Go Back   Cockos Incorporated Forums > REAPER Forums > REAPER General Discussion Forum

Reply
 
Thread Tools Display Modes
Old 11-06-2022, 10:31 AM   #1
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default performance settings with high core CPU's

As high core/thread cpu's are now the norm are there any Reaper performance settings that can be applied?

For example: Reaper always performs very well when playing back tracks when mixing bit can descend into crack and pop city if you record arm a track.

So in this case, if like me you have 16/32 cores/threads, would it be pertinent to let reaper use say 14/28 for general use ,and for monitoring set it to use the other 2/4 threads as discrete cores/threads just for rec armed tracks? In fact can reaper thread allocate like this?




M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-06-2022, 11:24 AM   #2
King Stupid
Human being with feelings
 
Join Date: Jan 2020
Posts: 1,172
Default

Quote:
Originally Posted by norbury brook View Post
As high core/thread cpu's are now the norm are there any Reaper performance settings that can be applied?

For example: Reaper always performs very well when playing back tracks when mixing bit can descend into crack and pop city if you record arm a track.

So in this case, if like me you have 16/32 cores/threads, would it be pertinent to let reaper use say 14/28 for general use ,and for monitoring set it to use the other 2/4 threads as discrete cores/threads just for rec armed tracks? In fact can reaper thread allocate like this?




M
Most processing in audio is linear, not good for parallel processing. Reaper is very good at parallel processing with it's native plugins. Some VST's, not so much. Apparently, how you have tracks arranged and how the VSTs are applied to them can affect what core things may be sent to for processing. It's why you will pften see one core pegged at 100%, a couple at 30 or 40% and the rest at ~5%.
King Stupid is offline   Reply With Quote
Old 11-07-2022, 03:28 AM   #3
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by King Stupid View Post
Most processing in audio is linear, not good for parallel processing. Reaper is very good at parallel processing with it's native plugins. Some VST's, not so much. Apparently, how you have tracks arranged and how the VSTs are applied to them can affect what core things may be sent to for processing. It's why you will pften see one core pegged at 100%, a couple at 30 or 40% and the rest at ~5%.
I'm very aware of that but that doesn't answer the question of whether one can allocate threads specifically for live monitoring vs playback.


M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-07-2022, 04:05 AM   #4
airon
Human being with feelings
 
airon's Avatar
 
Join Date: Aug 2006
Location: Berlin
Posts: 11,814
Default

A very juicy problem.

I'd love to know more about this as well, if anyone want so to share their experience in this area. Perhaps the devs can jump on if it's something they have not tackled yet.
__________________
Using Latch Preview (Video) - Faderport 16 setup for CSI 1.1 , CSI 3.10
Website
"My ego comes pre-shrunk" - Randy Thom
airon is offline   Reply With Quote
Old 11-07-2022, 06:30 AM   #5
King Stupid
Human being with feelings
 
Join Date: Jan 2020
Posts: 1,172
Default

Quote:
Originally Posted by norbury brook View Post
I'm very aware of that but that doesn't answer the question of whether one can allocate threads specifically for live monitoring vs playback.


M
Unlikely to be able to set different cores for specific tasks within Reaper. At least, not from what I've seen or read.

However, I could record 128 tracks simultanously, each with 3 JS effects and reeq running at 1.6ms without any glitching on my old 3700x.

Instrument VST's adn heavy plugins are, of course, a different matter.

What kind of instrumentation and plugins are we talking about on the track you arm that are causing glitching and popping? I wonder if it's jsut a little work that needs to be done to rectify this issue for you?
King Stupid is offline   Reply With Quote
Old 11-07-2022, 12:02 PM   #6
Uncovered Pitch
Human being with feelings
 
Join Date: Jul 2018
Posts: 101
Default

Maybe a bit too obvious but have you tried the setting at the bottom of the Buffering preferences page where you can specify the number of cores you want for live FX multiprocessing?
Uncovered Pitch is online now   Reply With Quote
Old 11-07-2022, 03:13 PM   #7
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by Uncovered Pitch View Post
Maybe a bit too obvious but have you tried the setting at the bottom of the Buffering preferences page where you can specify the number of cores you want for live FX multiprocessing?
That’s exactly what I’m talking about. Those settings I tried my suggestion today and it seemed to make no difference.

I was running a project with a full BFD 3 kit, a Sitalia kit, 3 instances of HIVE , Pianoteq 7 full version, Spitfire appasionata strings and was getting the odd pop and crackle @32 samples from my RME Raydat .

I know 32 samples is very low …0.7ms latency but I thought it was a good test of my idea of splitting Cores for playback/ record armed.


