Jump to content
irvin

CPU Use by different DAWs

Recommended Posts

Puzzled by the gigantic difference between DAWs when it comes to plugin use and CPU consumption.

Testing conditions: single track, 24-bit, 48Khz mono file imported.

DAW A uses 2.2% CPU when Plugin Z is engaged.

DAW B uses 19%CPU when plugin Z is engaged.

Chose plugin preset - not tweaking at all of anything else.

Is it the plugin? is it the DAW?

Share this post


Link to post
Share on other sites

If it´s exactly the same plugin version it´ll have to be the DAW. Which ones are you talking about or did you test? Are the projects in both DAWs exactly the same, like just one track with just one plugin or so? What´s the CPU load of the DAWs without any plugins?

Share this post


Link to post
Share on other sites

Same settings. Maybe we can all work on a simple, quick experiment that would shed a lot of light on this issue. The steps would be very simple:

1. Download multitrack mix from Mike Senior's site. Select file to use.

2. Create a single-track project at file's native bit and sample rates (24/44.1)

3. Insert plugin we all have

4. Record the performance

I have access to Studio One, Samplitude Pro X and Reaper. We would need to settle on a common plugin (I have Phoenixverb, for example, but others may have a different one) and just use it as a track insert with a specific preset running at 100% (no tweaking at all)

While different computers would report different CPU consumption, that's not important. The important part is the different consumption rates within the same computer.

What do you think?

Share this post


Link to post
Share on other sites

1. Download multitrack mix from Mike Senior's site. Select file to use.

-> no need to download if anyone doesn´t want to, as long as any audiofile has same bit/samplerate all is fine

2. Create a single-track project at file's native bit and sample rates (24/44.1)

-> you mean sample rate, right? Sam cannot create a VIP with a certain bitrate, it´ll read 16/24/32float by itself.

3. Insert plugin we all have

-> now that could be a difficult one. Who shall start posting his list?

I can offer Sam and Seq 11 and 12, from monday on Seq 13. Also Harrison Mixbus (totally useless btw). That´s it.

My plugins:

all Softube (32/64bit)

all Flux (32bit)

all IRCAM (64bit)

all Slate Digital (32/64bit) (exept the Drum stuff)

SPL Passeq (32bit)

VOS Thrillseeker (32bit)

all Audiocation (32/64bit)

Eventide Ultra Channel (32/64bit)

VeeMax (32/64bit)

Waves Masters Bundle (32/64bit)

Waves JJP Native Bundle (32/64bit)

all Breebaart (32bit)

Digital Fish Phones (32bit)

Lawo Collection (32bit)

Lexicon MPX Native Reverb (32/64bit)

Molot 03 (32/64bit)

NI Komplete 9 (32/64bit)

bx cleansweep V2 (32/64bit)

bx solo (32/64bit)

Share this post


Link to post
Share on other sites

I suggest we wait until Monday, to see who joins the test. Actual testing would take less than 30 minutes. I find all your suggestions sensible. Let's wait for other members' input.

Share this post


Link to post
Share on other sites

Single instance of reverb (B2 by 2C Audio) CPU consumption using the Dual Engine (4 Blessings) preset without any modifications on a mono, 24-bit, 44.1 k file:

Samplitude Pro X: 19% (19/20% MAX)

Studio One: 18%

Reaper: 2.3 %.

Multiple instances of the plugin result in an even bigger discrepancy: for example, 2 instances results in 40% DSP (whatever that means) use in Samplitude versus 4.6% CPU use in Reaper. Studio One shows 36%.

****************************************

Increasing the buffers within the audio interface results in better performance (at the cost of added latency), but not a really significant difference. Perhaps a non-optimal configuration of Studio One and Samplitude on my part?

I'd like to hear from other users and other DAWs.

Share this post


Link to post
Share on other sites

Hi Irvin,

you can't compare DSP % measurements and CPU % usage. I understand that you seem to have difficulties understanding what DSP % means (from your comment on the Pan issues form today, which was a bit off-topic to say the least), but I understand as well that you know that it is not the same.

