He also was running some screen capture and said he did not disable Google Drive. In my experience, Playback Only often is worse than Playback & Live (in spite of the theory).
He has the process buffer on Medium instead of Large and Playback Only. He also doesn't understand how Logic works given his explanation about the 1024 buffer. Furthermore, he had the threads in Logic on Automatic which do not use the 2 efficiency cores. He is instantiating every articulation in SCS (not sure if this included the performance legatos which are CPU hogs) as well as some of the BWW articulations. This is a lot more complicated than I assumed. But still impressed just how well my intel MBP holds its own.Ĭlick to expand.So I finally suffered through the video. The last 21 track were Areia, which has K6's reverb on by default (and is also more taxing than K5's reverb). I'm guessing it's a Live issue, but this is something I've seen reported many times so it's worth investigating how the buffer affects active instrument tracks.īTW, as for my final results in the current version of the test project I topped out at 241 tracks. (' The state of some Kontakt instances cannot be recalled correctly').įinally in the Live 11 beta most users find they actually get more CPU headroom with a lower buffer. Just keep in mind that my sample locations aren't the same as yours so you'll probably get stuck with Kontakt spitting out the new missing samples dialog. I have 3 of the Albions, I could easily fill up 250 tracks across all libraries. If not I think the best apples to apples comparison would be to use the same libraries. Or, knock on wood, Rosetta was doing some of its 1st-run translation behind the scenes? (Still not completely sure how that actually works).Ĭlick to expand.I don't have SF Symphonic Orchestra but I do have Spitfire Studio (not Pro) let me know if that works. Knock on wood this is what's actually going on and their playback numbers should actually be higher. Looking at the guy in this video's CPU meter when they show it, the meter seems to be shy a few cores. EDIT: I believe leaving Logic's multithreading set to Automatic is the problem. Hoping that someone can duplicate this guys test with an m1 Pro/Max and see if this really is the cost of Rosetta, or if what people have said is true about manually enabling all cores in Logic (not leaving it set to 'Automatic'). A few have stated that leaving it set to automatic excludes the efficiency cores. As I mentioned in my previous thread I've seen people say Logic will choke if they don't manually switch Logic's multithreading to use all 10 cores instead of the default automatic. I really hope this guy's results are some kind of user error. Not to mention that the memory is shared with the GPU. Macos is already super efficient at compressing data in use, even on intel. I've personally been skeptical about assumptions that 64 GB on M1 = 128 GB on intel, my test seems to confirm that. I know BWW uses more RAM but I landed at Logic showing 46-ish GB of memory in use with all 220 tracks playing back. Memory pressure is way down at the bottom in light green, no swap. Interestingly my memory usage isn't any worse.
guessing I should be able to land near 240 tracks.
I still have a bit of headroom as well, while everything is running between 90 & 95%, there isn't a single core spiking so I should be able to squeeze even more tracks out. I can also leave it looped with no dropouts. Not a single hiccup with all 220 tracks playing back. (He was only able to get around 200-205 IIRC). I'm able to playback all 220 tracks on my i9 MBP using either settings. I had to substitute Ark 1 for Berlin Winds, but I think it's safe to say that the CPU draw should be the same because they're both housed in Capsule, and he's using discrete articulations instead of multis. 256 buffer, medium buffer size, 'playback and live tracks' selected) I even used the same little melody, and copied their settings, (and tried mine as well. So I duplicated this guys test using my 8-core i9 MBP.
0 kommentar(er)
