much when the preset is already at the threshold hold.

I’m going to spend time at my new limit of 76% and see how it goes. Pray for me it doesn’t drop out again tonight.
 
À few months ago I added OUT1 EQ the dedicated page of the FM3 that usually is off. I encountered the same difficulties during the seton 2 of my patches. This function eats a few % more that you do not necessarily notice in the patch views of the UI. Turning out EQ off solved the problem. I actually can run patches up to 83-85% on my unit.
Great tip! I have both out 1/2 eq enabled but only use the out 2 for personal monitors which I do add a little eq on but I’ll disable out 1. Thanks for that.
 
If you haven’t already, reducing the ‘echo density’ in the reverb block has made a big difference for me.
I already have the reverb on economy, I don’t think I can take another hit to reverb but I’ll note that and explore it as an option.
 
Interesting and a bummer at the same
Time. So the controllers are an invisible culprit… I thought they were supposed to save cpu too! Off the top of my head these are features i use:

SC 1 on comp mix in different scenes
SC 2 on amp gain on different scenes
CS on out 1 level For +3dB
(CS 2 sometimes for tremolo in amp block)

So not a lot but who knows maybe too much when the preset is already at the threshold hold.

I’m going to spend time at my new limit of 76% and see how it goes. Pray for me it doesn’t drop out again tonight.
Interesting. I can run up to 83-85pct without stalling the patches but rarely use more than 3-4 controllers. Maybe also check fast/medium/slow action of the controllers and exp pedals, and use of the preamp and preamp quality in the cab blocks that also affect cpu usage.
 
A controller on Amp Drive asks a lot of the CPU.

"Depending upon the amp model it can take a lot of CPU to calculate the Input Drive network. Some amps have simple networks that are rapidly solved. Others, like the Hook Lead and Rhythm models have complex networks that require more math. If you attach a modifier to the Input Drive it is constantly recalculating the network which increases CPU usage."

"Some amps have complicated input drive circuitry that requires a lot of calculations. Adjusting the Input Drive control can result in crackling as the time required to recalculate the network is long. Don't attach a controller to the Input Drive on these amps." [https://forum.fractalaudio.com/threads/not-a-bug-possible-bug-in-carol-ann-od-2.153162/post-1821948]

"Takes a lot of math to calculate the Drive. Not a bug." [http://forum.fractalaudio.com/threads/not-a-bug-error-sound-with-solo100-rhy-if-it´s-drive-is-controlled-with-scene-controllers.132355/#post-1566823]
 
A controller on Amp Drive asks a lot of the CPU.

"Depending upon the amp model it can take a lot of CPU to calculate the Input Drive network. Some amps have simple networks that are rapidly solved. Others, like the Hook Lead and Rhythm models have complex networks that require more math. If you attach a modifier to the Input Drive it is constantly recalculating the network which increases CPU usage."

"Some amps have complicated input drive circuitry that requires a lot of calculations. Adjusting the Input Drive control can result in crackling as the time required to recalculate the network is long. Don't attach a controller to the Input Drive on these amps." [https://forum.fractalaudio.com/threads/not-a-bug-possible-bug-in-carol-ann-od-2.153162/post-1821948]

"Takes a lot of math to calculate the Drive. Not a bug." [http://forum.fractalaudio.com/threads/not-a-bug-error-sound-with-solo100-rhy-if-it´s-drive-is-controlled-with-scene-controllers.132355/#post-1566823]
Great post. Is it right to suppose that instead of using an Ampblock with for example 4 different scene controlled settings for Amp drive value, it would require less cpu usage by using instead the same Amp in 4 different channels A/B/C/D using different settings for the drive value?
 
Depending upon the amp model it can take a lot of CPU to calculate the Input Drive network.
Oh, thank you for sharing this bit. I just replaced Shiva with Cornfed in my template and everything became three times slower - up to the point, where removing pitch controllers from AMP's EQ (I'm crazy, I know) and a few other placed didn't actually help, the scenes are not switching until I stop playing. I've shared the preset with Fractal yesterday again, but maybe this is the answer to the issues I'm having.
. Adjusting the Input Drive control can result in crackling
I had this before FW7 and stopped using followers on Amp's input and gain because of that. But that's not the case any more, I didn't hear any crackling for a long long time. Only occasional volume jumps when you play a chord and it suddenly decides that it's time to change the input drive value at some point :)
 
A controller on Amp Drive asks a lot of the CPU.

