FM3 Firmware Version 5.03 beta 6 (release candidate)

Status
Not open for further replies.
rockstarjazzcat said:
Is it logical? Would it really? Right now there is no hard failure when we cross the 80% threshold. Would you want the sound to cut out at the manual stated limit of 80%? Again, headroom matters.
Click to expand...
There is nothing intuitive about 80% and above CPU usage = possibly muted output. I remember when I first encountered this behavior it left me baffled. I thought something had gone wrong with the device because I was supposed to have almost 20% CPU left. There's two things at play here: what the system actually does and how it is presented to the user.

For the user the scale could show 0-100% where 100% actually extends to the 85-100% actual usage region where the system tends to fail predictably. Basically negating the concept of headroom for the user presented value. It makes no difference in how the unit behaves, just in how the overload or close to it state is presented to the end user.

It's not really relevant if we know that we are at 85.6% CPU usage if that puts us in the muted output state. That could be just presented as 100% because in that 85.6-100% range the failure state remains and you need to remove a block or adjust settings to get back to something lower where the processing is functional again.
 
laxu said:
There is nothing intuitive about 80% and above CPU usage = possibly muted output. I remember when I first encountered this behavior it left me baffled. I thought something had gone wrong with the device because I was supposed to have almost 20% CPU left. There's two things at play here: what the system actually does and how it is presented to the user.

For the user the scale could show 0-100% where 100% actually extends to the 85-100% actual usage region where the system tends to fail predictably. Basically negating the concept of headroom for the user presented value. It makes no difference in how the unit behaves, just in how the overload or close to it state is presented to the end user.

It's not really relevant if we know that we are at 85.6% CPU usage if that puts us in the muted output state. That could be just presented as 100% because in that 85.6-100% range the failure state remains and you need to remove a block or adjust settings to get back to something lower where the processing is functional again.
Click to expand...
I agree with this, however my lizard brain now knows how it works and if it's changed I'll still just always know that's how it works thus making any changes in how I see it (physically, with my eyes) kind of pointless. But I still want you to know I agree with you lol.
 
laxu said:
There is nothing intuitive about 80% and above CPU usage = possibly muted output. I remember when I first encountered this behavior it left me baffled. I thought something had gone wrong with the device because I was supposed to have almost 20% CPU left. There's two things at play here: what the system actually does and how it is presented to the user.

For the user the scale could show 0-100% where 100% actually extends to the 85-100% actual usage region where the system tends to fail predictably. Basically negating the concept of headroom for the user presented value. It makes no difference in how the unit behaves, just in how the overload or close to it state is presented to the end user.

It's not really relevant if we know that we are at 85.6% CPU usage if that puts us in the muted output state. That could be just presented as 100% because in that 85.6-100% range the failure state remains and you need to remove a block or adjust settings to get back to something lower where the processing is functional again.
Click to expand...
The thing is, how do you decide what CPU percentage to call “100%?” All that other background stuff and housekeeping — that doesn’t use a constant amount of CPU. Maybe you’ve decided that 82% equals 100%. And right now, all that background stuff is only using 14%. You could wind up with a preset that reports 104% CPU. Or a preset that goes into CPU blockage at 98%. Not so intuitive anymore. ;)

Or, you could just leave the CPU meter as it is, reporting actual CPU usage. With the knowledge that if you’re running above 80% CPU usage, something could change at any moment that would shut down your audio.
 
Rex said:
The thing is, how do you decide what CPU percentage to call “100%?” All that other background stuff and housekeeping — that doesn’t use a constant amount of CPU. Maybe you’ve decided that 82% equals 100%. And right now, all that background stuff is only using 14%. You could wind up with a preset that reports 104% CPU. Or a preset that goes into CPU blockage at 98%. Not so intuitive anymore. ;)

Or, you could just leave the CPU meter as it is, reporting actual CPU usage. With the knowledge that if you’re running above 80% CPU usage, something could change at any moment that would shut down your audio.
Click to expand...
It should not be too hard to evaluate what to call the "unsafe" range by testing block combinations and other usage. My experience is that it's about 85% and up but that might be off by a percent or two. That's why I proposed it's a range rather than a single value, you could have a reported value of 100% that still works and just know that it is at the very edge of not working. And you can't get above 100% values because you clamp the reported value down. Similarly you won't get muted output at 98% because that is always below the CPU usage that causes it.
 
ElectricPhase said:
It's just a matter of labeling, isn't it? I mean, when you attach a % symbol to 80, you are mentally programmed to assume that there's 20 left.
Click to expand...

