Not a Bug (Resolved) CPU behavior has me stumped...

John K.

John K.

Member
FM9 Turbo Preset

For some reason, Scene 2 in this preset drops to around 72.8% CPU usage for me compared to the ~81ish% on the rest of the scenes.
Weirdest thing is that this even happens between Scene 2 and Scene 3 where the only difference is a Drive block turned on in Scene 3. But when I just turn on the drive block in Scene 2, it stays at 72.8%.

Can anyone test this preset out and see if they get the same thing happening? And if anyone has any idea why.... I'm stumped.


Notes:
  • I use Input 2
  • I have Out 3 > In 3 as an effects loop but in kill-dry and parallel.
 

Attachments

  • FINAL TEST.syx
    24.1 KB · Views: 1
You're probably changing channels in a block to a higher CPU algorithm somewhere. Whether the block is active or bypassed has no effect on CPU usage. It's processing the audio either way. That's the only way to get seamless on/off transitions.
 
Yep. Your pitch block is changing channels and going from Virtual Capo to Quad Chromatic. The latter uses more CPU.

If you don't need the Quad Chromatic shifter in that scene, change the Pitch block back to channel C and the CPU will stay the same as Scene 2.
 
Last edited:
mr_fender said:
Yep. Your pitch block is changing channels and going from Virtual Capo to Quad Chromatic. The latter uses more CPU.

If you don't need the Quad Chromatic shifter in that scene, change the Pitch block back to channel C and the CPU will stay the same as Scene 2.
Click to expand...
Thank you!! I was going nuts trying to find the change but I was only checking active blocks.
I knew active/bypass shouldn't affect CPU usage but I thought the same applied to channels as well.
 
John K. said:
I knew active/bypass shouldn't affect CPU usage but I thought the same applied to channels as well.
Click to expand...
You’re right. Engaged/bypassed has no effect on CPU.

If a block is engaged, it uses whatever CPU it uses. If it’s on a channel that needs more CPU, it uses more CPU. And if that block is bypassed, there is no change. If the bypassed block is on a channel that needs more CPU, it still uses more CPU.

Which is another way of saying, “Bypassing a block doesn’t change how much CPU it uses.” Which you already knew. ;)
 
Rex said:
You’re right. Engaged/bypassed has no effect on CPU.

If a block is engaged, it uses whatever CPU it uses. If it’s on a channel that needs more CPU, it uses more CPU. And if that block is bypassed, there is no change. If the bypassed block is on a channel that needs more CPU, it still uses more CPU.

Which is another way of saying, “Bypassing a block doesn’t change how much CPU it uses.” Which you already knew. ;)
Click to expand...
Yeah, I thought it was a matter of the CPU reserving whatever the highest CPU required on a block regardless of bypass or channel status.

This is interesting because now I could see some potential ways to squeeze some more effects into the same kitchen sink preset as long as I don't need two of the same higher CPU usage effects at the same time.


Thanks all for the help!
 
John K. said:
Yeah, I thought it was a matter of the CPU reserving whatever the highest CPU required on a block regardless of bypass or channel status.

This is interesting because now I could see some potential ways to squeeze some more effects into the same kitchen sink preset as long as I don't need two of the same higher CPU usage effects at the same time.


Thanks all for the help!
Click to expand...
Yes, that's a really good strategy for squeezing every last drop out of a preset...
 
unix-guy said:
Yes, that's a really good strategy for squeezing every last drop out of a preset...
Click to expand...
I'm returning to Fractal after a couple years so I don't 100% remember, but I vaguely remember/misremember there being a list of CPU usage per effect but I can't seem to find it again. I found some old one for each type of block, but not for each effect. I suppose the list grew in that time as well.
I'm really only looking for what the lowest CPU usage effect is for each block type, really. Do you know if there's a list like that somewhere?

The list I'm thinking of might have been for Helix...
 
John K. said:
I'm returning to Fractal after a couple years so I don't 100% remember, but I vaguely remember/misremember there being a list of CPU usage per effect but I can't seem to find it again. I found some old one for each type of block, but not for each effect. I suppose the list grew in that time as well.
I'm really only looking for what the lowest CPU usage effect is for each block type, really. Do you know if there's a list like that somewhere?

