Dewdman42
Senior Member
I wanted to post these test results for everyone's FYI and any further discussion which might come out of it.
The main purpose of this test was to take a real world orchestra mockup and play it using several different DAW's, both with and without VEP7, to compare them to each other, and also to see the impact of using VEP together with a DAW in terms of performance.
UPDATE: Cubase 10.1.30 drastically improved performance on the mac, see new results here now.
Full Spreadsheet here:
Procedure
Summary
Most of the tested situations averaged around 35% cpu usage over the test. In the graph above you can see a cluster of graph lines, some slightly better or worse then others, but all reasonably close and acceptable in performance, both with and without VEP. DP was the worst performer, but only slightly; LPX performed significantly better then everything else with and without VEP, but especially with the new AU3 version of the VEP plugin. In general, everything performed slightly better with VEP than without, except for DP9 which performed better alone, probably related to pregen feature. (See below)
System Specs
Here are some additional tests that would be interesting to do
The main purpose of this test was to take a real world orchestra mockup and play it using several different DAW's, both with and without VEP7, to compare them to each other, and also to see the impact of using VEP together with a DAW in terms of performance.
UPDATE: Cubase 10.1.30 drastically improved performance on the mac, see new results here now.
Full Spreadsheet here:
Procedure
- 90 track score from VSL's website was used as test project. This contains 90 tracks of ViPro, with MirPro on each track and Miracle reverb on the master bus.
- The same 90 track score was used in all cases, and every attempt made to optimize each DAW to similar optimized settings
- Audio Buffer of 1024 used in all cases.
- Score was played through same section of score, for 3 minutes, sampling the average CPU % every 3 seconds, using following command line, started at same point of score playback:
Code:
iostat -w 3 -I -c 60
Summary
Most of the tested situations averaged around 35% cpu usage over the test. In the graph above you can see a cluster of graph lines, some slightly better or worse then others, but all reasonably close and acceptable in performance, both with and without VEP. DP was the worst performer, but only slightly; LPX performed significantly better then everything else with and without VEP, but especially with the new AU3 version of the VEP plugin. In general, everything performed slightly better with VEP than without, except for DP9 which performed better alone, probably related to pregen feature. (See below)
- LogicPro alone: 25% average cpu usage.
- LogicPro+VEP-AU3: 24% average cpu, this is the clear winner in terms of CPU usage!
- StudioOne+VEP 33% average cpu usage
- StudioOne alone: 35% average cpu
- Cubase+VEP: 33% average cpu usage.
- Cubase10 alone: 34% average cpu usage
- DP9+VEP: 38% avg cpu
- DP9 alone: 35% average cpu. Audio buffer at 2048. Also tried DP10 which performed 5% worse then DP9 and the GUI as extremely laggy compared to DP9, so there is that.
- Reaper5+VEP: 35% avg cpu
- Reaper Alone: 33% avg cpu
System Specs
- MacPro 5,1 12 cores x 3.33ghz, 128gb ram, OSX 10.14.5 (Mojave), RX580 video
- LogicPro 10.4.4
- Cubase 10.0.30
- Studio One 4.5.1
- DP 9.52
- Reaper 5.978
- VEP 7.0.826
- Audio Buffer at 1024
Here are some additional tests that would be interesting to do
- How many tracks can be added before it starts dropping out audio, at various buffer sizes.
- All of the above at lower audio buffer sizes
- A test using generic instruments so that everyone can try the same test on their DAW, instead of using a real world cue, try to max out the max number of tracks in each case to see how big each DAW scenario can go before running out of steam with dropouts.
- How low latency can each scenario go before getting drop outs
Last edited: