|
|
|
01-03-2019, 08:03 AM
|
#1
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Reaper Performance Thread
After many many many years of using Reaper it seems to me that I have not learned enough about setting it up to use my computer. I'm just starting to read more about it but already I'm seeing huge improvements. I was freezing everything to free up cpu but it seems my i980x 6 core has more horsepower than I have been using. Here are just a few tweaks that have already gotten rid of crackling I was getting when I armed tracks for recording.
I would like to see a thread that is dedicated to performance tweaks that could even become STICKY. Feel free to post them all here. If anyone can make a glossary of terms that are common to these issues that would be great. Things like.
Crackling & Popping.
Skipping & Stuttering
etc.......
I don't know if there are any side effects of my new settings but they got rid of the crackling and dropped RT CPU a lot !
I find that if I ARM more than 4 tracks the crackling comes back though.
Unarming all tracks gives me this..... Rearming one track puts RT CPU at 36%
Rearming a single track with an FX on it pushes RT CPU to 67% and crackling is back. That's just arming one track during playback.
Last edited by Coachz; 01-03-2019 at 08:16 AM.
|
|
|
01-03-2019, 09:22 AM
|
#2
|
Human being with feelings
Join Date: Jul 2009
Posts: 7,570
|
I would not enable 12 cores for live multiprocessing. Try 4 and some taken off from the audio processing threads option at the top.
on my 8-core i7 I use 6 and 2 for these options and it works pretty well. Seen other people getting good results with similar settings.
|
|
|
01-03-2019, 09:32 AM
|
#3
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Thanks for the reply. So you would disable the auto-detect the number of audio processing threads and set it to ??? And also set live to 4.
|
|
|
01-03-2019, 10:25 AM
|
#4
|
Human being with feelings
Join Date: Jan 2017
Location: Los Angeles
Posts: 1,161
|
Can you add some more details such as: audio device, driver, buffer size, sample rate? Please.
Last edited by ErBird; 01-03-2019 at 10:44 AM.
|
|
|
01-03-2019, 10:30 AM
|
#5
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Soundcard: RME 9652
to Digimax FS interface using analog cables looping DAC outputs back to inputs
@512 RTL = 24.974ms ( 1101 samples)
@256 RTL = 13.355ms (589 samples)
@128 RTL = 7.540ms (332 samples)
@ 64 RTL = 4.683ms (205 samples)
@ 32 RTL = 3.193ms (141 samples)
44.1khz sample rate
I run 128 buffer.
I'm playing back about 45 tracks of audio. Would an SSD drive help for the Reaper Project drive ? I record wav 24 bit pcm.
CPU is i980x and 12gb ram
Crackles are happening above RT CPU of 55%
|
|
|
01-03-2019, 12:17 PM
|
#6
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
This is a good topic, I would also like to know how some are running large 100+ track sessions with lots of vsts. I have followed the usual audio settings for Win10 such as disabling c-states/onboard audio/removed excess apps, but all of my Reaper audio settings are pretty much default.
|
|
|
01-04-2019, 12:51 AM
|
#7
|
Human being with feelings
Join Date: Aug 2015
Location: Florence, Italy
Posts: 463
|
Quote:
Originally Posted by EpicSounds
I would not enable 12 cores for live multiprocessing. Try 4 and some taken off from the audio processing threads option at the top.
on my 8-core i7 I use 6 and 2 for these options and it works pretty well. Seen other people getting good results with similar settings.
|
I have an 8-core too and I enabled 5 core on FX live multiprocessing but I didn't think of unchecking the multiprocessing auto-detect on top of this page. I'll try it out.
I usally use 256 samples but with a track I'm making (lots of automation, a huge amount of it) I had to turn it up to 512 to have a smoother playing.
