Possible bug? Reason doesn't use all cores, but load metter is red.

This forum is for discussing Reason. Questions, answers, ideas, and opinions... all apply.
calmbird
Posts: 12
Joined: 08 May 2017

04 Jun 2017

Hi.
My Reason can't handle some ammount of VSTs, load measure becomes red, sound becomes bad and squicky.
However I run task manager, and I can see less than half of the threads are in use.
Do you maybe have same problem?
Attachments
reasonbug.jpg
reasonbug.jpg (820.66 KiB) Viewed 7630 times
Last edited by calmbird on 04 Jun 2017, edited 1 time in total.

WongoTheSane
Moderator
Posts: 1851
Joined: 14 Sep 2015
Location: Paris, France

04 Jun 2017

When using computationally-intensive programs (like Reason), virtual cores are not effective (they are meant to speed up multiple tasks happening simultaneously, like browsing Internet and watching a movie at the same time). Say you have a core that can perform 100 operations per second: virtualizing it splits that core in two logical units, capable of 50 operations per second each (simplifying for the sake of demonstration). You're gaining on multitasking, but not on raw power.

The Task Manager is a bit misleading in this case: it gives you the impression that your CPU is only used at 50%, when in reality it's already maxed up (in the example above, you would have the first core doing 100 operations per seconds, and nothing left for the second virtual core; if you consider the total number of operations per second, you're maxed up, but if you consider per core, only half the cores are busy - it's purely theoretical, but that's what the Task Manager is showing you).

So in the case of Reason, forget virtual cores and only count physical ones. Among those four, three of them are busy actually crunching the sounds, but the fourth one is dedicated to everything else: drawing the sequencer/rack/mixer, handling MIDI, and managing the other three (and performing other menial tasks like system I/O, clock, other programs in the background, etc). This fourth one is generally less busy than the other three, so you might think that you still have unused power that could be leveraged, but in reality it would cause more problems than it would solve.

In the end, what your screen capture tells me is that your CPU is pretty much maxed out. I'm afraid you'll have to resort to the CPU-saving techniques that we all use: bounce in place, use Master Send effects instead of duplicating effects in each patch, use "economical" devices for things that don't require high quality (for instance, use a Subtractor for subs instead of an Expanse or a ReSpire)... Or upgrade your CPU to a better one. A never-ending pursuit though! :)

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

04 Jun 2017

WongoTheSane wrote:So in the case of Reason, forget virtual cores and only count physical ones. Among those four, three of them are busy actually crunching the sounds, but the fourth one is dedicated to everything else: drawing the sequencer/rack/mixer, handling MIDI, and managing the other three (and performing other menial tasks like system I/O, clock, other programs in the background, etc). This fourth one is generally less busy than the other three, so you might think that you still have unused power that could be leveraged, but in reality it would cause more problems than it would solve.
Just some clarity for those who are chiming in late: Reason utilizes Hyperthreading (virtual cores) for many things with the exception of audio DSP (for instance, disc i/o and high-quality stretch and resample). A great way to demonstrate this is to load a lot of large audio files into the sequencer and change the tempo to something dramatically different than the original:

Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

04 Jun 2017

Also, the Windows Task Manager CPU usage number shows the percentage of real time Windows considers a processor core working rather than idle. 100% CPU usage on a core signifies that Windows thinks that core is sustaining execution without pause. When two threads are running on two logical cores which share a single physical core, the execution speed (intructions per second) of each thread varies hugely. At best, it will be about 30 % faster than traditional (non-HyperThreading) switching between two threads on a single core (The two HT-threads only run some instructions in parallel, under some conditions, but most are executed sequentially).

Reason audio rendering threads can not work reliably on threads with such varying execution speed.
When two threads are running in parallel on two physical cores, the execution speed varies much less, and only because of memory, caches and such. I.e. the way Reason dedicates threads to cores is only reliable with physical cores, and not with logical cores that execute using HyperThreading.

If Reason would dedicate all your cores to audio rendering, including logical cores, it would be much less reliable and responsive. The DSP meter measures the percentage of real time left between deliveries of audio to/from the interface driver.

If it shows 0%, almost no time is spent in Reason to deliver audio to/from the interface driver.
If it shows 100%, the interface driver wants more audio before Reason has finished rendering the previous batch.

You can manually change part of Reason's multi-core implementation by assigning more "audio worker" threads. This is done by using the following shortcuts in Reason:
  • Adding on Windows: Ctrl Shift Alt +
  • Removing on Windows: Ctrl Shift Alt -
  • Adding on Mac: Command Option Shift +
  • Removing on Mac: Command Option Shift -
This setting is NOT saved between sessions. Changing this setting is kind of a trade-off between different priorities, for example GUI performance, RAM-intensive operations, audio playback and so on could be impacted. So try it and see if it improves performance in your case. However, we recommend using this command sparingly or simply keeping it on the original setting.

