Reaper - Long recording Test Results.....

[LEX]

I have begun the Reaper 3.63 Long recording test.

System Specs:

MAC Mini - Core Duo 2ghz Intel
2 gigs RAM
OS - Leopard 10.5.8

Record Drive:
FW 400 via FW800 to FW400 cable

Recording specs:
48k, 24bit
64 mono tracks
Waveform recording on.

LEX

[LEX]

Okay, so I am retesting.

I forgot I was testing. 3 hours 45 minutes later it was still recording.

I'm re recording an hour now to ensure what results I saw were in comparison.

More soon.

LEX

[LEX]

I don't know what to say, so you say it. Its even better than PT.

The file labeled "Forgot Reaper", I forgot it was still recording.

So I felt it best to retest.

"At Stop" is after I stopped the recording, not before.

It is shocking!

Don't think Nuendo or Cubase is going to come close to what Reaper can do in Long recordings.

I'd suggest even that Hugh, Get Reaper and not worry about it.
Though he might not even be able to import the tracks.


LEX

[LEX]

Here is the Reaper Files Folder.

LEX

[John Lance]

So these folks figured out how to get exactness along with low system resources.

Is this system you are testing on capable of 128+ tracks at the sample rates already specified in the N5 thread? I would like to see an apples to apples (relative) test on these issues between the products.

[LEX]

So these folks figured out how to get exactness along with low system resources.

Is this system you are testing on capable of 128+ tracks at the sample rates already specified in the N5 thread? I would like to see an apples to apples (relative) test on these issues between the products.

I wanted to keep my testing sessions consistent with the number of tracks.

LEX

[TAFKAT]

LEX,

Can you just note the results please..

Clicking on the screen shots and then poking around for the result is O.K, if you know exactly what you are looking at, but it would be better just to note them IMO.

I have no idea what all of the different references are supposed to represent , so can we just have some side by side numbers compared to Cubendo , Please.

Thanks

[TerryG]

I wanted to keep my testing sessions consistent with the number of tracks.

LEX
?
I don't understand.

If the goal here is determining systems limitations and software efficiency (as it should be), we're going to need testing with 128 tracks @96K in PT and Reaper in both Mac and PC to compare with existing tests in Cubendo.

[LEX]

Reaper - 128 tracks, 48k, 24bit

Reaper remains steady at 675meg real memory, and 1.6 gigs VM throughout record

Reaper Playback is 1.06gigs real memory and 1.95 gigs VM.

Terry, this test system doesn't have a 96k interface attached plus I need to free up more space to test 96k on a few drives.
I haven't had the time to transfer and backup drives.

I also want to compare with Hugh's original test and collect numbers based on 48k as I think more people will be recording 48k than 96k.

LEX

[efernan]

Hmm... Reaper doesn't look too good... it's worst than Cubase/Nuendo 5!

Seems that Cubase/Nuendo 3 (maybe 4.x also) or PT is the one to use for (long) live recording.

Cheers!

[LEX]

Hmm... Reaper doesn't look too good... it's worst than Cubase/Nuendo 5!

Seems that Cubase/Nuendo 3 (maybe 4.x also) or PT is the one to use for (long) live recording.

Cheers!

Actually its not. Cubase threw a 2 many tracks error on the first run. The reaper RAM footprint is lower than Cubase.

Cubase CPU usage is through the top in comparison to Reaper. Cubase test will be posted soon.

LEX

[John Lance]

Hmm... Reaper doesn't look too good... it's worst than Cubase/Nuendo 5!

Seems that Cubase/Nuendo 3 (maybe 4.x also) or PT is the one to use for (long) live recording.

Cheers!
Reaper is doing quite well in tests here. I've got a project in Nuendo that I'm "roughly" moving over to Reaper and so far at 44.1K with this project:
Nuendo4.x: 423 Megabytes of memory in use.
Reaper 3.x: 154 Megabytes of memory in use.

Cubase should be similar to Nuendo.

It is "roughly" because at this point I'm not using plugin for plugin, but my experience so far is that Reaper itself is comparatively light on resources. I am looking to install Reaper on a laptop due to this.

With my own recording tests on the same hardware with Nuendo and Reaper, Reaper 3.65 was able to out do Nuendo4 on a practical basis with Nuendo doing 60 tracks and Reaper doing 128, both near maxing out and at a 96K sample rate. That says a lot to me. I'm not going to reproduce the results tables here (I didn't copy them all down, just did the tests to my own satisfaction).

[Audiocave]

You guys are aware that Cubendo's memory resources per channel have a lot to do with the actual channel? You cannot have all of the things that Cubendo has immediately available - per channel - without burning some memory. Of course Reaper will be more memory efficient with lots of channels. You can't have your cake and eat it also.

The inspector alone probably adds a certain high kb amount or more of memory to each channel. Even thought it follows selection, those settings - per channel - certainly are stored in memory for each channel. Honestly guys, this comparison is not apples to apples. Things like being able to re-arrange automation lanes (something Reaper can't do) certainly also takes memory space. There are lots of things in the Cubendo UI that Reaper doesn't have to reserve memory for.

It's very cool because it's a resource efficient host but making comparisons to things like Cubase aren't exactly valid comparisons unless you want Steiny to do away with the inspector and all of the other per-channel customizations that burn memory. Add 3 different panning options - per channel - to Reaper, and it's memory use will probably creep up a bit.