I have an Audient iD22.
|
|
|
01-04-2019, 03:46 AM
|
#8
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
512 on mine would be 24.974ms and very audible for recording.
|
|
|
01-04-2019, 03:54 AM
|
#9
|
Human being with feelings
Join Date: Aug 2015
Location: Florence, Italy
Posts: 463
|
Quote:
Originally Posted by Coachz
512 on mine would be 24.974ms and very audible for recording.
|
On my workstation it translates in 15-12 ms VS 9-5 ms for 256 samples.
|
|
|
01-04-2019, 04:50 AM
|
#10
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
The Steinberg UR22 has lower latency than an RME 9652 ? wow
|
|
|
01-04-2019, 06:10 AM
|
#11
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Using stealth sends requires a lot of sends. I set mine up to be able to feed 10 FX and it results in over 500 sends. Well it turns out they are not free. I went back and muted all of them except for 5 of them and I got a big difference.
RT CPU Value in perf monitor ..................
5 stealth sends unmuted with about 495 muted
7.2 idle
1 track armed with FX on it
48 idle
53 during playback
about 500 stealth sends unmuted
7.2 idle
1 track armed with FX on it
70 idle
82 during playback
|
|
|
01-04-2019, 07:39 AM
|
#12
|
Human being with feelings
Join Date: Aug 2015
Location: Florence, Italy
Posts: 463
|
Quote:
Originally Posted by Coachz
The Steinberg UR22 has lower latency than an RME 9652 ? wow
|
No my signature is not updated. I'm using an iD22 for primary soundcard, I'm also using the UR22 for secondary, then I have a third soundcard for movies and other things.
I wrote this in my previous post.
|
|
|
01-04-2019, 06:39 PM
|
#13
|
Human being with feelings
Join Date: May 2017
Posts: 3,201
|
What is a stealth send?
|
|
|
01-04-2019, 07:05 PM
|
#14
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by Philbo King
What is a stealth send?
|
The sending child tracks to a folder or bus on higher channels such as 5/6, then sending the parent bus/folder's 5/6 to say a reverb track's 1/2.
This avoids issues with send proportions changing when changing the bus folder volume (which occurs since the sends on children bypass the folder/bus and go directly to the reverb track).
It also fixes an issue when using VCAs to solve the same problem, VCA's introduce the side-effect of not being able to place non-linear FX on the bus/folder - since VCAs control child tracks and will hit the parent's FX harder. So VCAs solve one problem and cause another.
Sending through the parent on unused channels solves all of this.
__________________
Music is what feelings sound like.
Last edited by karbomusic; 01-04-2019 at 07:11 PM.
|
|
|
01-05-2019, 03:46 AM
|
#15
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
And correct me if I'm wrong but all of that could be avoided with one checkbox in prefs.
X ... Parent folder faders adjust all child sends proportionaly.
|
|
|
01-05-2019, 04:35 AM
|
#16
|
Human being with feelings
Join Date: Apr 2017
Location: South
Posts: 587
|
Quote:
Originally Posted by Coachz
And correct me if I'm wrong but all of that could be avoided with one checkbox in prefs.
X ... Parent folder faders adjust all child sends proportionaly.
|
PLEASE
|
|
|
01-06-2019, 07:59 AM
|
#17
|
Human being with feelings
Join Date: Oct 2017
Location: Black Forest
Posts: 5,054
|
What also helped me a great deal in terms of performance is the setting:
Buffer Preference: Thread Priority Highest, Behavior 0 - Relaxed
|
|
|
01-06-2019, 08:16 AM
|
#18
|
Banned
Join Date: Dec 2016
Location: England
Posts: 2,432
|
hey- is there any windoze10 users,or cockos, please mind explaining this reaper advanced disk io dialog>
^like,which settings would seem appropriate optimizations for either mac,linux or win10 users?
please+ty.
|
|
|
01-06-2019, 08:46 AM
|
#19
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Quote:
Originally Posted by _Stevie_
What also helped me a great deal in terms of performance is the setting:
Buffer Preference: Thread Priority Highest, Behavior 0 - Relaxed
|
Thanks for sharing. Do you have a way to measure performance ? RT CPU or such ?
|
|
|
01-06-2019, 10:45 AM
|
#20
|
Human being with feelings
Join Date: Oct 2017
Location: Black Forest
Posts: 5,054
|
I just used the internal Reaper performance meter.
