Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
CPU usage on mono vs stereo
#1
Hey there, I just came to the limit on my Mac Pro 6 Core with 53 stereo tracks. Same behaviour on C32 V6 and V8, 2048 buffer, all set for big projects.
Some of them are bounced mono tracks from the client, so this rises the question about how much I could save by converting them (back) into mono.

Having had a CPU crackles issue for the first time in 12 years on Mixbus I also wonder about the actual CPU window, because it still doesn't show very much usage.
Is it saved/held back for eventual full use of eq and comps?

I'm asking for expertise here, not opinions.
Mac Pro 5,1 | 6x 3,4Ghz | 48GB | OS X 10.14  | Macbook Pro M1 | 16GB | OS X 14.4 | Metric Halo 2882 3d 
http://www.sounddesign-pro.com
Reply
#2
How do you know that CPU causes the cracks and not DSP?
If MB does ot have enough speed/cpu to transfer data it tells so by a popup note.
Mono or stereo tracks are just half info. The load mainly on DSP comes from stareo plugins that need more power.
For example one Waves Abbey Road Chambers or Plates stereo plugin can increase DSP near by 20% that would need 50-60 plus tracks to hit the same amount.

I do not know if it is possible in Mac but on Windows there are small CPU Unpark programs that make the Cores all actve and need not wait for the OS to tell the cores to wake up.
Not opinion, experienceSmile
best
Tassy
Win7/64, Mixbus32C, Mixbus2.5 the QueenSmile UR22, Dynaudio BM5A MKII, Pc all SSD,
Reply
#3
No PlugIns involved, just the plain Mixbus setup, nothing else- but I found the answer to my second question in the manual:
Quote: • Like a hardware mixer, the DSP horsepower for EQs and compressors is preallocated. Enabling a compressor or EQ will not increase the CPU usage of the computer.
That explains why only 15% DSP usage (which equals 150 % on ten virtual cores) already create crackles.

So, to specify my first question:
How is the CPU drain ratio (and therefore DSP allocation) between a mono track and a stereo track. It's surely not 1:2 or 1:1 but something in between.
However, converting the mono tracks on my project did the job.
Mac Pro 5,1 | 6x 3,4Ghz | 48GB | OS X 10.14  | Macbook Pro M1 | 16GB | OS X 14.4 | Metric Halo 2882 3d 
http://www.sounddesign-pro.com
Reply
#4
(07-08-2022, 05:03 PM)Stephan S Wrote: That explains why only 15% DSP usage (which equals 150 % on ten virtual cores) already create crackles.
I would see that as 1.5% spread across 10 cores.... also try at 1024 I have had better success than running higher or lower for mixing.

(07-08-2022, 05:03 PM)Stephan S Wrote: However, converting the mono tracks on my project did the job.

I often find people who work in Broadcast make the mistake of exporting everything as stereo when it is really only a mono track.
They seem to think a stereo pair can be made louder than a mono signal panned equal parts L+R.
Often worth importing all as mono and using phase cancel on one side to determine exactly what is dual mono vs real stereo.
I then delete the mono tracks that are duplicates.
Macmini 8,1 | OS X 13.6.3 | 3 GHz i5 32G | Scarlett 18i20 | Mixbus 10 | PT_2024.3.1 .....  Macmini 9,1 | OS X 14.4.1 | M1 2020 | Mixbus 10 | Resolve 18.6.5
Reply
#5
I use Flux Stereo Tool Goniometer and then just delete one of the split files in the interchange folder before reimporting. Need to educate clients- or do extra charges.
Wasn't Mixbus supporting interleave files in the past?
Mac Pro 5,1 | 6x 3,4Ghz | 48GB | OS X 10.14  | Macbook Pro M1 | 16GB | OS X 14.4 | Metric Halo 2882 3d 
http://www.sounddesign-pro.com
Reply
#6
Mixbus32c has supported interleaved files as long as Ive used it….so back to v4. It imports. It exports. It displays as stereo.

Given that analog desks have no stereo channels, and honestly when they do, they're control ganged mono….what youre experiencing isnt related to “not supporting stereo files”….its that mixbus’s entire ethos is that all channel DSP is on, running and ready. You thunk some other daw uses less than 2x the DSP because it uses 1/100th the CPU for a given channel by default. It still uses 2x for two channels of DSP. Other than the tiny potential savings of one file system call, theres no performance advantage in using interleaved files under the hood.

I will solve your issue, though. In the import dialog, choose “one track per channel”. It would take you 1min to solo and flip polarity on one of each pair and delete the ones that cancel. Otherwise, with a stereo guitar, it imports as two channels, you flip polarity on one, solo both….if they cancel, its mono and you remove one of the two (the one you inverted the polarity to save clicks). Trust me, you will never win trying to educate clients. You will frustrate them and potentially get overed mono summed tracks. Kind of a lose/lose.
Win10pro(2004) : i7 8700/RX570 8gb/16gb/970evo : RME PCIe Multiface : Mixbus 32c 4.3 & 7.2
Other DAWs: Logic 10.4 (MacBook) Cubase 10.5 (PC)
Music: https://jamielang.bandcamp.com
Reply
#7
I found the solution to my performance problems, this is important for all Metric Halo 3d users:

https://forum.harrisonconsoles.com/threa...l#pid60513
Mac Pro 5,1 | 6x 3,4Ghz | 48GB | OS X 10.14  | Macbook Pro M1 | 16GB | OS X 14.4 | Metric Halo 2882 3d 
http://www.sounddesign-pro.com
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)