If you would be interested in apple2apple figures, you could have used resmon.exe.

Besides that, running a single VST on a single track input is not telling a lot about the efficiency of a DAW on a multi-core, multi-threaded modern computer. Check out bench mark tests run by ADK and others.

BR,

Sebastian

Share this post


Link to post
Share on other sites

...running a single VST on a single track input is not telling a lot about the efficiency of a DAW on a multi-core, multi-threaded modern computer.

That´s not the question in post Nr.1. The questions is: How come that the same plugin uses different CPU power when used in different DAWs. The question is not about the DAW itself, nor about CPUs. So it´s a relative comparison.

But you´re right with:

...you can't compare DSP % measurements and CPU % usage...

...If you would be interested in apple2apple figures, you could have used resmon.exe

Share this post


Link to post
Share on other sites

So it´s a relative comparison.

Still, there is the question if it is a meaningful comparison. It's a fact that a pig weighs less than an oxen. A Vegetarian will eat neither animal, so he will not care.

-S

Share this post


Link to post
Share on other sites

We´re talking about 2 DAWs that are loading the same plugins. So, if you want, there are 2 vegetarians eating pigs or oxens, for whatever reason, completely unimportant. The question is: Why does one of them gain more weight, the other one less.

flower-talk-mode: off

Share this post


Link to post
Share on other sites

Hi Irvin,you can't compare DSP % measurements and CPU % usage. I understand that you seem to have difficulties understanding what DSP % means (from your comment on the Pan issues form today, which was a bit off-topic to say the least), but I understand as well that you know that it is not the same.If you would be interested in apple2apple figures, you could have used resmon.exe.Besides that, running a single VST on a single track input is not telling a lot about the efficiency of a DAW on a multi-core, multi-threaded modern computer. Check out bench mark tests run by ADK and others.BR,Sebastian

No need to get too technical - let's put it this way: samplitude chokes on 6 instances of plugin x, while reaper hums along with 10 of them.

Can you understand that?

Share this post


Link to post
Share on other sites

Ironically, there's this thread (almost identical in content) at KVR:

http://www.ikmultimedia.com/forum/viewtopic.php?f=12&t=8755

Posted by none other than Kraznet, the well-known "Samplitude videos" guy. He makes the same "mistake":

*****************

I've just been doing some tests using SampleTank 3 inserted to a single track with 16 instruments loaded

There is a massive disparity of CPU usage with Samplitude and Studio One compared to Reaper. This is just with SampleTank sitting there without any recording taking place.

Samplitude - 44% (51% max)

Studio One 51%

Reaper 5.07% !!

Is Reaper really that efficient?

***********************************

The bottom line is no matter how technical we get (and we shouldn't, this is not a geek forum, but a user forum for people using a simple DAW, a forum for musicians), the observed disparity in CPU usage is a legitimate issue.

Share this post


Link to post
Share on other sites

Samplitude - 44% (51% max)

Studio One 51%

Reaper 5.07% !!

Is Reaper really that efficient?

Or does Reaper maybe not do the job right?

Actually, the right way to do it is to use ResMon. Check CPU load of the DAW without the plugin, then check it with the plugin loaded. The difference is the number you need.

Share this post


Link to post
Share on other sites

No need to get too technical - let's put it this way: samplitude chokes on 6 instances of plugin x, while reaper hums along with 10 of them.

First of all, don't oversimplify. Second, what you are trying to find out is technical, no? Third, what you observe now can't be shown, or does not principally relate to the performance figures you find from a single instance.

Share this post


Link to post
Share on other sites

observed disparity in CPU usage

Well, I tried to make clear that the actual observation you made is that different DAWs have different performance metrics. I am not sure if I want to go deeper than that as my observation was that you have already been jumping to conclusions.