The project I was using had around 1200 (disabled) tracks.
without that tweak
IDLE and Playback:
Total CPU: 16 - 47% (it was highly fluctuating)
RT CPU: around 9.5%
with the tweak:
IDLE and Playback:
Total CPU: 0.9 - 2.2%
RT CPU: around 7.4%
|
|
|
01-06-2019, 11:01 AM
|
#21
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Quote:
Originally Posted by _Stevie_
I just used the internal Reaper performance meter.
The project I was using had around 1200 (disabled) tracks.
without that tweak
IDLE and Playback:
Total CPU: 16 - 47% (it was highly fluctuating)
RT CPU: around 9.5%
with the tweak:
IDLE and Playback:
Total CPU: 0.9 - 2.2%
RT CPU: around 7.4%
|
Nice. I tried your tweak and got improvement. Thanks.
This is with about 50 tracks playing and one track armed.
Default
Relaxed
Last edited by Coachz; 01-08-2019 at 04:10 AM.
|
|
|
01-08-2019, 01:10 AM
|
#22
|
Human being with feelings
Join Date: Aug 2015
Location: Florence, Italy
Posts: 463
|
Quote:
Originally Posted by _Stevie_
What also helped me a great deal in terms of performance is the setting:
Buffer Preference: Thread Priority Highest, Behavior 0 - Relaxed
|
I'll try it too when I came back from work.
|
|
|
01-08-2019, 04:19 AM
|
#23
|
Human being with feelings
Join Date: Apr 2017
Location: South
Posts: 587
|
Thanks for this.
Learning a lot
|
|
|
01-08-2019, 12:05 PM
|
#24
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
hey everyone, I am confused about something I discovered the other day, the relationship between a plugin's PDC and CPU performance.
I have been using the SSL channel on a recording track. It's pdc measures 64, or 1 sample. I recently found a new channel strip, Channel X. It's PDC is 0, so I thought it would be better to record with.
But, RT cpu meter will measure 10% with the SSL, and 30% with Channel X. BIG difference.
I was under the assumption that a higher PDC would mean higher CPU usage to compensate for the latency, but apparently that is not the case.
Can anyone shed light on this?
|
|
|
01-08-2019, 12:20 PM
|
#25
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
Quote:
Originally Posted by karbomusic
The sending child tracks to a folder or bus on higher channels such as 5/6, then sending the parent bus/folder's 5/6 to say a reverb track's 1/2.
This avoids issues with send proportions changing when changing the bus folder volume (which occurs since the sends on children bypass the folder/bus and go directly to the reverb track).
It also fixes an issue when using VCAs to solve the same problem, VCA's introduce the side-effect of not being able to place non-linear FX on the bus/folder - since VCAs control child tracks and will hit the parent's FX harder. So VCAs solve one problem and cause another.
Sending through the parent on unused channels solves all of this.
|
karbo, thanks for explaining this. I've been trying to find more about stealth sends but with no luck.
|
|
|
01-08-2019, 11:06 PM
|
#26
|
Human being with feelings
Join Date: Jun 2013
Location: Krefeld, Germany
Posts: 14,687
|
Quote:
Originally Posted by poetnprophet
I was under the assumption that a higher PDC would mean higher CPU usage to compensate for the latency, but apparently that is not the case.
|
On playback Reaper just feeds the tracks with data from different locations in the media file to compensate the tracks' latency. No CPU overhead at all.
In Stop mode ("Live usage") I suppose Reaper introduces appropriate delays, but that does not consume much CPU.
