Right now, I need to change to the settings page to see the CPU usage.
It would be more usable to see this always somewhere on the screen.
Other Apps, for example, feature just a small line that gets larger and shorter, this way it does not consume much space.
Comments
Yeah I was thinking about something (colour line or even a dot) above the play button to let you know - just something to know how close we’re getting to the danger zone lol. Nothing too in detail as the meters in these things only work as a guide and are never 100% accurate for knowing when the dreaded crackles will hit!
I find I rarely check the CPU meter. Everything seems really stable. I’m impressed by that! Maybe CPU meter is a bigger issue on older iPads? Maybe there could be a user defined toggle for a floating CPU meter?
There are many AU Synth, where already 2 instances may reach the CPU limit!
From what you wrote, i get the impression that you stay in NS2 only, without AU Synths. This is not comparable, i think.
What dendy wrote
Are there any stats (or should we start some) on how many of which AUv3 on what devices causes problems? It may be useful one day to have a chart that shows what users should expect based off of hardware & app combinations. I don't have the skills to collate that sort of info. Just a thought really.
You are right - not comparable. I’m using both internal & external. Sometimes more of one than the other, somtimes external keyboards. I haven’t noticed a pattern yet, but I think some AU are bigger CPU hogs than others. 6-7 AU + internal sounds seems to be working for me, but I probably am just doing simple things and not complex projects like some people here.
My impression of CPU metering is it’s not a good thing to be obsessing over. I don’t think it’s very accurate, and the only real measure of when things are overloaded is when you start to get audio dropouts.
I’m perfectly happy to have it somewhat buried so I’m not paying attention to it until I need to. It’s a musical instrument, not an airplane. I’d rather not be distracted by the dials.
I realize I’m probably largely alone in that opinion.
No, you're right. It's false accuracy too, because the iOS device can do dynamic CPU throttling without the app knowing about it (which can result in false alarms: having more CPU reserves than you think or CPU peaks being less of a problem than they appear).
It’s funny at 34% cpu I experienced some slight crackling. Only because I had two active signals with the AU Replicant 2 on it. My first time lol. I even had a song with 20 tracks and this never happened.. thank goodness we can change the buffering on here. It helped a lot. Thanks for that Matt!
No you are not
Because of what @brambos said, information value of this meter is low.
Check just for fun this project
https://www.dropbox.com/s/pux5sjgtujf249c/TestLoad.nsa?dl=0
There is 32 Obsidian tracks with reverb on each one. It starts play 8 notes, then after 1 bard second track starts play, etc sec..
Now, hit play and watch CPU meter, it's funny. It starts rise, but at some point it stops (at my iPad AIR 1 with buffer set to 11ms it's around 50%) and almost whole time, as more and more obsidians and reverbs are added to project, it STILL oscillates around 50%. Like they are running for free
))) Only before end it jumps to 60-70%
Bottom of line - isn't that insane ? 32x 8 voices + 32 reverbs on A7 CPU ?!?
I don't think people still
really understand how much efficient is this app :-)
@dendy Try the same with Model D or Zeeon
Yes, I've encountered this too and it relates to what Bram said. iOS dynamically changes the CPU clock rate based upon the load, but often the buffer has to overrun once before iOS thinks, 'crap, I better up my game!'.
If you can get away with a little more latency then increasing the buffer size usually solves the problem by giving it more headroom (in time) to handle the odd CPU peak. I'd recommend the lowest setting when you're using it for live performance with only a few instruments. Medium's a good compromise, and High gives the best battery life if you're OK with the extra latency.
Also as Bram said, the CPU % on the meter can be misleading as as it often goes down when you ask it to do more - again, this is because iOS has stepped it up a gear.
Still, it's a lot better than the early iOS 9 days. Back then iOS never looked at the audio thread when considering the CPU load and many devices had a processor which was either clocked at 50 or 100%. As a result, nearly every audio app in the store used only 50% of the CPU's actual capability.
@Blip Interactive
Wait, buffer size affects battery life ? Why ?
I'll just add that the latency on these devices is incredibly low.
The "high" latency setting in NS2 is ~10ms.
By comparison "high" on my desktop ASIO driver is probably about 40ms.
The fact that an iPad can pull off latency at that low of a level is incredible (although, with crackling, maybe "pull off" isn't exactly the best way of saying it).
If you can get by on medium or high latency then I'd suggest that for sure. I hate crackling/clicks/pops. Ruins my music making mojo almost as much as slow load times and the latency added would be in the 2ms range per step. Negligible IMO.
Higher CPU usage. Those little squirrels eat more the faster they have to run.
I’m working on my “learn NS2” song so I’m only using native NS2 plugins, no AU’s, and it’s got 30+ tracks... like 5 string parts, 4 pad parts, 5 to 6 lead’s, 3 or 4 basses, 3 slates, 5 or so sends with tons of processing on the tracks and I can’t get my 2017 13” iPad Pro to go over 25% no matter what latency I set it to. And I write dense music, there’s lots of stuff going on. I just can’t get over it...
...it’s baffling how efficient it is.
fucking genial explanation... made me laugh loudly
Looking forward for that song !