[efernan]

Hi,

I was talking about the RAM used for caching the peak images. Not the resources the base program uses itself. It doesn't matter if an empty project in Reaper consumes less RAM, if it uses a lot of memory while doing long, big live recording. I don't have the time now, but from memory it looks it uses bigger peak files, and most probably it caches all of them in RAM much like any other DAW application. If you can, check how much memory Reaper uses progresively while recording 128 tracks / 48 kHz (or what you're using for comparing the apps).

BTW, here we have two separate issues: one is how much RAM a DAW app uses for caching the peak images (and so limiting the extension of long, big live recordings) and the other is the how efficient is the I/O disk data transfer and so the possibility of doing these huge recordings or not.

Cheers!

[LEX]

Hi,

I was talking about the RAM used for caching the peak images. Not the resources the base program uses itself. It doesn't matter if an empty project in Reaper consumes less RAM, if it uses a lot of memory while doing long big live recording. I don't have the time now, but from memory it looks it uses bigger peak files, and most probably it caches all of them in RAM much like any other DAW application. If you can, check how much memory Reaper uses progresively while recording 128 tracks / 48 kHz (or what you're using for comparing the apps).

Cheers!

I've stated this already. You really need to read the thread and check my snapshots.

Reaper's memory footprint does NOT increase through the recording. It stays at around 665 megs through the entire recording.

Read back through, get caught up and comment.

LEX

[kdm]

It doesn't matter if an empty project in Reaper consumes less RAM, if it uses a lot of memory while doing long big live recording. I don't have the time now, but from memory it looks it uses bigger peak files, and most probably it caches all of them in RAM much like any other DAW application. If you can, check how much memory Reaper uses progresively while recording 128 tracks / 48 kHz (or what you're using for comparing the apps).

Cheers!

That's what Lex tested and it was far lower than Cubase/Nuendo's.


The inspector alone probably adds a certain high kb amount or more of memory to each channel. Even thought it follows selection, those settings - per channel - certainly are stored in memory for each channel. Honestly guys, this comparison is not apples to apples. Things like being able to re-arrange automation lanes (something Reaper can't do) certainly also takes memory space. There are lots of things in the Cubendo UI that Reaper doesn't have to reserve memory for.

It's very cool because it's a resource efficient host but making comparisons to things like Cubase aren't exactly valid comparisons.

The track inspector and other other GUI factors don't cause Cubase's ram usage to climb during a long record session - peak files are the issue and they are 4x larger than Nuendo 2/3. That's the problem - it limits large live recordings some users require (and could limit us on large post projects as well).

These tests are all about finding the most efficient way to get the job done, not compare apples to apples in terms of applications' general day to day ram usage.

[Audiocave]

The track inspector and other other GUI factors don't cause Cubase's ram usage to climb during a long record session - peak files are the issue and they are 4x larger than Nuendo 2/3. That's the problem - it limits large live recordings some users require (and could limit us on large post projects as well).

These tests are all about finding the most efficient way to get the job done, not compare apples to apples in terms of applications' general day to day ram usage.

How does this all work if you turn off "Create Audio Images During Recording"? Thanks.

[kdm]

How does this all work if you turn off "Create Audio Images During Recording"? Thanks.

Read back - I think Lex or Terry tested that. Doesn't really help if you need to edit or work soon after the recording.

[John Lance]

I should have further qualified my statement above when giving my own track count limitations.

It is not the memory footprint in this case that kept the channel count from increasing in my own tests. The real issue in my testing was the i/o activity on the one hard drive I was using for recording, and that along with Reaper's and Nuendo's usage of that resource became the limiting factor.

Nuendo has much more i/o to the hard drive for the relative same conditions, and others here are noting ram usage issues over time.

[TerryG]

How does this all work if you turn off "Create Audio Images During Recording"? Thanks.

In a 32-bit system, you can record far longer in Cubendo with "Create Audio Images During Recording" disabled, but if you exceed the time you would have been able to safely record with it enabled, as soon as your session is finished you'll be in serious trouble.
You won't have enough RAM to cache the Peak files as they're being generated, and your system will crash, possibly corrupting the entire project.
The only recourse will be to import the audio files into a 64-bit system with at least enough RAM to generate the peak files (because they need to be generated before you can accomplish anything -editing-, etc) Then, you'll need to finish the sessions in 64-bit, or weed out enough of the pool through editing/discarding tracks that it becomes safe once again to load into a 32-bit system.
So, the best policy is to always record with "Create Audio Images During Recording" engaged, then monitor your RAM useage and stop recording before you exceed your RAM limitation. Anything beyond 1.6GB for the Cubendo.exe seems to be the point where 32-bit systems risk session failure.

[efernan]

Hi again,


I've stated this already. You really need to read the thread and check my snapshots.

Reaper's memory footprint does NOT increase through the recording. It stays at around 665 megs through the entire recording.

Read back through, get caught up and comment.

LEX

Sorry, I got confused with the test project and its settings. There are a couple of threads about this and I got them somewhat mixed up. You know... 48 kHz versus 96 kHz, 64 tracks versus 128 tracks, etc.

Based on the peak files Reaper generates it seemed at first sight that it will use more memory than Cubase/Nuendo 3 (and not really worst than Cubase/Nuendo 5, but about the same), but... I've done a quick test and discovered that Reaper doesn't use cached peak images! As you stated, Reaper memory footprint stays pretty constant during recording. I see it grow but it seems not related to the peak images. Got to dig into it more.

So it looks like Reaper is very efficient for its own memory usage compared to Cubase/Nuendo (yeah, no news).

Anyway, I guess that all this live recording subject need further investigation. I'm sorry I don't have the time now for participating. I have to deliver two new systems tomorrow morning! And they are for producing music for advertising. You know... very long projects of, ehem, about, ehem, 15-60 seconds.

Cheers!