Clips with Plugins Render Speed

WaveLab
1

I’m looking for ways to speed up my renders. I always initially render regions/cd tracks from an album’s worth of regions/tracks in a montage, and have the rendered files “open in new montage”, so the renders open placed in order in a new montage, with the rendered files named as I would like them (each render file named as the cd track marker in the original montage). Renders are to the original sampling rate.

Each clip/cd track has many different clip plugins, and some of the plugins are quite slow. Oftentimes, sampling rate is 96K or 192K, or the program time is very long, further slowing things down.

I have a 6-core Xeon. I ran a test, rendering six 96K clips with plugins from a montage as I normally do, and the time to complete was 18 minutes 30 seconds. During the render, CPU reading was about 18% throughout.

I then separated the six clip/region/cd tracks into six separate montages, moving the clip/region/cd track to the head of each montage. Then I loaded the six montages into the batch processor, and set it for 4 cores. Time to complete was 6 minutes 30 seconds. CPU reading was about 68% for most of the process.

This is a major difference to me, almost 3 times faster, and just what I’m looking for, but I need to have the rendered files open, automatically placed in order in a new montage, and be named as the original region/cd track. That’s the part that isn’t being done in the batch processor. Also separating each region/track into separate montages to accommodate the batch processor is not going to work for me.

Is there a way for Wavelab to incorporate this speed increase somehow, invisibly behind the scenes, so that I can render regions/tracks “normally” as I do from one montage, but utilize the speed gain from using more cores, if I’m only rendering regions/cd tracks? The difference is so large I’m going to have to take advantage of it somehow, if it can’t be optioned into the Wavelab montage regions/tracks renders in some way.

ps. I initially set the batch processor to “all cores”, but all 6 cores maxed out and CPU sat at 100%, which freaked me out, so I set it down to 4 cores. Is it ok to run at 100% if it doesn’t crash? It seemed to be working ok until I cancelled it.

pps. Lastly, are there any disadvantages to running with the extra speed, at about 68% CPU, besides having less available CPU for other simultaneous tasks?

2

Did you try the WaveLab 9 multi render feature? (control + click on the tabs you want to render).
However, not all options are possible in this mode.

Is it ok to run at 100% if it doesn’t crash?

Yes, if you yout CPUs have good fans.

Lastly, are there any disadvantages to running with the extra speed, at about 68% CPU, besides having less available CPU for other simultaneous tasks?

No. But the External Gear plugin can’t be used.
Also, since more plugins get instantiated, you may run into limits eg. with UAD plugins.

3

Thanks PG. I tried the multi render feature, and see what you mean about it offering multicore support, but it seems only multiple montages or multiple audio files would benefit from that (unless I’m missing something).

I have only one montage. I need to run each of the (say) 12 clips/cd tracks in that montage in separate tasks. Like I said, I can separate each of those clips/tracks into 12 separate montages, name the montages as the cd track, run all of them in the batch processor using multi tasks or cores, then open the renders in a new montage in order, but all of that is a lot of extra work.

I just tried my test again, except using all 6 cores this time, and my time to complete was 4 minutes 50 seconds, which is amazing. Unfortunately, render time has become one of the major issues here, and the fact that Wavelab can deal with multiple tasks and cores in the Batch Processor is amazing. I just need to find a workable way to use that. I can’t continue to render in one task at 18% of the computer’s speed, when I want to be able to render at 80 or 90%.

If Wavelab can’t be easily changed to do this, and since Wavelab can already do what I need (with a lot of extra work), it seems like I should be able to do this with Wavelab scripting (if I knew Wavelab scripting). But basically the process would be:

  1. Create separate temporary montages for each of the clip/cd tracks, and move the clip/cd track to the head of each montage. Name the montages after the cd track names.
  2. Add all of the temporary montages to a batch process set for 5 or 6 tasks or cores.
  3. Add the result files, in the original order if at all possible, to a new montage.