Source.
Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

WongoTheSane
Moderator
Posts: 1851
Joined: 14 Sep 2015
Location: Paris, France

04 Jun 2017

EnochLight wrote:Also, the Windows Task Manager CPU usage number shows the percentage of real time Windows considers a processor core working rather than idle. 100% CPU usage on a core signifies that Windows thinks that core is sustaining execution without pause.
Is there any reason why they don't add both hyperthreads before averaging? In OP's screenshot, you can see that Task Manager shows 55%, while the first three cores are at 100%, which means it's 75% busy already. Never understood that (and it's the same with Windows 7). Is it a marketing decision? Or to avoid support calls like "it says 100% but half my cores are doing nothing"?

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

04 Jun 2017

WongoTheSane wrote:
EnochLight wrote:Also, the Windows Task Manager CPU usage number shows the percentage of real time Windows considers a processor core working rather than idle. 100% CPU usage on a core signifies that Windows thinks that core is sustaining execution without pause.
Is there any reason why they don't add both hyperthreads before averaging? In OP's screenshot, you can see that Task Manager shows 55%, while the first three cores are at 100%, which means it's 75% busy already. Never understood that (and it's the same with Windows 7). Is it a marketing decision? Or to avoid support calls like "it says 100% but half my cores are doing nothing"?
No idea. Not sure if it's for technical reasons or designed on purpose as you describe.
Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

calmbird
Posts: 12
Joined: 08 May 2017

04 Jun 2017

Thanx for responses!

I've checked with Reaper, with some multithread options turned on, and my observations have following conclusions.

I run Reaper with 14 tracks, each track includes:
Predator device,
2x RC48 Reverb,
RC24
Molot
loaded with elements Izotope Ozone 7

Up to 13 tracks everything runs smoothly.
And as you can see, task manager is showing 100% utilization.

Normaly Reason could handle up to 4 of such tracks.

I think something is wrong with way Reason utilizes those threads/cores.
Attachments
reaper.png
reaper.png (998.8 KiB) Viewed 7534 times

ShawnG
Posts: 120
Joined: 31 Aug 2015

05 Jun 2017

would certainly concede that Reason lags behind Reaper in cpu utilization.

but 3 algorithmic reverbs PER TRACK? + Ozone?

that's no way to rack up a solid track count in any daw, hopefully this is just some sort of stress test?

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

06 Jun 2017

calmbird wrote:Thanx for responses!

I've checked with Reaper, with some multithread options turned on, and my observations have following conclusions.

I run Reaper with 14 tracks, each track includes:
Predator device,
2x RC48 Reverb,
RC24
Molot
loaded with elements Izotope Ozone 7

Up to 13 tracks everything runs smoothly.
And as you can see, task manager is showing 100% utilization.

Normaly Reason could handle up to 4 of such tracks.

I think something is wrong with way Reason utilizes those threads/cores.
Have you tried this?

viewtopic.php?p=336306#p336306
Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

User avatar
poshook
Posts: 10
Joined: 06 Jun 2017

06 Jun 2017

EnochLight wrote:
calmbird wrote:Thanx for responses!

I've checked with Reaper, with some multithread options turned on, and my observations have following conclusions.

I run Reaper with 14 tracks, each track includes:
Predator device,
2x RC48 Reverb,
RC24
Molot
loaded with elements Izotope Ozone 7

Up to 13 tracks everything runs smoothly.
And as you can see, task manager is showing 100% utilization.

Normaly Reason could handle up to 4 of such tracks.

I think something is wrong with way Reason utilizes those threads/cores.
Have you tried this?

viewtopic.php?p=336306#p336306
It helps a little. In my test I reached from 29 instances to 34, but still only 1/2 from what I get in Logic (72 instances) or Ableton (58)

calmbird
Posts: 12
Joined: 08 May 2017

06 Jun 2017

ShawnG wrote:would certainly concede that Reason lags behind Reaper in cpu utilization.

but 3 algorithmic reverbs PER TRACK? + Ozone?

that's no way to rack up a solid track count in any daw, hopefully this is just some sort of stress test?
Yes it's just a stress test. I was manage to run this setup in reaper almost without sound cracking.
Reason couldn't handle even 1/3 of the load.

calmbird
Posts: 12
Joined: 08 May 2017

06 Jun 2017

EnochLight wrote:
calmbird wrote:Thanx for responses!

I've checked with Reaper, with some multithread options turned on, and my observations have following conclusions.

