I've never understood why aux busses seem to take such a huge toll on processing power?ccrane wrote:DP 9.13 STRESS TEST
I just installed DP 9.13 AND 9.02 on my Mac. DP 8.13 is also on this Mac.
As I have some DP projects that require a buffer of 1024 in order to playback properly, I was concerned with DP 9.13’s performance on these projects given that I can not raise the buffer any higher than 1028. Here are the results of my tests.
COMPUTER: 2012 Mac Pro 5,1 2 x 2.4Ghz 6 Core Xeon, 12 GB Ram
OS: Sierra 10.12.4
DP SEQ FILE SIZE: 14 MB
DP SEQ RAM USEAGE: 3.80-4 GB (with about 4GB system RAM to spare)
PLUG-INS IN PROJECT: WAVES, KUSH, PSP, IZOTOPE, XFER RECORDS, KEILWERTH, SPL, EVENTIDE, MDA, SOUNDTOYS, SUGARBYTES
VIs IN PROJECT: VPS, REFX, PAPEN, SYNAPSE
TRACK COUNT: ABOUT 150 (mostly audio and auxes)
TRACK’S BUSSING/ROUTING: Most tracks are bussed to one of five auxes (set A), those five auxes are then further bussed to either one of two aux inputs (set B). Finally, those two auxes are then bussed to the input of an Audio track where the Final Mix is recorded.
BUFFER: 1024 for all DP versions (max my system will allow)
RESULTS:
DP 8.07:
DP Processing Meter: 95%
Performance: Not exactly snappy but works.
CPU Usage in Apple’s Activity Monitor: Around 26%
DP 9.02:
DP Processing Meter: 90%
CPU Usage in Apple’s Activity Monitor: Around 26%
Performance: Similar to 8.13, not snappy but works.
DP 9.13:
DP Processing Meter: 80%
CPU Usage in Apple’s Activity Monitor: Around 25%
Performance: Plays, but barely.
Selecting with mouse is extremely sluggish, 5 sec or more spinning beachball for anything, including something as simple as changing the cursor type.
STEPS TAKEN TO HELP DP 9.13 BETTER PLAY BACK SEQUENCE:
1) Used Load feature to import into empty DP 9 sequence.
Result: NO PERCEPTIBLE CHANGE
2) Increased prime seconds from 125 to to 500.
Result: NO PERCEPTIBLE CHANGE
3) Deleted unused takes.
Result: NO PERCEPTIBLE CHANGE
4) Deleted all plug-ins from Master Fader.
Result: NO PERCEPTIBLE CHANGE
5) Routed ALL tracks to MASTER FADER OUT, removing all aux bussing in sequence:
Result: SLIGHT IMPROVEMENT
Still sluggish but can change curser type and scroll through input/output lists much faster. Playback is still not as smooth as DP 8.13.
6) Rerouted one CPU intensive reverb (Eventide Blackhole - or a similarly CPU demanding reverb) directly to MASTER FADER OUT:
Result: SLIGHT IMPROVEMENT
Still sluggish but can change curser type and scroll through input/output lists much faster.
CONCLUSIONS BASED ON THIS ONE TEST:
1) If you have a project that previously ran at your maximum buffer setting in a version of DP prior to 9.13, your sequence may be unable to play properly in DP 9.13.
Previous to DP 9.13, you could double the Host Buffer Multiplier in order to slightly increase performance. That feature is no longer available in 9.13.
2) As I have noticed in previous versions of DP, extensive bussing comes at a cost.
Bussing auxes to additional auxes, especially if there are intensive reverb plug-ins involved, appears to greatly increase the strain on DP.
3) Unexpectedly, the load on the CPU appears to be slightly less in DP 9.13 even though performance seems to indicate the opposite of that. MOTU tech support previously suggested doubling the buffer setting in DP 9.13 which seems to support the likelihood of increased CPU load. Not sure why DP’s and Apple’s metering suggests otherwise. Maybe there is something on my system that is skewing those particular readings.
4) As usual, Apple’s Activity Monitor shows much less strain than DP’s Processing Meter. I recently read that DP, Logic and Pro Tools are all close to each other in how many plug-ins can be run so one might assume they would also exhibit the same lower than expected Activity Monitor CPU readings - which leads me to:
5) I may need a more powerful computer.
I can have a huge template, but adding even just a few auxes just slams the performance meter and slows things to molasses.