M
__________________
https://www.marcuscliffe.com/

Last edited by norbury brook; 11-08-2022 at 07:23 AM.
norbury brook is offline   Reply With Quote
Old 11-08-2022, 12:34 PM   #8
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 12,536
Default

For live sound - ie. live performance through the computer DAW mixer that requires imperceptible lag - the most bang for the buck is to start with an audio interface with a lower baseline latency. That leads to landing on a higher block size to achieve the same target latency vs with a slower interface.

Shop for thunderbolt connecting models or some of the higher performance USB or firewire models. A faster interface will let you achieve, for example, a 6 or 7 ms round trip latency with a 128 sample block size whereas a slower unit will require the block size dialed down to 64 or 32 samples to hit that target.

You can't run any plugin live that has a higher internal lag than what your block size is set to. So the faster interface also leads to more and higher latency plugins being able to be used live.

All of this is no matter how fast of a computer you have. You would shut down that $60,000 film house Mac Pro with an errant setup just the same.

For audio in general, shop for fastest single core performance first.
For live audio, shop for a faster audio interface first and foremost.
serr is offline   Reply With Quote
Old 11-08-2022, 03:23 PM   #9
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by serr View Post
For live sound - ie. live performance through the computer DAW mixer that requires imperceptible lag - the most bang for the buck is to start with an audio interface with a lower baseline latency. That leads to landing on a higher block size to achieve the same target latency vs with a slower interface.

Shop for thunderbolt connecting models or some of the higher performance USB or firewire models. A faster interface will let you achieve, for example, a 6 or 7 ms round trip latency with a 128 sample block size whereas a slower unit will require the block size dialed down to 64 or 32 samples to hit that target.

You can't run any plugin live that has a higher internal lag than what your block size is set to. So the faster interface also leads to more and higher latency plugins being able to be used live.

All of this is no matter how fast of a computer you have. You would shut down that $60,000 film house Mac Pro with an errant setup just the same.

For audio in general, shop for fastest single core performance first.
For live audio, shop for a faster audio interface first and foremost.
I have an RME RAYDAT, that's as good as it gets latency wise.

I've been working on this project all day today now adding lots more stuff and a few plugins so i can send a rough mix to my client.Including a whole additional set of Spitfire samples across 6 tracks. 4 more HIve instances. Kontact Symphobia, another Pianoteq instance, and acustica aqua plugins .softube and Weiss mastering limiter.

This was still all done @32 samples and 0.7ms latency... I don't think I'd get better performance from any other interface.

M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-08-2022, 04:38 PM   #10
King Stupid
Human being with feelings
 
Join Date: Jan 2020
Posts: 1,172
Default

Quote:
Originally Posted by norbury brook View Post
I have an RME RAYDAT, that's as good as it gets latency wise.

I've been working on this project all day today now adding lots more stuff and a few plugins so i can send a rough mix to my client.Including a whole additional set of Spitfire samples across 6 tracks. 4 more HIve instances. Kontact Symphobia, another Pianoteq instance, and acustica aqua plugins .softube and Weiss mastering limiter.

This was still all done @32 samples and 0.7ms latency... I don't think I'd get better performance from any other interface.

M
Well, could be 32 at 96 might just be too much for it. Does your issue get better at 1.3ms? (either 32/48 or 64/96)?
King Stupid is offline   Reply With Quote
Old 11-09-2022, 05:19 AM   #11
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by King Stupid View Post
Well, could be 32 at 96 might just be too much for it. Does your issue get better at 1.3ms? (either 32/48 or 64/96)?
It's not really a problem generally but it's an area Reaper isn't as good as Cubase say for example.

My post wasn't about solving a problem per se but about tweaking for max performance and whether reaper could be allocated cores/threads for specifics.

Having tried it now I see it can't so that answers my own question.

My last post was to show that I'm getting pretty stunning performance from my machine and RME Raydat.

nearly at the end of a project and still running 32 samples/ @0.7ms latency even with Aquas and Wiess plugins on the master.





M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-10-2022, 03:20 AM   #12
brk303
Human being with feelings
 
Join Date: Nov 2016
Location: Serbia
Posts: 172
Default

Quote:
Originally Posted by norbury brook View Post
@0.7ms latency even with Aquas and Wiess plugins on the master.
Not sure about Aquas, but Weiss is not a zero latency plugin, if you chase lowest latency you should try to use zero latency plugins and among those ones that use least CPU. Otherwise it defies the purpose.

