ASIO meter behaviour.

Hi,

Here are a couple of pics of the weird behavior of Cubase ASIO/Real-Time meters. I see wildly different readings depending on which track/instrument I open In the Key Editor.

If I enable Steinberg Power Scheme both meters will hit the red, also, the same happens if I disable ASIO Guard.
Surely that should not happen, I mean, the project should run fine without ASIO Guard enabled. After all the project is certainly not heavy, just a few VSTIs and VST plugins.


Can you please list what plugins you have loaded and the number of instances of each? Are any of those plugins 32 bit?

If you start a new empty project and load a single instance of Halion Sonic SE, then open the Key Editor, do you see a change in the metering? In Devices, Device Setup, VST Audio System, what is your Input Latency?

Hi, Chris,

This is the project I referred to in my original post.

3 x Acustic-Audio Nebula (1 x reverb 2 x EQ)

16 x Slate Virtual Mix Rack (Virtual Consoles)

1 x Acon Digital Multiply (CPU hog!)

2 x DDMF LP10

1 x UAD Fairchild

2 x Cubase EnvelopeShaper

1 x DDMF 6144

1 x UAD Roland CE-1

1 x UAD Pultec EQ

1 x UAD SSL Buss Compressor

1 x Waves L3-LL Ultra Stereo

16 x Slate Virtual Tape Machines

2 x EZDrummer

1 x Omnisphere

1 x Trilian (two channels)

1 x Polysix

1 x Wavestation

2 x Elysia Empressor


None of the plugs are 32bit.

Input Latency 12.630

ASIO Guard Level: normal

Steinberg Audio Power Scheme not activated.


New project with Halion Sonic SE, same ^ settings, ‘real time peak’ meter shows some activity when Key Editor is open. Doesn’t show any activity when Project page is open.

No activity in ‘average load’ meter in either page.


Mauri.

Can you please post a screen shot of the project view and Key Editor view of the VST Performance meter in the HSSE project (the complete screen please)? If you raise the buffer size on your audio interface, does the VST Performance change?

Are you running Hyper-threading on the system? Are you running any anti-virus?

Hi, Chris,

Here are two pics of the Key Editor and Project page with buffer and ASIO Guard settings.


Two more with different buffer settings…


Two more with different buffer, ASIO Guard set on Low as with 1024 buffer it can only be set to Low.

Not a lot difference in this test project and with these settings. The meter behavior is a lot more pronounced in a heavier project.

Hyperthreading enabled but after testing with it disabled no real difference. No antivirus running.


Unfortunately, I’ve had some weird asio meter behavior as well. It seems that the red light indicating peak is ALWAYS on… the meter will stay down low, and then I look and the light is on again. I hope they can fix the issue. I have tried with and without ASIO guard… with and without hyperthreading.

I can say that with hyperthreading, my computer runs much better - everything is snappier - so I would greatly prefer to not have to disable it. Also, because of the way I work, I don’t mind longer latencies. Even with longer latencies though, I still have ASIO meter weirdness.

As a general recommendation, we suggest having Hyperthreading off.

I noticed from the screen shots that the MIDI track is not in Record Enable. If you put the MIDI track in Record Enable, do you see the same rise in the VST Performance meter? Generally, I would expect this as the Virtual Instrument is now excepting incoming MIDI and trying to be ready to create the sound accordingly.

In the Key Editor, this occurs automatically because you have the “Acoustic Feedback” option engaged. If you turn off “Acoustic Feedback” (speaker icon in the top left corner) in the Key Editor, does the metering stay higher than when the Key Editor is off (and Record Enable is off)?

I thought the Steinberg recommendation to disable Hyperthreading was an outdated recommendation? Now we have to ask again…

“To Hyperthread or not Hyperthread? That is the question!”

Here’s the official statement from Cubase 7 on Hyper-threading and ASIO Guard: https://www.steinberg.net/nc/en/support/knowledgebase_new/show_details/kb_show/hyper-threading-and-asio-guard.html

Depending on the project setup, with ASIO guard on, hopefully the audio dropouts from Hyper-threading would be minimized, but as a general recommendation, I would still suggest that having Hyper-threading off is the better way to operate. Again, it depends on the setup. If you were only using audio plugins, it’s likely that ASIO Guard and Hyper-threading could be a good selection. But if you are using multi-timbral VST instruments with Disk Streaming, ASIO Guard cannot help with that, so a user would be better to have Hyper-threading off. I hope this information helps to clarify.

