Multicore performance (or lack thereof) in Audition
I’ll just post a (slightly modified for context) message that I sent to Jason Levine as an open letter since there is literally no information or outreach about this out there.
TL;DR, I’m very close to switching DAWs (likely Pro Tools) even though I don’t want to. Between this and a lack of a real strip silence feature that’s also usable in the nondestructive environment, and which doesn’t force me to ripple my track and/or export to hundreds of useless markers, it’s becoming less and less viable to continue editing in Audition.
Please forgive me for contacting you directly. I’ve exhausted all other search options to the extent my capabilities allow, including emailing Adobe support to no avail.
I’ve got a 2019 16” MacBook Pro with 8GB 5500M and a 2018 Mac mini connected to a 16GB Vega 64 eGPU. The former is the 8-core i9 with 32GB RAM. The latter is a 6-Core i7 with 32GB RAM.
I’m running simple multitrack podcast recording and editing sessions in the latest version of Audition. The MBP is used primarily for editing, the mini is what we normally use for recording. We’re recording at 48/32 in mono. I typically apply plugins from iZotope and Waves. My effects chain is usually 7-8 items long per track. Usually 2-4 tracks. One track per individual on the podcast. They’re recording at same rates (at worst, 44.1/16, but assume it’s usually 48/32).
After doing my initial edit I usually do a mixdown to a single WAV at the same rates. This way I can deal with a single track for fine-tuning my pacing, using markers to denote chapters, and mastering to yet another WAV. This is my final, which I later encode (including artwork, links, chapters) to MP3 (160Kbps, mono) in a phenomenal little app by Marco Arment called Forecast (which uses all 8 or 6 cores just beautifully).
My biggest issue is the first mixdown on either device. I’ve noticed on iStat Menus that on either computer only a single CPU core will fire up. On the 16” MBP, it doesn’t even look at the discrete GPU. It stays on the Intel the whole time (even with energy settings that force the discreet GPU on adapter power). On the Mac mini, the eGPU (which I’ve set in macOS to specifically use with Audition, Premiere, Lightroom, etc.) fires up to full bore, but still doesn’t seem fully leveraged. On the MBP a typical 70-80 minute show with two tracks will take over an hour to mix down. On the mini, even with Audition clearly leveraging the eGPU (to what extent or why, I do not know as I thought a mixdown of audio only would be mostly processor intensive), it still takes about 50-60 minutes. In fact, I suspect the difference is more about slightly better thermal performance than the eGPU.
Why isn’t Audition leveraging more of the processor cores? It seems like Creative Cloud applications for 2020 made fairly good leaps in leveraging discrete GPUs in certain apps, even if non-workstation class, but multicore processors are seriously underused across the entire creative suite. Is it just me? Am I missing a setting? It just seems like this stuff should be so much faster. As I take on more client editing work this becomes a real issue. I can use my Mac mini as a passive little render farm so I can free up my MBP, but it’s still taking that hour.
Do you have any insight on this?
I genuinely appreciate your time and once again apologize for reaching out via this forum.