And yet, quibblers quibble. None of the chatter about CPU limits has ANYTHING
to do with the performance. It's just moving the goalposts--a move which would
require Fractal to spend time and money on reworking it just to appease an handful
of folks who think it somehow matters.
 
laxu said:
It should not be too hard to evaluate what to call the "unsafe" range by testing block combinations and other usage. My experience is that it's about 85% and up but that might be off by a percent or two. That's why I proposed it's a range rather than a single value, you could have a reported value of 100% that still works and just know that it is at the very edge of not working. And you can't get above 100% values because you clamp the reported value down. Similarly you won't get muted output at 98% because that is always below the CPU usage that causes it.
Click to expand...
If I’m reading you right, you’re proposing that we take the highest value known to always work, and call that 100%. Is that correct? I’ve seen a preset mute at 82%. I’ve also seen an 87% preset continue to work just fine. That’s a 6% range. With your proposal, we’d shave off that top 6% and not let the user have it. Seems a high price just to make a number go all the way to 100.

I’d rather have it report the true CPU usage, as it does now
 
Last edited:
ElectricPhase said:
It's just a matter of labeling, isn't it? I mean, when you attach a % symbol to 80, you are mentally programmed to assume that there's 20 left.
Click to expand...
I don't think it is. I'm reminded of the loudness wars. Do we need a meter showing processor spikes up into the 20%? Maybe it's that I grew up with traditional VU meters, causing me to not want to see metering dumbed down.
 
rockstarjazzcat said:
Is it logical? Would it really? Right now there is no hard failure when we cross the 80% threshold. Would you want the sound to cut out at the manual stated limit of 80%? Again, headroom matters.
Click to expand...
That I wanted to say is that for me is difficult to know when cpu cuts the sound (80, 84, 88...). Is depend on the Fm3 in that moment. There could be an easier way to make things, or not. Is possible that this is the best way to interpretate the cpu limit. For me it isn't and also with every update I have the feeling that the Fm3 sucs more cpu and my pressets stop to work because Fm3 cuts the sound sometimes in some pressets and in other presets always. I set the reverb in economy and the IR in normal quality but Fm3 now cuts the sound. There a lot of people with the same problem and Fm9 is very expensive and maybe is the unic solution if you want that your device sound better updating firmwares.
 
The best thing I did to free up CPU in the FM3 was to break my addiction to having a reverb as the last effect on every modeler or profiler patch I've used for decades. The FM3 has excellent and capable delays that do a lot of what I used reverbs for in the past. And another thing; it is much cheaper and easier to get good reverb in a pedal or PA than it is to get Cygnus guitar tone.

I've sounded better in the mix with the FM3 than I ever have with the Kemper during years of use. There, I said it. And the last time I powered up the Kemper I was disappointed. Wow... Maybe the lack of reverb?
 
MKB99 said:
The best thing I did to free up CPU in the FM3 was to break my addiction to having a reverb as the last effect on every modeler or profiler patch I've used for decades.
Click to expand...
I never use reverb in a live setting. Why would I?
Every room already has it's own ambience. Throwing reverb on top of that and it starts to sound muddy in a roomful of people drinking, partying, and dancing.
I keep it tight. And some delay on the leads.
Not saying I don't like reverb. I love it when I'm here at home and start messing around with it. But for real gigs at venues? Nope.
 
MKB99 said:
The best thing I did to free up CPU in the FM3 was to break my addiction to having a reverb as the last effect on every modeler or profiler patch I've used for decades. The FM3 has excellent and capable delays that do a lot of what I used reverbs for in the past. And another thing; it is much cheaper and easier to get good reverb in a pedal or PA than it is to get Cygnus guitar tone.

I've sounded better in the mix with the FM3 than I ever have with the Kemper during years of use. There, I said it. And the last time I powered up the Kemper I was disappointed. Wow... Maybe the lack of reverb?
Click to expand...

Yup. Especially given Reverb doesn't always translate that well live, as it can clash
with the inherent reflections in whatever room we might be playing/performing in.

Edit: Robbie beat me to the CPU punch. :)
 
RICARH said:
That I wanted to say is that for me is difficult to know when cpu cuts the sound (80, 84, 88...). Is depend on the Fm3 in that moment. There could be an easier way to make things, or not. Is possible that this is the best way to interpretate the cpu limit. For me it isn't and also with every update I have the feeling that the Fm3 sucs more cpu and my pressets stop to work because Fm3 cuts the sound sometimes in some pressets and in other presets always. I set the reverb in economy and the IR in normal quality but Fm3 now cuts the sound. There a lot of people with the same problem and Fm9 is very expensive and maybe is the unic solution if you want that your device sound better updating firmwares.
Click to expand...

Have you considered that since the FM3 was released (and even since a year ago) numerous
features have been added (Ultra Res IRs, updated Cygnus modeling, numerous delays, CFNB),
and all of that for no cost to the user?

Also, you can roll back to a previous firmware version that may omit and leave out some of those
added features and save you some CPU.
 
Status
Not open for further replies.
Back
Top Bottom