The list I'm thinking of might have been for Helix...
Click to expand...
There was one for Axe Fx II... And at one point maybe for Axe Fx III. But it's hard to keep current and I think it went away.

I just experiment with the blocks to see what works best.
 
John K. said:
Do you know if there's a list like that somewhere?
Click to expand...
I and @chris made this one 4 years ago, I don't know if anyone made another one more recently.
https://forum.fractalaudio.com/threads/fm3-vs-axe-fx-iii-cpu-comparison-chart.160045/

It is for axe fx and fm3 only but most blocks on the fm9 use the same amount of cpu as on the fm3 (when comparing turbo vs turbo or standard vs standard), with the exception of reverbs which run on a dedicated core on the fm9, cab2 which uses more cpu than cab1 and maybe a few other exceptions I don't recall atm.
 
Last edited:
DLC86 said:
I and @chris made this one 4 years ago, I don't know if anyone made another one more recently.
https://forum.fractalaudio.com/threads/fm3-vs-axe-fx-iii-cpu-comparison-chart.160045/

It is for axe fx and fm3 only but most blocks on the fm9 use the same amount of cpu as on the fm3 (when comparing turbo vs turbo or standard vs standard), with the exception of reverbs which run on a dedicated core on the fm9, cab2 which uses more cpu than cab1 and maybe a few other exceptions I don't recall atm.
Click to expand...
Reverb and Delay both run on dedicated cores on the FM9.
 
DLC86 said:
I and @chris made this one 4 years ago, I don't know if anyone made another one more recently.
https://forum.fractalaudio.com/threads/fm3-vs-axe-fx-iii-cpu-comparison-chart.160045/

It is for axe fx and fm3 only but most blocks on the fm9 use the same amount of cpu as on the fm3 (when comparing turbo vs turbo or standard vs standard), with the exception of reverbs which run on a dedicated core on the fm9, cab2 which uses more cpu than cab1 and maybe a few other exceptions I don't recall atm.
Click to expand...
This is great! I'll have to dig through this.

Thank you!
 
John K. said:
Yeah, I thought it was a matter of the CPU reserving whatever the highest CPU required on a block regardless of bypass or channel status.

This is interesting because now I could see some potential ways to squeeze some more effects into the same kitchen sink preset as long as I don't need two of the same higher CPU usage effects at the same time.
Click to expand...

You can definitely save some CPU by inventorying all your blocks and channels in each and defaulting them to an effect type that's the least CPU intensive when not in use.

Doing that procedure has gotten me a cumulative ~10% CPU savings over the years. I saved about 5% CPU once by changing two unused channels in my Pitch block to less intensive ones.
 
unix-guy said:
Reverb and Delay both run on dedicated cores on the FM9.
Click to expand...

Yea...it's awesome that this is the case. Offloading those functions to dedicated CPU cores really allows you to get a little crazy stuffing in all sorts of other FX blocks with impunity....with "Utra Quality" reverb...!
 
Stratoblaster said:
Yea...it's awesome that this is the case. Offloading those functions to dedicated CPU cores really allows you to get a little crazy stuffing in all sorts of other FX blocks with impunity....with "Utra Quality" reverb...!
Click to expand...
It can be a double edged sword as well. I'm sure there's technical reasons as to why all the cores can't just flex to be used for anything in the unit, but I'm using an outboard reverb and it is a bit of a shame that I can't use that DSP for other things. I do still use a reverb block in the FM9 so there's that, I suppose.
 
John K. said:
It can be a double edged sword as well. I'm sure there's technical reasons as to why all the cores can't just flex to be used for anything in the unit, but I'm using an outboard reverb and it is a bit of a shame that I can't use that DSP for other things. I do still use a reverb block in the FM9 so there's that, I suppose.
Click to expand...
Did you play around with the plex block yet? It can do ambient verbs better than the reverb block imho.
 
You must log in or register to reply here.
Back
Top Bottom