You are not logged in. Please login or register.
RME User Forum → Miscellaneous → Multi DAW-Livehost only 4 possible?
I have a setup with one live-host(daw) per core. When I got my new workstation xeon dual cpu 8 core setup a new problem arose. No matter how I connect the live-hosts to the asio driver, regarding input and outputs, 4 live-hosts is the limit.
What to do, I need 7 or more..?
On further investigation, there seems to be an instability regarding this issue. Sometimes 4 hosts can run, sometimes only 3...
Unclear request. Did that work before, 7 hosts? Then it obviously is not on the RME driver side. Or did you never try more than 3 or 4 before?
Before I never tried more than 4 instances of vsthost and that was stable. I was when I got my new workstation with two cpus, that I noticed this: Going for 5 instances of vsthost, the 5th would simply not start. I searched internet for limitations/correct ways of how to connect everything. I tried connecting the live-hosts with different inputs and the same outputs(Analog 1 and 2) and also with separate inputs and outputs.
Then I tried different live-hosts together... The software I am using is Live Professor 2, vsthost and bidule.
When staring Live Professor 2 first, I notice different results each day. Yesterday, I was able to to start 3 instances of vsthost after Live Professor 2 was started. Today only 2. It seams like if I start Live Professor 2, then I can only start 1 -3 instances of vsthost. It varies.
If I restart, never run Live Professsor 2, only run vsthost, I can have 4 but no more, instances of vsthost. This is reproducible.
It is confusing...
And what RME interface and what driver is this?
UCX - Firewire connected, driver 3.122, hardware revision 0.27.
Have you tried testing using MME or whatever other audio drivers might be available in VSTHost (not RME drivers)? This would rule out the RME ASIO Drivers.
Is it possible the dual-socket SMP architecture is affecting this as well? It's different from a single-socket multi-core SMP...
Thanks for the suggestion, I tried connecting everything via JackD audio, and all vsthost instances started, more than 4. But not click and pop free. So it is not usable. But for debugging it is ok.
I just started 6 instances of VSThost x64 on both a Fireface 800 (FW) and Fireface UFX (USB), running Windows 10 Creators.
Thanks Timur, for testing. Good to see that it is working for you. We have slightly different audio interfaces and I use 32 bit vsthost, 64 bit Win10. Can it be that these differences matter?
I will do some more testing, but my result are conclusive. Just now I tried staring as many vsthost's as possible, no more than 3 was possible. I find it puzzling that the number varies.
Just tested 64bit vsthost, asio. 3 instances ran, the 4th did not...
Tested mme driver instead of asio, 5 instances no problem.
The USB driver is the same, so the interface hardware shouldn't matter too much.
Your image is not to clear, correct me if I am wrong. I will specify more clearly. The starting of the program vsthost is one thing, this I can do as many instances as I like, but I am talking about asio connected running instances that can process. Looking at your image, I see in the bottom information bar of many of your vsthosts, where the DSP meter is the following sign: ?%. This is what happens to me too. When this sign comes up, the program is not processing. There will be no audio. So I see you to have the same problem as me.
There is a limit of how many DAWs - LiveHosts you can connect to the asio driver at the same time. F eks if I start 4 instances of vsthost, that are processing, then start Live Professor 2 it will give an error messages and will not work/process. So to be clear, the program is started but not capable to connect to the asio driver and process audio.
Now we know more thanks for helping Timur :-)
From the readme:
V 1.088 (09/21/2015)
- ASIO: Number of simultaneous hosts in multiclient operation raised from 4 to 8
You should make sure that the number of WDM devices is set to 0. After that you can easily test the max number of ASIO hosts on the existing I/Os of the UCX by using different programs that support ASIO. Including DIGICheck which is also a real ASIO Host.
Today clean start of workstation. Set WDM from 1 to 0 and restart. Only 2 instances able to process. You see number 3 have: ?%, that means not processing. So even if the documentation says going from 4 to 8 connections, it still is not working stable.
https://goo.gl/photos/iFRNg5KvyXk9oDD8A
I now fired up 5 clean vsthosts (64bit) hosts, with the option in the program to to connect to all asio channels at once. No just a select few. And there you go all 5 running... :-)
https://goo.gl/photos/tTfKgaNNtDd8uay48
Starting Live Professor 2, first. Then see how many vsthost instances that can process, 2.
https://goo.gl/photos/evoncqd8GSwFpYw68
If anyone wonders why this matters, why do I, do it like this: I like to use my computer as a vst monster processor. I love generating sound-scapes. I have fund no DAW that can put each track on a separate cpu core and have low latency processing. That is why I use vsthost. Each instance process huge number of vsts and totalmix does the mixing and loopback. I have once instance for my e-drums, one for vsti synth. 1 instance for reverb and echos. 1 instance for final mixing. Etc Etc...
So having invested in a xeon workstation with 2 cpus, 8 cores, I am intending to use all cores. This workstation can be upgraded to a total of 12 cores... My scenario should not be too strange for music composition... or...?
Thanks so fare for all suggestions, I really appreciate it :-)
Is it case closed now or will you guys at RME have a look at it?
First of all these tests don't tell much. I tried to explain that above. To rule out the VST Host is not the culprit you should use DIFFERENT programs all the way. Like one VST Host, one Reaper, one DIGICheck, and so on.