BR,

Sebastian

Share this post


Link to post
Share on other sites

Samplitude - 44% (51% max)

Studio One 51%

Reaper 5.07% !!

Is Reaper really that efficient?

Or does Reaper maybe not do the job right?

Actually, the right way to do it is to use ResMon. Check CPU load of the DAW without the plugin, then check it with the plugin loaded. The difference is the number you need.

I did extensive testing on this subject and was left without any doubt that Reaper uses some kind of voodoo when processing SampleTank 3 compared to nearly all the other DAW's. Although this seems to be isolated to SampleTank because when I did similar tests using Kontakt 5 not only was the CPU usage in Samplitude considerably lower (around 2-4%) but also the difference between CPU usage in Samplitude compared to Reaper was very similar. Anyway I did my tests, reported them to SamDev and moved on.

Regards

Kraznet

Share this post


Link to post
Share on other sites

observed disparity in CPU usage

Well, I tried to make clear that the actual observation you made is that different DAWs have different performance metrics. I am not sure if I want to go deeper than that as my observation was that you have already been jumping to conclusions.BR,Sebastian

Yes, I have been 'jumping to conclusions': I concluded that Samplitude is not that efficient compared to Reaper. Not too difficult to determine based on simple observation: whenever the 2 DAWs use the same number of plugins, Samplitude comes to its knees way before the other one. And no, there is no need to get too technical, especially since none of us knows nearly enough to determine exactly what the reason is - we are all really guessing. It's not like using Performance Monitor will clarify much...let's get real...lol

In the end, it may just be that you guessed right when you have repeatedly stated that Samplitude is an 'old' application and that nothing short of a full re-write of the code will do. Very damaging statements on your part (at least to Magix), but it may be true. In any case, the question remains un-answered, as does the one pertaining to the mysterious way Samplitude reports CPU usage. So much so, that even well-respected Magix evangelist Kraznet is just as confused and intrigued as I am....

Share this post


Link to post
Share on other sites

Samplitude - 44% (51% max)

Studio One 51%

Reaper 5.07% !!

Is Reaper really that efficient?

Or does Reaper maybe not do the job right?

Actually, the right way to do it is to use ResMon. Check CPU load of the DAW without the plugin, then check it with the plugin loaded. The difference is the number you need.

I did extensive testing on this subject and was left without any doubt that Reaper uses some kind of voodoo when processing SampleTank 3 compared to nearly all the other DAW's. Although this seems to be isolated to SampleTank because when I did similar tests using Kontakt 5 not only was the CPU usage in Samplitude considerably lower (around 2-4%) but also the difference between CPU usage in Samplitude compared to Reaper was very similar. Anyway I did my tests, reported them to SamDev and moved on.

Regards

Kraznet

There is a very simple test: keep adding instances of any plugins and see which application chokes first. Samplitude does, by a large margin, regardless of plugin type or name.

Share this post


Link to post
Share on other sites

It's not like using Performance Monitor will clarify much...let's get real...lol

Well, have you even tried?

As far as "getting real" is concerned, I would like to ask in how far the "number of plugins" is a real-world related scenario. A real-world scenario would in my opinion include a complete project layout, including the use of plug-ins on Objects, Tracks, Buses, and Master, plus the use of VSTis.

Most people will opt for another method: They will simply use the DAW, and see if they like it. That's the non-technical approach.

For any "semi-technical" assertions: No, thanks!

In any case, the question remains un-answered, as does the one pertaining to the mysterious way Samplitude reports CPU usage

Samplitude doesn't report CPU load. The fact that you keep saying that allows for the conclusion that you don't understand either approach, DSP% and CPU%. Explaining the difference is then some kind of "mission impossible".