I run Reaper with 14 tracks, each track includes:
Predator device,
2x RC48 Reverb,
RC24
Molot
loaded with elements Izotope Ozone 7

Up to 13 tracks everything runs smoothly.
And as you can see, task manager is showing 100% utilization.

Normaly Reason could handle up to 4 of such tracks.

I think something is wrong with way Reason utilizes those threads/cores.
Have you tried this?

viewtopic.php?p=336306#p336306
Yes, Reason became laggy and unresponsive. Tried both, increasing and reducing.
I also had a situation that Reason load meter was showing 20% of CPU usage, thought there was 0 devices.
Something is wrong. I hope it's not only me?

calmbird
Posts: 12
Joined: 08 May 2017

06 Jun 2017

poshook wrote:
EnochLight wrote:
calmbird wrote:Thanx for responses!

I've checked with Reaper, with some multithread options turned on, and my observations have following conclusions.

I run Reaper with 14 tracks, each track includes:
Predator device,
2x RC48 Reverb,
RC24
Molot
loaded with elements Izotope Ozone 7

Up to 13 tracks everything runs smoothly.
And as you can see, task manager is showing 100% utilization.

Normaly Reason could handle up to 4 of such tracks.

I think something is wrong with way Reason utilizes those threads/cores.
Have you tried this?

viewtopic.php?p=336306#p336306
It helps a little. In my test I reached from 29 instances to 34, but still only 1/2 from what I get in Logic (72 instances) or Ableton (58)
Lucky you, I can't even reach 1/3 of Reaper performance. I like Reason alot but really, I'm waiting for some fixes from Properellhead.

calmbird
Posts: 12
Joined: 08 May 2017

06 Jun 2017

Please check this one out.
I've just created new track, added one reverb and Izotope Ozone 7 chain.
DPS meter showing 75% cpu usage, while in Reaper was manage to add 13 Ozone instances and still everything runs smoothly.
Attachments
reason.jpg
reason.jpg (673.85 KiB) Viewed 7336 times

ShawnG
Posts: 120
Joined: 31 Aug 2015

06 Jun 2017

with those 2 plugs in the mastering section, my usage is 16 percent. my processor is an older i5, but running at 3.8 ghz

I then made a project with 13 mix channels, and filled them each with an instance of ozone 7 and one instance of RC 48 in the master section along with the usual reason master suite and send effects. and my cpu usage is 57%

what sampling rate and latency buffer is your interface using? is that sampling rate and buffer the same that reaper is running? did you go into reason's options menu and set your cpu usage limit to none, or is it still on the default 80%?

I can't directly compare with reaper on my rig for you because reaper is one of 2 DAWS (the other being FL studio) that I actively feel immense hatred towards and will not allow on my machine. :) But I have zero doubts that Reaper blows the doors off Reason in performance. some DAWS have vst handling such that a plug in that is not passing audio through will not be processed, whether its loaded or not. don't know whether reaper is one of those for sure, but it would not surprise me if it is. Reason is definitely NOT one of those, and if you have a plug loaded, it will consume full cpu resources at idle.

calmbird
Posts: 12
Joined: 08 May 2017

07 Jun 2017

ShawnG wrote:with those 2 plugs in the mastering section, my usage is 16 percent. my processor is an older i5, but running at 3.8 ghz

I then made a project with 13 mix channels, and filled them each with an instance of ozone 7 and one instance of RC 48 in the master section along with the usual reason master suite and send effects. and my cpu usage is 57%

what sampling rate and latency buffer is your interface using? is that sampling rate and buffer the same that reaper is running? did you go into reason's options menu and set your cpu usage limit to none, or is it still on the default 80%?

I can't directly compare with reaper on my rig for you because reaper is one of 2 DAWS (the other being FL studio) that I actively feel immense hatred towards and will not allow on my machine. :) But I have zero doubts that Reaper blows the doors off Reason in performance. some DAWS have vst handling such that a plug in that is not passing audio through will not be processed, whether its loaded or not. don't know whether reaper is one of those for sure, but it would not surprise me if it is. Reason is definitely NOT one of those, and if you have a plug loaded, it will consume full cpu resources at idle.
My sampling rate is 88Khz, no CPU limit and 9ms Focusrite 2i2 asio driver.

RodJustice
Posts: 13
Joined: 12 May 2017

12 Jun 2017

I'm glad to see this thread.

Props need to fix this because it's unacceptable.
Must be some glitch in the code because for some people with older CPUs everything is good.