Of course we can recheck the 8 limit setting in the driver code, but right now is summer holiday time, so this will not happen in the next two weeks.
Hi MC, are you getting aggravated by my persistence. That's good :-)
I will do your 8 different daw test a little later. Here I did a 8 instances of vsthost connected to jackd test, all instances started, connected and are processing... I will personally challenge you, that this is a more normal setup than 8 different daws, that's almost kinky :-)
Do have a great summer, thumbs up.
https://goo.gl/photos/rmnCiaGC1CpervTW9
> Hi MC, are you getting aggravated by my persistence. That's good :-)
Sorry, but this makes no sense to me. Better be to the point.
Without clear explanations this will not get resolved. For example why is the input port always greyed out and set to WDM in your screenshots? Can you set it to ASIO too?
When you select asio in vsthost, the input port gets grayed out. You only select asio in the output, and by doing so it disables the the input gui option, not the input channels. You get all the asio inputs and outputs. Just how the program is designed, bad gui if you like. The audio system Jackd on windows uses asio as the connection method inside the daw/live-host. That is why you see ASIO: Jack Router. So this test was to examin if there where any short comings in vsthost regarding connections to ASIO. So I am running everything via the asio to jackd bridge. As fare as vsthost knows, it can only see asio. The conclusion of this test is that vsthost has no limit of how many instances you can connect to asio. If you then see the other tests I did, with direct connection to rme asio driver, you see that, the rme asio driver has problems accepting more than 4 connections and that it is also a little unstable, in that the number of connections varies between 2-4 (this I don't know why).
https://goo.gl/photos/GcWhhcvQH6JrAqkXA
ASIO Channels are available, even if input gets grayed out:
> Hi MC, are you getting aggravated by my persistence. That's good :-)
Sorry, but this makes no sense to me. Better be to the point.
What this can translate to in simpler English, is if MC is getting mildly frustrated by my intensity of topic. So there is no intent on my side for any disrespect. This is the type of humor you use to loosen up a conversion. No need to worry :-)
Thanks for the explanation in post 24, that will help the developers to understand the problem.
I did some analysis and managed to get 8 instances of VSTHost running, or 1 instance of LiveProfessor + 7 instances of VSTHost. From what I can tell the issue seems to be with VSTHost, it doesn't play nice with other software or even itself.
There were no issues getting 8 instances of other software to use the FF USB driver concurrently. But trying to replace one instance of other software with one instance of VSTHost would sometimes fail. If that happened I either had to restart VSTHost or close a second instance of the other programs running.
Here is the catch: VSThost sometimes seems to block two (or maybe even more?) connections to the FF driver. If that happens then you have to find out which of your many VSTHost instances is the offender and close it. One by one you have to find all "offending" VSTHost instance to replace it with "good" ones (that only use 1 connection).
In my short testing of around 1 hour of time I did not see this happen with any of the other programs I ran, so there is a good chance that VSTHost is the one to blame.
At one point I had audio buffers breaking up, which cleared itself by closing ASIO program instances. Since VSTHost was running when this happened this might have been caused by VSTHost as well.
Thanks Timur, valuable info. How can we combine this info with the fact that Jackd vst bridge runs flawless with vsthost as demonstrated above, 8 instances? If vsthost behaves badly with direct rme asio, but not with jack asio-bridge, what conclusions can be made by that? So jackd asio bridge has unlimited connections, so I a few gets blocked, than that does not matter, that is why it apparently seams to work ok...
I will investigate this further.
Here I did test of a small bass to midi host, completely different company/person than vsthost. It uses asio. Fresh start of windows 10. No other daws or jackd running. Starting as many instances as possible. 4 ok, 5th fails with asio driver error.
Look at the gray cpu meter on the working instances. Look at the yellow triangle with the ! sign in the most front window.
Here I did the same test with BlueCat Audio Patchwork, it uses asio. I started 6 instances loaded a patch for processing in each. This program is not so sophisticated in showing errors and cpu usage. So had to use the windows cpu meter to see what was processing. 4 instances of 6 was running properly.
Either these use the same libraries/functions to implement their functionality or the problem only arises when multiple instances of the same program are used. Normal DAWs usually don't allow multiple instances or enforce Rewire.
I will do some tests on that tomorrow.
Curiously I am not able to reproduce this behavior anymore. VSTHost keeps working properly up to 8 instances now. Here is something you can try to maybe fix it for your setup as well:
- Open the first instance of VSTHost.
- Disable menu -> Maschine -> Ausführen.
- Open as many additional instances as you need.
- Enable Maschine -> Ausführen in all of them (which is only possible one after the other).
If this does not work then try downloading Reaper, disable "Check for multiple instances" (General) and start playback in 8 instances. Report back if that works and try starting 8 instances of VSTHost after you did the Reaper test.
I am selling the RME UCX, I can not take this constant tweaking any more. I have used all available time tweaking and not making music. Will go with Linux and the RME hdsp 9632. If this combo does not give low round-trip latency noting will, and in that case I will go mostly analog.
I can not do the reaper test now. Anyhow thanks again Timur :-)
RME User Forum → Miscellaneous → Multi DAW-Livehost only 4 possible?
Powered by PunBB, supported by Informer Technologies, Inc.
