Page 1 of 4

Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 10:49 am
by TomViolenz
So here is my set:

My live set has 15 audio tracks full of clips with rendered stems (44.1Khz 32Bit - same as set) plus 1 DrumRack track with 14 different one shots from Samplers.

In addition I have EQ8s on each track, I use several buses to route and process the audio and I also use 6 sends with normal FX on them.

CPU at idle is about 20%, running normaly at 35% and maxes out at 50% (This is on a 2012 MBP i7 Quadcore with 500GB SSD and 16 GB of RAM (both after market) running on Mavericks, Live 9.2 64Bit)

This is all good, but now to the problem:

Pretty inconsistently, but often (way too often for a live set), the CPU all the sudden spikes up to insane values (22.000% and such) the audio doesn't really crackle but basically slows to a crawl.
That problem only goes away again when I restart Live.

Interestingly OSXs ActivityMonitor does not reflect these high CPU loads it shows around 200% use (which on a 4 core with hypethreading is not that high since each core can have 100%).
Also all the other values in ActivityMonitor show nothing out of the ordinary.

One thing I noticed is that in Live the D symbol from harddisk overuse lights up some of the time when that happens, but not always.
OSXs DiskUtility showns no signs of a defective harddisk. And considering that it is an SSD, it should not have problems supplying the data of some 16-30 32Bit samples in a timely fashion.

One odd thing I noticed seems to be that this problem only happens when I use follow actions to automatically trigger the next clip, while if I manually trigger them it doesn't. Very curious!

So I'm at the end of my ideas. I would currently almost be happy if it turns out to be a faulty SSD, since they can be replaced. But as it is now I can hardly play this set to practice it. Playing with that live? Forget it! :(

Does anyone have an idea what could be at fault here, or maybe has had some similar problems in the past? Or at least some further ideas how to narrow down the cause?

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 11:04 am
by fishmonkey
getting a disk overload is definitely a problem. what model SSD is it?

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 11:08 am
by TomViolenz
Samsung SSD 840 Evo 500GB

But if it is the SSD, why does DiskUtility say everything is fine?
And if DiskUtility is not good enough to anyalize the SSD, do you know of a good (preferably free) program that is?

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 11:17 am
by fishmonkey
Disk Utility checks the directory structure and some system file permissions. it is by no means a comprehensive drive diagnosis tool.

does your SSD have the latest firmware updates?

http://www.samsung.com/au/support/model/MZ-7TE500BW

http://techreport.com/review/27212/sams ... h-old-data

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 11:20 am
by [jur]
Sometimes a clip warped with Complex/Pro can cause this kind of issue, especially with extreme transpose or stretch settings. Do you have any in your set?
I think OS X activity monitor is averaging a lot and thus don't always shows these brief peaks.

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 11:23 am
by TomViolenz
I'll check for the firmware, good call

No my warp mode it re-pitch. I use only my own stems, so I don't need to timestretch them

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 12:12 pm
by TomViolenz
Ok maybe I'm stupid, but I fail to see how to do that firmware update without burning a CD with the image.

A CD! 8O
Is this 1995?!
Oh no, I double checked it's 2015!
Ergo: I don't have any. :x
(And most Macs don't even have a drive anymore! - Though mine still does)

Surely there must be an easier way?!
Why didn't say simply supply a .dmg, like everyone else does? :x

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 12:30 pm
by Zyko1
I dont know much about mac... But couldnt it be something cpu-energy-saving related? If u use a midicontroller to trigger the clips and dont move your mouse at all while performing, maybe the system doesnt register midi-inputs (clip triggering) as an input like from the mouse, and therefore after a while suddenly cuts the cpu resources thats why it shows this strange numbers like 222000%...

Just an idea from a "outside" viewpoint. :-)

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 12:41 pm
by TomViolenz
No, happens the same with mouse and all the controllers.

Also when I use my controllers I had times before where the screen saver turned on (so looong time no interaction with the mouse), still never had the issue.

I have to say, it does sound like a hard disk issue (with the D turning on and all), so I will def make that firmware update (as soon as I manage to actually make it :x ).

I will then report back if it worked 8)

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 1:44 pm
by TomViolenz
Wow! This is slightly infuriating! :x

So Samsung is too lazy to make a OSX version of their Firmware update utility, so it has to run off some Dos hack AND the customer has to take care of making it bootable while suggesting to use tech (CDs DVDs) that most Macs don't even have anymore.

After searching the web near and far, I found you can also make a USB drive bootable, but for that it needs to be formatted or something, so you lose all previous data on it. (No, I don't tend to have empty ones laying around, just in case :roll: ).

Samsung must REALLY hate Apple (and its users)! :x

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 3:32 pm
by TomViolenz
Hm, so while I have not updated the firmware yet, I'm starting to be reasonably sure it's not the SSD. :(

I loaded all clips I played into RAM as well as all the Samplers I use. AND the problem still persists. The high CPU doesn't even go back down after play back is stopped, it jumps up and down for no apparent reason at all (And ActivityMonitor does not reflect the spikes at all) :?

Any other ideas?

Re: Puzzling issue - anyone know what's up?!

Posted: Fri Sep 18, 2015 7:32 pm
by beats me
TomViolenz wrote:Samsung must REALLY hate Apple (and its users)! :x
Well, there is the billion $ lawsuits Apple and Samsung have been firing at each other for the past 5 years or so. So there’s that. :wink:

Re: Puzzling issue - anyone know what's up?!

Posted: Sat Sep 19, 2015 2:00 am
by fishmonkey
the load meter in Live is not directly comparable to the CPU meter in Activity Monitor.

the load meter in Live is giving you an estimate of how much of the available audio processing headroom is left over. it only takes one core of your CPU to be overwhelmed before you get audio glitches.

Re: Puzzling issue - anyone know what's up?!

Posted: Sat Sep 19, 2015 1:42 pm
by zigzag
If i were you i would start seing which track is to blame, i.e. I would start disabling deleting one by one to see if you can identify which one may be causing it. You will know in a matter of minutes.
Let us know how it went

Re: Puzzling issue - anyone know what's up?!

Posted: Sat Sep 19, 2015 2:10 pm
by TomViolenz
zigzag wrote:If i were you i would start seing which track is to blame, i.e. I would start disabling deleting one by one to see if you can identify which one may be causing it. You will know in a matter of minutes.
Let us know how it went

That's what I've been doing in the back ground now for days now. There is just no real pattern discenable. (accept that follow action relation strangely - but even that does not always trigger it at the same clips/scenes)
Also I did not change this set around much at all, meaning it is still pretty similar to when it was running smoothly.

That's why I tend to think it must be hardware related.
And even though the problem still occurs, even after I put all samples into RAM, I still think it might be SSD related, since the D symbol still turns on regularly.

So I will try the firmware update mentioned above next, as soon as I pick up an empty USB thumbdrive somewhere.