10-19-2023, 03:08 AM (This post was last modified: 10-19-2023, 03:26 AM by johne53.)
(10-18-2023, 05:26 PM)sunrat Wrote: Windows needs a lot of tuning to be optimal for audio, although a few tweaks are more essential and effective than others.
There's a few other important differences too... for example, Intel and AMD use different timing strategies (Intel shares the same timer among all cores whereas AMD gives each core its own timer). But the AMD timers aren't synced so this can cause havoc for an OS like Windows which is constantly switching threads between the different cores. Another important difference is that Windows is a message driven OS (you might've heard of Windows messaging?) whereas Linux and MacOS both use a technique called signalling (signalling gives more consistent time calculations).
Mixbus would need to use different strategies to make allowances for the different processor types and for messaging. In fact, MB does allow for the different processor types but not for the differences between messaging and signalling (there were advantages to this in the early days...) IIRC though, Jack made allowances in both cases. It might not be feasible to switch to Jack but if it's possible, you should see more consistent DSP readings - tho' that never worked for Tassy for some reason!!
Knowledge is knowing a tomato is a fruit... Wisdom is knowing you don't put tomatoes in a fruit salad !!
10-19-2023, 03:32 AM (This post was last modified: 10-19-2023, 03:36 AM by Tassy.)
From MB version 9.1 Jack works now in my windows and can save around 15% DSP or more.
And from 9.1 it is all the same whether Jack is chosen by the opening audio window or from the openned session options.
Win7/64, Mixbus32C, Mixbus2.5 the Queen UR22, Dynaudio BM5A MKII, Pc all SSD,
Thanks for all the advice, everyone. Tomorrow is my "day" for sitting in the studio and applying all these tweaks, reconnecting keyboards, scanning VSTs ... you know the score. I'll post here if I have any impact on the DSP.
(10-17-2023, 11:14 PM)TheBassBagwhan Wrote: Hi everyone, I've rebuilt my PC with a fresh Win 10 Install, an i5 10-core CPU and 32Gb Ram and new mobo ... and figured it was worth investing in 32c 9.2. I've used Version 7 in the past, so I'm not a complete newbie.
With just one track created in a new MB session, no recordings, and absolutely no plugins whatsoever, I'm seeing 35-40% DSP use. CPU according to Task Manager is about 2%
Perhaps rather than address this particular problem - unless that's worthwhile - can anyone point me to good threads or MB resources in regards to making sure I've got basic settings and setups correct? Meaning, I should try and work my way through these things myself before starting to cry for help.
I'm using a Presonus 1810C interface and it's definitely selected as the audio device, 48 Khz rate and 256 buffers for a start.
Likely, I have a lot of basic Windows tweaking to do first, but it did seem an odd issue right off the bat. There are a lot of background processes happening that I need to cull, but that CPU use suggests that's not a problem.
Thanks for any help.
I guess you're talking about a 12th/13th Generation Intel, with P and E cores? Then this behaviour might just be the way it is. I run a I7 12700 and compared to my two older computers (I5 3rd Gen and I7 4th Gen) the idle DSP shows much higher (it's actually lowest on the old I5 2core/4thread CPU).
I guess it's an efficiency thing, of your 10 cores probably 8 are parked on idle, until there is a real load.
On my 12th Gen the idle DSP is as mentioned fairly high (in the high 20ies with an empty session), but it won't go up as fast and you have way more headroom CPU wise.
I also run process lasso for my own peace of mind to avoid core parking, efficiency is for me not an issue.
(10-18-2023, 08:32 AM)Tassy Wrote: The guy has Windows not Linux, so what dsp is on yours is just a good fact that has nothing to do with his question..
Of course it's related. We are working on the same hardware and sound drivers are usually programmed very close to the hardware, so there is not such a big difference. It might be mor a configuration issue or other applications which interfere here.
I will try to load the same project on the same Laptop when I boot Windows, Could be intresting...
I have done the test and was a littlebit irritated because windows was using more than three times DSP power (45%). However, after doing the performace optimizations mentioned in this thread the DSP power went down to 15%. And of course the Linux system is optimized as well.
UR22, Dynaudio BM5A MKII, Pc all SSD,