My band plays live via Reaper and I use 44k/128 samples with ~7 ms latency, and everyone is fine with that.
0.7ms sounds awesome, but I doubt you really need it that low for any practical purpose.
brk303 is offline   Reply With Quote
Old 11-10-2022, 04:50 AM   #13
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by brk303 View Post
Not sure about Aquas, but Weiss is not a zero latency plugin, if you chase lowest latency you should try to use zero latency plugins and among those ones that use least CPU. Otherwise it defies the purpose.

My band plays live via Reaper and I use 44k/128 samples with ~7 ms latency, and everyone is fine with that.
0.7ms sounds awesome, but I doubt you really need it that low for any practical purpose.
yes of course, but I mentioned the weiss and the acustica Aquas on the master bus because they're a real heavy cpu munching set of plugins and i'm still running fine @32 samples

i was really seeing how my machine with the new RME RAyDAt could perform with this project. I was expecting at any point to have to raise the buffers but it just kept going.

Amazing

Ironically I'm actually using total mix and hardware reverbs when record monitoring so it's a zero latency regardless of project setting as I'm not monitoring through reaper, even though at 0.7ms i could easily.





M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-11-2022, 06:42 PM   #14
Lynx_TWO
Human being with feelings
 
Lynx_TWO's Avatar
 
Join Date: Dec 2019
Location: St Petersburg FL
Posts: 989
Default

My last setup at 3.2ghz boost had 28 cores and 56 threads (dual Xeon CPU) and yes, you can limit the number of cores and threads used by Reaper, but I don’t think you can actually assign specific cores to specific tasks, unfortunately…. Now I use AMD with 12 cores at 4.2ghz boost and it outperforms the old system. When it comes to audio, higher frequency trumps multiple cores. To a point, of course; I’m betting a 4 core 5ghz system would probably trump a 10ghz single core CPU (which don’t exist outside the lab)
__________________
My mixes from the Cambridge multitracks library
SoundCloud link & Youtube (ThemTube?) link
My preferred adjectives are “Handsome” and “Brilliant”
Lynx_TWO is offline   Reply With Quote
Old 11-24-2022, 07:33 AM   #15
airon
Human being with feelings
 
airon's Avatar
 
Join Date: Aug 2006
Location: Berlin
Posts: 11,814
Default

Quote:
Originally Posted by norbury brook View Post
yes of course, but I mentioned the weiss and the acustica Aquas on the master bus because they're a real heavy cpu munching set of plugins and i'm still running fine @32 samples

i was really seeing how my machine with the new RME RAyDAt could perform with this project. I was expecting at any point to have to raise the buffers but it just kept going.

Amazing

Ironically I'm actually using total mix and hardware reverbs when record monitoring so it's a zero latency regardless of project setting as I'm not monitoring through reaper, even though at 0.7ms i could easily.

M
Would you mind sharing what converters you're using ?

I'm planning ahead for whatever distant future my RME Fireface UC dies on me, which is probably never but who knows. I'd be swinging over to a ADAT optical port(Raydat/Digi/etc.) in to an 8-channel converter for running my 5.1 system.
__________________
Using Latch Preview (Video) - Faderport 16 setup for CSI 1.1 , CSI 3.10
Website
"My ego comes pre-shrunk" - Randy Thom
airon is offline   Reply With Quote
Old 11-24-2022, 07:54 AM   #16
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by airon View Post
Would you mind sharing what converters you're using ?

I'm planning ahead for whatever distant future my RME Fireface UC dies on me, which is probably never but who knows. I'd be swinging over to a ADAT optical port(Raydat/Digi/etc.) in to an 8-channel converter for running my 5.1 system.
Of course

Ferrofish pulse 16 really great sounding converters.

And if you use all 4 ADAT ports on the Raydat and Ferrofish you can keep using all 16 IO even at 96k

I don’t need 32 analogue IO so this suits me fine.


M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-25-2022, 03:44 AM   #17
airon
Human being with feelings
 
airon's Avatar
 
Join Date: Aug 2006
Location: Berlin
Posts: 11,814
Default

Interesting indeed. There are more choices than I thought.

There's Ferrofish with 16 I/O is about 1k here in Germany as well. THanks for pointing that one out. I had no idea they existed.

Focusrite Clarett+ OctoPre is 800 and has eight mic preamps that I'll never need(probably).

Quite a few options that are more expensive and usually have I/O options that aren't necessary for me. Then again, maybe I'll want to run MADI or Dante converters instead of optical lines. If that happens, I'll likely just grab an RME converter unit and a PCIe RME MADI card.