-Michael
|
|
|
01-09-2019, 12:49 AM
|
#27
|
Human being with feelings
Join Date: Aug 2015
Location: Florence, Italy
Posts: 463
|
Quote:
Originally Posted by _Stevie_
What also helped me a great deal in terms of performance is the setting:
Buffer Preference: Thread Priority Highest, Behavior 0 - Relaxed
|
Tried it, no improvements here.
|
|
|
01-09-2019, 10:28 AM
|
#28
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
Here's something else I discovered, I'm curious if anyone has some insight.
Yesterday I changed my recording template. Previously I just had recorded audio in folders, divided by category...typical. I mostly record vocals, so vocal folders go to a vocal bus. All vox fx go to the vocal bus. Everything else goes to a sub-master, then the master...simple. My new template incorporated VCA's for each folder, FX went to an fx buss and then the sub master. All other busses went to a dry sub, then the submaster.
Now my submaster track can barely run Ozone. Previous projects with the previous template, I can have 70 tracks and 150 plugs including ozone, sometimes with another instance of ozone after that and runs no problem using just 15% real time cpu. Now with this new template, my first project has only 27 tracks (20 of them are just the template and routing) with maybe 40 plugins and the CPU is at 160% as soon as I activate ozone. I'm so confused.
|
|
|
01-09-2019, 10:57 AM
|
#29
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by poetnprophet
Here's something else I discovered, I'm curious if anyone has some insight.
|
One of the users who remembers better than I can chime in on the specifics but it matters/changes based on folder structure and who/what has VSTs inserted... because it may force a bunch of VSTs onto a single CPU core.
Quote:
Tried it, no improvements here.
|
To everyone:
Just a sanity check that all these settings have multiple choices exactly because no two use cases or hardware configs are the same. So some set of thread behaviors or other tweaks may be the bee's knees for one user and the worst possible performance for another. I think the thread idea is great, just keep in mind that changeable settings exist because there truly is no one-size-fits-all combo.
__________________
Music is what feelings sound like.
|
|
|
01-09-2019, 12:46 PM
|
#30
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Yeah Hardware is different and Hardware is the same. The more we find in common the more settings will work in common. Luckily we're all using Reaper so we have some of the ame resources being required.
With my latest stealth sends setup I muted every send and will just be enabling them as I need them because I found unmuted 500 sends takes up a lot of RT CPU.
Also I have every effect offline when I load my initial Reaper template, even my effects bus which has 10 different effects tracks. When I begin to send that's what I'm going to start in a bellinis resources and if needed freeze Tracks Of course. I also have buttons in the toolbar to set tracks offline and online for all of the effects on them to speed that process up. I also have the PDC latency checked on my midi tracks and I'm recording onto. I don't know either way but I'm going to try to see if I can notice what it's doing audibly overtime
With that combination I'm unstoppable!
|
|
|
01-09-2019, 12:55 PM
|
#31
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
I'm not sure if things have changed since I changed my video card, but my last big project (from last year at this time) was 70-100 tracks and roughly 250 VSTs. Never had to freeze or disable anything and ran like a champ.
__________________
Music is what feelings sound like.
|
|
|
01-09-2019, 01:14 PM
|
#32
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
can you give some particulars on your setup? CPU, Ram, buffer size, audio card?
|
|
|
01-09-2019, 01:20 PM
|
#33
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by Coachz
can you give some particulars on your setup? CPU, Ram, buffer size, audio card?
|
i7-2600k @ 4.5GHz + RME UFX + 24GB RAM + 1024 buffer. I think I have my thread behavior set to the one that says "automatic (experimental)".