"Depending upon the amp model it can take a lot of CPU to calculate the Input Drive network. Some amps have simple networks that are rapidly solved. Others, like the Hook Lead and Rhythm models have complex networks that require more math. If you attach a modifier to the Input Drive it is constantly recalculating the network which increases CPU usage."

"Some amps have complicated input drive circuitry that requires a lot of calculations. Adjusting the Input Drive control can result in crackling as the time required to recalculate the network is long. Don't attach a controller to the Input Drive on these amps." [https://forum.fractalaudio.com/threads/not-a-bug-possible-bug-in-carol-ann-od-2.153162/post-1821948]

"Takes a lot of math to calculate the Drive. Not a bug." [http://forum.fractalaudio.com/threads/not-a-bug-error-sound-with-solo100-rhy-if-it´s-drive-is-controlled-with-scene-controllers.132355/#post-1566823]

very interesting, i guess this applies to both the gain + input trim parameter?
 
A controller on Amp Drive asks a lot of the CPU.

"Depending upon the amp model it can take a lot of CPU to calculate the Input Drive network. Some amps have simple networks that are rapidly solved. Others, like the Hook Lead and Rhythm models have complex networks that require more math. If you attach a modifier to the Input Drive it is constantly recalculating the network which increases CPU usage."

"Some amps have complicated input drive circuitry that requires a lot of calculations. Adjusting the Input Drive control can result in crackling as the time required to recalculate the network is long. Don't attach a controller to the Input Drive on these amps." [https://forum.fractalaudio.com/threads/not-a-bug-possible-bug-in-carol-ann-od-2.153162/post-1821948]

"Takes a lot of math to calculate the Drive. Not a bug." [http://forum.fractalaudio.com/threads/not-a-bug-error-sound-with-solo100-rhy-if-it´s-drive-is-controlled-with-scene-controllers.132355/#post-1566823]
Is the dirty Shirley one of these amps you think?
 
Found out the hard way tonight... I knew running presets to 80% was a bad idea so I've kept things hovering around 77-78% thinking I was safe. WRONG! Mid gig I change to my Michael Jackson preset to play "Black and White". Here comes the big opening guitar riff and... nothing! FML :mad: unit froze had to restart and of course it takes like 30-45 seconds, quite long when the band leader is looking at you saying wtf.

so moral of the story, don't go past 76% because from what I've seen on the FM3 is a fairly large spike in CPU (up to 2-3%) around whatever the baseline CPU is at. I don't experience freezes with the FM9 when it hit's 80% but I'm going to follow suit with that unit too.
I'm sorry to hear that, mate!

At least you did not experience this nasty VERY LOUD high pitched note during your show. See what happened to me yesterday during rehearsal:


My FM3 totally froze, I needed to switch it off. Last time I checked CPU was 75%. It happens right after I switch channels for the chorus. Never happened to me before.
 
I'm sorry to hear that, mate!

At least you did not experience this nasty VERY LOUD high pitched note during your show. See what happened to me yesterday during rehearsal:


My FM3 totally froze, I needed to switch it off. Last time I checked CPU was 75%. It happens right after I switch channels for the chorus. Never happened to me before.

😳😳😳
 
I was really digging your sound just before you decided to go alt saxo, @SteveFromBerlin !

Never happened to me before.
Happened to me more than once - the worst part of that is not the freeze itself, but that "Output" knobs do nothing in this case. I expected them to be "after" everything as manual says, but once the unit goes sqeak, they will not reduce the sound, only shutting the unit off helps.
 
I was really digging your sound just before you decided to go alt saxo, @SteveFromBerlin !


Happened to me more than once - the worst part of that is not the freeze itself, but that "Output" knobs do nothing in this case. I expected them to be "after" everything as manual says, but once the unit goes sqeak, they will not reduce the sound, only shutting the unit off helps.
Yes, you are right. I was really to surprised to find out about that. That's why it was taking me so long while leaning over. So it's definitely a CPU-related problem, right? I'll see what I can do to reduce the load of the patch. Still don't get why someone decides to use a scale up to 100 and you can't go over 75. That doesn't make any sense to me at all!
 
I'm sorry to hear that, mate!

At least you did not experience this nasty VERY LOUD high pitched note during your show. See what happened to me yesterday during rehearsal:


My FM3 totally froze, I needed to switch it off. Last time I checked CPU was 75%. It happens right after I switch channels for the chorus. Never happened to me before.

Oh no that’s actually really bad bro. At least it was just rehearsal. I wish the FM3 booted up quicker too. Since dyna cabs installed it seems to take longer. During my shows freeze experience that 30+ second it takes feels like the whole song.
 
Back
Top Bottom