Very odd!

calmbird
Posts: 12
Joined: 08 May 2017

19 Jun 2017

I have discovered solution that can increase Reason performance alot (still not as much as I would expect).
So in Windows task manager just set priority of reason from normal to high.

rigglebeats
Posts: 3
Joined: 20 Jun 2017

20 Jun 2017

I am also suffering with very high cpu usage, sometimes with even just one vst open,
where as with other daw's I have no problems at all, emailed propellerheads and I was just given a corporate response that did not help at all if I am totally honest,

seems like a streaming issue were reason cannot handle loads of voices at once, especially when using certain kontakt libraries in reason, NI Thrill is one that runs like total ass

I am running a 2500k that is overclocked to 4.5 ghz, 16gb ram, reason is running of a ssd,
should be more than enough to handle reason,

Does anyone know if there are going to be more updates to reason 9?, will be a lil annoyed if they don't bring out an update till reason 10 then we have to pay for the upgrade.

rigglebeats
Posts: 3
Joined: 20 Jun 2017

20 Jun 2017

calmbird wrote:
19 Jun 2017
I have discovered solution that can increase Reason performance alot (still not as much as I would expect).
So in Windows task manager just set priority of reason from normal to high.
I had a little test and it seemed like putting reason to high actually bumped up the cpu usage by about 4% for me

User avatar
sequence
Posts: 47
Joined: 20 Jan 2015

20 Jun 2017

Just to shime in that I did experience a really bad performance when using Omnisphere compared to Reaper,
I opened the same patches and dropped the same midi on both, and reason cannot even start playing without showing the computer too slow message....

Honestly I left reason when Re's came out, because I tried and got hooked on the whole new world of VST land... But always missed the reason sequencer, which I was using since version Reason 2... VST came out, so I was like yeahhhhh I can use the sequencer again..... nope.... performance really suck !!

I do hope they are aware of it, and trying to fix with the priority it deserves.

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

20 Jun 2017

sequence wrote:
20 Jun 2017
Just to shime in that I did experience a really bad performance when using Omnisphere compared to Reaper,
I opened the same patches and dropped the same midi on both, and reason cannot even start playing without showing the computer too slow message....
Well that's good to know, and a stark difference from what James Benard claimed in his recent Spectrasonics demo video about Reason performance. Sadly, I was trying to buy a used copy of Omnisphere 2 from a user over at KVR, but he/she dropped off the face of the planet after they told me they had put in a request for license transfer over 3 weeks ago. Glad I didn't pick it up, if performance truly is an issue.
sequence wrote:
20 Jun 2017
nope.... performance really suck !!

I do hope they are aware of it, and trying to fix with the priority it deserves.
Yes, they are very aware and of course working to improve it, as mentioned by Mattias over at KVR here:

http://www.kvraudio.com/forum/viewtopic ... 5#p6811115
Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

User avatar
davidvilla
Posts: 208
Joined: 16 Jan 2015

20 Jun 2017

EnochLight wrote:
20 Jun 2017

Yes, they are very aware and of course working to improve it, as mentioned by Mattias over at KVR here:

http://www.kvraudio.com/forum/viewtopic ... 5#p6811115
So that guy who ran the Slate VBC test - I did the same thing and Reason choked at ONLY 18 instances of VBC rack. Logic let me run OVER 100 and I could have added more but I got bored.


Screen Shot 2017-06-20 at 12.29.13 PM.png
Screen Shot 2017-06-20 at 12.29.13 PM.png (698.12 KiB) Viewed 6941 times

User avatar
EnochLight
Moderator
Posts: 8405
Joined: 17 Jan 2015
Location: Imladris

20 Jun 2017

davidvilla wrote:
20 Jun 2017
EnochLight wrote:
20 Jun 2017

Yes, they are very aware and of course working to improve it, as mentioned by Mattias over at KVR here:

http://www.kvraudio.com/forum/viewtopic ... 5#p6811115
So that guy who ran the Slate VBC test - I did the same thing and Reason choked at ONLY 18 instances of VBC rack. Logic let me run OVER 100 and I could have added more but I got bored.



Screen Shot 2017-06-20 at 12.29.13 PM.png
It looks like the two of you were piling on instances of Slate VBC onto 1 mix channel, correct?
Win 10 | Ableton Live 11 Suite |  Reason 12 | i7 3770k @ 3.5 Ghz | 16 GB RAM | RME Babyface Pro | Akai MPC Live 2 & Akai Force | Roland System 8, MX1, TB3 | Dreadbox Typhon | Korg Minilogue XD

User avatar
davidvilla
Posts: 208
Joined: 16 Jan 2015

20 Jun 2017

Ok I see, so using 4 different instruments, piling Slate VBC on each of them, I can do 35 instances of VBC rack. Still nowhere close to Logic though.

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 25 guests