__________________
Music is what feelings sound like.
|
|
|
01-09-2019, 01:41 PM
|
#34
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Quote:
Originally Posted by karbomusic
i7-2600k @ 4.5GHz + RME UFX + 24GB RAM + 1024 buffer. I think I have my thread behavior set to the one that says "automatic (experimental)".
|
1024 is a big buffer!
|
|
|
01-09-2019, 01:50 PM
|
#35
|
Human being with feelings
Join Date: May 2009
Posts: 29,260
|
Quote:
Originally Posted by Coachz
1024 is a big buffer!
|
I'd run larger if it let me. I have no reason to use smaller buffers when mixing ever unless...
1. I have some automation that has to happen so fast that 1024 samples is too long (21 ms @48k) which pretty much never happens for me. Automation resolution follows block size.
2. I have to track and monitor through reaper late in the project, which is rarely going to be the case by the time the project has grown to that complexity. 99% of that would have occurred during initial tracking. The exception is the occasional overdub near the end of the project that also needed to be monitored through reaper, but that's rare and if it were to be an issue, it's just a quick overdub, I'd drop the buffer and just turn off a few of the offending VSTs if it were to start popping/clicking.
Kind of my point with #2 is tracking and mixing are ideally completely different phases, I'd think something is wrong if I'm going that far back in the process mid mix.
__________________
Music is what feelings sound like.
|
|
|
01-09-2019, 02:03 PM
|
#36
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Makes sense
|
|
|
01-10-2019, 07:44 PM
|
#37
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
Quote:
Originally Posted by karbomusic
One of the users who remembers better than I can chime in on the specifics but it matters/changes based on folder structure and who/what has VSTs inserted... because it may force a bunch of VSTs onto a single CPU core.
|
just following up, I removed my dry mixbus, and seems the issue went away. very strange that having just a couple of audio sends would cause this as it was basically just a pass thru track.
|
|
|
01-11-2019, 03:53 AM
|
#38
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Quote:
Originally Posted by poetnprophet
just following up, I removed my dry mixbus, and seems the issue went away. very strange that having just a couple of audio sends would cause this as it was basically just a pass thru track.
|
Try it with 500 sends like I have . Oye vey!
|
|
|
01-11-2019, 10:54 AM
|
#39
|
Human being with feelings
Join Date: Jan 2018
Posts: 1,651
|
ha, well here's another update. I had a recording session last night using my template. I had recorded 2 vocal tracks over an instrumental no problem. I went to record a 3rd and I got huge latency and glitches. CPU was only at 20%. I couldn't figure out why, ended up recording with no plugins at all just to get through the session. The previous day I had recorded 12 tracks no problem at all.
This morning, I realized that I had changed some default settings based on this thread.
-buffer preference behavior, from default to relaxed.
-allow fx multi, from unchecked to 5
-auto detect threads, from default 12 to 7.
It seemed to work pretty well when mixing, getting some slight CPU decrease. But for recording live it seems to be counter-productive.
So tonight i will revert my settings to default and hope that gets me back to where I was.
|
|
|
01-11-2019, 12:29 PM
|
#40
|
Human being with feelings
Join Date: Oct 2010
Location: Charleston, SC
Posts: 12,769
|
Quote:
Originally Posted by poetnprophet
ha, well here's another update. I had a recording session last night using my template. I had recorded 2 vocal tracks over an instrumental no problem. I went to record a 3rd and I got huge latency and glitches. CPU was only at 20%. I couldn't figure out why, ended up recording with no plugins at all just to get through the session. The previous day I had recorded 12 tracks no problem at all.
This morning, I realized that I had changed some default settings based on this thread.
-buffer preference behavior, from default to relaxed.
-allow fx multi, from unchecked to 5
-auto detect threads, from default 12 to 7.
It seemed to work pretty well when mixing, getting some slight CPU decrease. But for recording live it seems to be counter-productive.
So tonight i will revert my settings to default and hope that gets me back to where I was.
|
What was your RT CPU at when having problems ?
|
|
|
Thread Tools |
|
Display Modes |
Linear Mode
|
Posting Rules
|
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
HTML code is Off
|
|
|
All times are GMT -7. The time now is 01:09 PM.
|