Is that a possibility?

re: External Gear plugin, I don’t use it.
re: UAD, I’m using 3 UAD plugins on each of these clips and have no problem with 6 instances of each being open during the processing.

4

I see the need for this, and I have added it to a future plan. Since WaveLab uses multi processing in various contexts, that would be coherent.

This being said, scripting won’t help you today, to achieve your alternate workflow.

5

Thank you Philippe. I really appreciate it.

6

PG, I found I can get this same speed increase by starting each region/track render separately from the montage, one after the other, using the “one region” selector in the render dialog. I previously thought this wouldn’t give me any speed benefit (I thought the separate tasks would split up over the 18% CPU usage), but instead it seems they’re additive up to 100%.

Because I’d like to use this as my temporary solution, I wanted to ask how many tasks I can safely add to the render queue from the montage. Can I start 12 region/track tasks one after the other and they’ll just take care of themselves, CPU wise? Will the montage render queue accept 12 or more tasks? This method isn’t ideal for me in many ways (having to number the tracks to keep them in order, manually add to a new montage, and then remove the numbers because I don’t want them at this point), and I’d rather limit to 5 cores than let it use all 6, but the speed difference is so compelling I think I’m just going to have to do it this way until you’ve incorporated changes into the program. Thanks again for offering to do that. I really doubt that other programs are as versatile as Wavelab in handling things like this.

7

There is no internal limit. It’s up to your ressources, number of cores, memory, UAD limits, etc.
Hence this is up to you to find.

8

This is all very interesting though for me, standard rendering speed seems OK on my iMac 3.5GHz Quad and I always prefer the “Multiple Outputs” rendering of each track in one command so that the numerical prefix can be applied automatically vs. doing it manually for every track among other things.

@Bob: If it doesn’t mess up anything else, you could batch rename your track markers to contain the numerical prefix so that even if you render the CD tracks one at at time, they end up with the numerical prefix automatically.

I think I tested this recently and even though WL can’t compute the numerical prefix when you render one region, it still adds the correct track number/track total in the ID3v2 metadata so that’s good.

I will have look into the “control+click” option for rendering to see if it helps any of my workflows as I didn’t know it was there.

Even though my render speeds are OK (UAD slows things down though), I would always welcome a way for WL to naturally do faster rendering.

9

The speed might be fine, what you’ve gotten used to, as I did, rendering the whole album as regions/cd tracks in a fairly reasonable amount of time. But the speed can be 3,4,or more times faster if the changes are put in place, or if done manually. I only got the speed complaint from someone else. Which made me look at the CPU utilization during the render, and finding it was only running at the speed of one core/task, if that’s the right terminology (as it does in other programs, like Reaper when rendering regions/tracks. I’ve yet to find a workable way around that in Reaper either, besides possibly adding all one at a time, which I haven’t tried yet. It might be do-able, but I haven’t found it yet).

I would think you would probably see CPU at about 25-30% during the render if you have a Quad core.

I’ve mostly looked at the single thread rating of processors in the past when buying, thinking most operations would be as single tasks, but if things like this can really benefit from more cores, as multiple tasks, I would probably consider even more cores next time. I don’t think there could be any speed benefit when rendering to one big long file, which I think has to run as a single task (and run at the slower speed), but these other cases where renders are to multiple files can have a huge speed benefit I think.

Thanks Justin. Sorry, I neglected to say I’ve made batch marker rename presets to add numbers and to remove numbers, so it’s pretty easy. It really is amazing what you can do with Wavelab.

10

Thanks Justin. Sorry, I neglected to say I’ve made batch marker rename presets to add numbers and to remove numbers, so it’s pretty easy. It really is amazing what you can do with Wavelab.[/quote]

Yes. 70% of the time when I have a feature request for WaveLab, it’s already there.

11

I am supporting of this future feature (since I also usually render out 75 clips from one montage).

Having that be parallel rendering with multtiple course would save me TONS of time.