BR,

Sebastian

Share this post


Link to post
Share on other sites

It's not like using Performance Monitor will clarify much...let's get real...lol

Well, have you even tried?As far as "getting real" is concerned, I would like to ask in how far the "number of plugins" is a real-world related scenario. A real-world scenario would in my opinion include a complete project layout, including the use of plug-ins on Objects, Tracks, Buses, and Master, plus the use of VSTis.Most people will opt for another method: They will simply use the DAW, and see if they like it. That's the non-technical approach.For any "semi-technical" assertions: No, thanks!

In any case, the question remains un-answered, as does the one pertaining to the mysterious way Samplitude reports CPU usage

Samplitude doesn't report CPU load. The fact that you keep saying that allows for the conclusion that you don't understand either approach, DSP% and CPU%. Explaining the difference is then some kind of mission impossible.BR,Sebastian

You keep trying to goad me into a fuzzy pseudo-technical debate, but quite frankly I'm not interested...lol...

Simple fact is: the way Samplitude reports CPU consumption (or whatever you want to call it, we all know what we are talking about) is cryptic at best. Don't blame users for that - hell, even Kraznet (a guy who has dedicated countless hours to the program) does not know what the numbers mean! The issue is real...

And yes, I do maintain my very simple, easy-to-understand position: Samplitude shows a significantly lower capacity for running plugins when compared to other applications. If you can't explain it (or even accept it) it's fine - let other more knowledgeable people contribute to the discussion. I don't know, you don't know. That's ok.

Share this post


Link to post
Share on other sites

If DSP and CPU is that much different, then maybe someone should clarify that in the manual, otherwise misunderstandings are pre-programmed.

Here's one from the Sam ProX manual, page 479:

Performance display: This option activates the DSP performance display in the lower left corner. If the display shows values of 100% or more, then the processor is overloaded.

As far as I'm aware processor and CPU are the same, and please don't try and come up with some BS-arguments against it, I'm talking about a common understanding.

Share this post


Link to post
Share on other sites

Simple fact is: the way Samplitude reports CPU consumption (or whatever you want to call it, we all know what we are talking about) is cryptic at best.

Samplitude doesn't report CPU use, and it is also good that it doesn't because CPU use is not directly related to the performance of a DAW. I.e., running a single signal through a DAW running on a Computer with 8 threads will result in max. 12,5% CPU load. Any little bit higher will result in drop-outs. Samplitude reflects that circumstance by displaying app. 100% DSP load in that situation, because there is no way to use any other thread for the task (that is running the processing on track, Master, or Bus). Would you distribute the same processing load to two tracks, you would see half the DSP% figures. Therefore, the way Samplitude does measure load is very much appropriate, while a CPU% metric would understate the situation.

[i tired the "mission impossible", hope you appreciate.]

Don't blame users for that

My complaint you are juggling with numbers you don't understand, and which you declare that you don't understand them. That attitude is not ok, and neither is the attempt to defend your attitude by the allegation that your were "prevented" to understand by "mysteries". That's your problem, not MAGIX', not mine, not anybody else's.

Samplitude shows a significantly lower capacity for running plugins when compared to other applications.

Fair enough. However, your contributions add very little insight to your respective observation. My observations don't concur to that directly. I have seen worse, same, and better performance in Samplitude and other DAWs.

-S

Share this post


Link to post
Share on other sites

Well said, Matfle. Welcome to the 'confused club'. I suspect *all* Samplitude users belong to it, even those pretending to know better...lol

Share this post


Link to post
Share on other sites

If DSP and CPU is that much different, then maybe someone should clarify that in the manual, otherwise misunderstandings are pre-programmed.

The manual should not try to explain "differences", but what Samplitude does. Explaining differences is making assumptions about how others do things. That has a risk of being wrong.

As far as I'm aware processor and CPU are the same, and please don't try and come up with some BS-arguments against it, I'm talking about a common understanding.

Actually, you will see that the processor is overloaded. But you may see that the CPU load is less than 10% at the same time. This is because there is an additional overhead of Scheduling, and also of multi-tasking the user-land process cannot do much about.

-S

Share this post


Link to post
Share on other sites
Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...