FWIW: I think Mauri recently changed his hardware to the latest and newest hardware available, and what he’s stating is that it shows spikes galore in cubase even when it shouldn’t do this. He’s doing the same thing as before, but with the newer system the performance went downwards in stead of the expected upwards.
I’ve seen a lot of spiking issues in the recent past (mostly not on my system) but i do see a consistency in it, that it looks like very determinated by driver issues. Very brand related, and the way you use it.
Recently i also had to change over, and i’m really lucky this time. Absolutly rock solid for the moment and no spikes at all, even with wifi enabled. And it isn’t the latest/badest system on the planet, but still powerfull.
So for me as a conclusion is that performance for cubase is very system/driver-dependend even when you buy the most recent machinery. It would make sence to pin down the bad drivers. (but it could be that is also a variable matter over time)

again, FWIW, thanks for reading.

kind regards,R
R.

The performance on my new machine is really now a lot better overall than my previous Q6600/ASUS P5E/8GB RAM but the ASIO meter readings don’t seem to be as steady.
I disabled Hyper-Threading and OCd to 4.2GHz which has calmed things down quite nicely. I’m pretty happy now.

There is still an annoying issue with slow loading times of certain items in the Mix Console. In one project EazyDrummer Overheads can take up to 20-30secs, when loading another project, Trilian channel B takes a long time etc.
Also, loading a project with rather large (say, three instances of AITB Mammoth Nebula EQ and reverb libraries) can take up to something like 30secs! That was not the case with my old machine.

On a slightly different tack, I have five SSDs in my dual boot system, two Samsung EVOs and three Crucial M4s which means not all the drives operate at the same speed. It made me wonder how that might affect things, would it affect how Cubase performs? And, is it always a good idea to disable Indexing on all the drives, as I’ve done?

Thanks to everyone for chiming in, especially Chris Dobbs, pretty cool to have ‘official’ Steinberg involvement here!

Mauri.

On the 5 SSD drives, are they all connected by SATA? Are you grouping any of them into a RAID or partitioning them?

Glad to be of help!

Hi,Chris,

No RAID, but during the installation of my two Windows OSs this is what I have, see attachment. Those partitions were created during the installations, I didn’t see any options, it was automatic.

Disk 0 (Cubase) is a Samsung EVO as is Disk 2 (Samples 1)

Disk 1 (Internet) is Crucial m4 as are Disk 3 (Audio) and Disks 4 (Samples 2)

Somehow it all seems somewhat messy…

Mauri.

What 2 Windows OS’s did you install?

If you install Cubase onto the C Drive and load a new empty project on the C Drive, do you have the same issue with audio playback there?

My two OSs, W7 Home Premium x86 Internet, and W7 x64 Cubase are on separate SSDs. When I load Cubase OS, it shows the drive letter as C:\ there, when I load Internet OS that shows as C:\ there. So, Cubase is actually on C Drive.

I’ll do an edit a bit later and post a screenshot of Win Management Console - Disk Management as seen from the Cubase OS.

Cubase performance is now quite good but some Mix Console channel loading times are very slow, and the items that are slow, seem kind of random from project to project, I can’t find any commonality (except for Nebula libraries) as I use pretty much the same plugin set in all my projects, but something odd is going on.

I forgot to mention this before, a while back I partitioned the Internet Drive and installed W10 Preview version to learn a bit about that. I think the ‘System Reserved’ (D:) partition was created then. When I uninstalled W10, (D:) became redundant, I took a semi educated guess, and deleted that and merged the space with 'Samples 2 (E:) without any funny side effects, or advantage. Apparently it is not wise to delete the EFI System Partition, even if you figure out how to do it!

Now I have an opportunity to rebuild the machine using an ASUS X99 Pro board and will have to reinstall the whole setup, so we’ll see how that turns out…

Thanks very much for your time and help :slight_smile:!

In the project that is having the behavior issue, what plugins do you have loaded (please list each with the number of instances loaded in the project)?

Personally, on the new system, I would suggest that you start by only testing 1 version of the operating system (Win 7 64 would be my suggestion).

Hi, Chris,

The issue (mainly unusually long loading times now) exists with all my projects, across the board. The list of plugins in that particular project I posted earlier is a good example of the plugs I generally use.

Maybe I should start a new topic about the slow loading times which is now my main concern?

The meter behavior has somewhat settled down after I OC’d the CPU to 4.2GHz, disabled Hyper-Threading and found the optimal ASIO Guard settings etc. for my system.

Mauri.

Sorry for the duplication on the plugin question. Personally, I would suggest taking the project down to a single 3rd party plugin manufacturer and verifying each one without the others.