And then there's the budget option Behringer ADA8200 for just 211 Euros here. I guess that could work too, though I'd prefer something as reliable as RME. Depends on the budget I suppose.
__________________
Using Latch Preview (Video) - Faderport 16 setup for CSI 1.1 , CSI 3.10
Website
"My ego comes pre-shrunk" - Randy Thom
airon is offline   Reply With Quote
Old 11-25-2022, 07:55 AM   #18
mozart999uk
Human being with feelings
 
Join Date: Nov 2010
Posts: 1,710
Default

Quote:
Originally Posted by serr View Post
For live sound - ie. live performance through the computer DAW mixer that requires imperceptible lag - the most bang for the buck is to start with an audio interface with a lower baseline latency. That leads to landing on a higher block size to achieve the same target latency vs with a slower interface.

Shop for thunderbolt connecting models or some of the higher performance USB or firewire models. A faster interface will let you achieve, for example, a 6 or 7 ms round trip latency with a 128 sample block size whereas a slower unit will require the block size dialed down to 64 or 32 samples to hit that target.

You can't run any plugin live that has a higher internal lag than what your block size is set to. So the faster interface also leads to more and higher latency plugins being able to be used live.

All of this is no matter how fast of a computer you have. You would shut down that $60,000 film house Mac Pro with an errant setup just the same.

For audio in general, shop for fastest single core performance first.
For live audio, shop for a faster audio interface first and foremost.
No way a USB or thunderbolt is gonna be faster than a RME raydat 🤣
mozart999uk is offline   Reply With Quote
Old 11-25-2022, 07:57 AM   #19
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by mozart999uk View Post
No way a USB or thunderbolt is gonna be faster than a RME raydat 🤣
Well I think the presonus quantum is technically lower latency but doesn’t perform as well.

Yep, the PCIe RME cards are amazing, and it’s an investment that will last you 15 years or more


M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Old 11-25-2022, 08:51 AM   #20
Hermetech Mastering
Human being with feelings
 
Hermetech Mastering's Avatar
 
Join Date: May 2019
Location: Milan, Italy
Posts: 506
Default

True, been using the HDSPe AES daily since June 2011 here, in two different PCs. Never had a single problem with it!
Hermetech Mastering is offline   Reply With Quote
Old 11-26-2022, 09:18 AM   #21
PitchSlap
Human being with feelings
 
PitchSlap's Avatar
 
Join Date: Jan 2008
Location: Vancouver, BC
Posts: 3,776
Default

Make sure you set Thread behavior priory to default (3) under Buffering. This made a big difference for performance issues I was having (could be AMD specific).
__________________
FRs: v5 Media Explorer Requests, Global Quantization, Session View
Win10 Pro 64-bit, Reaper 6(x64), AMD 3950x, Aorus X570 Master, 64GB DDR4 3600, PowerColor Red Devil 5700XT, EVO 970 2TB, 10TB HD, Define R6

Last edited by PitchSlap; 11-26-2022 at 06:30 PM.
PitchSlap is offline   Reply With Quote
Old 11-26-2022, 10:20 AM   #22
80icio
Human being with feelings
 
Join Date: Mar 2016
Location: Italy
Posts: 311
Default

Quote:
Originally Posted by norbury brook View Post
As high core/thread cpu's are now the norm are there any Reaper performance settings that can be applied?

For example: Reaper always performs very well when playing back tracks when mixing bit can descend into crack and pop city if you record arm a track.

So in this case, if like me you have 16/32 cores/threads, would it be pertinent to let reaper use say 14/28 for general use ,and for monitoring set it to use the other 2/4 threads as discrete cores/threads just for rec armed tracks? In fact can reaper thread allocate like this?




M
Putting a higher media buffer time (my set up is 1600 ms instead of default 1200ms) fixed a lot of pops and crackles at low buffer sizes when recording tracks on a heavy processing session
80icio is offline   Reply With Quote
Old 11-27-2022, 03:25 AM   #23
norbury brook
Human being with feelings
 
norbury brook's Avatar
 
Join Date: Mar 2007
Location: London UK
Posts: 3,378
Default

Quote:
Originally Posted by 80icio View Post
Putting a higher media buffer time (my set up is 1600 ms instead of default 1200ms) fixed a lot of pops and crackles at low buffer sizes when recording tracks on a heavy processing session
interesting , I have gone the other way though and have all my Reaper system buffer settings at 50% of the default.

So , you're saying by making the 'media buffer'? larger you can record at low latency with better performance on a project that already has fairly heavy processing ?


M
__________________
https://www.marcuscliffe.com/
norbury brook is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 08:19 PM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.