Cubase 12 Upgrade ASIO-Guard Issue

Thank you WEM and LarsErik for your detailed replies. It’s great to get this issue documented.

Although we’re all approaching it from different angles, we all share the same end result with overloads/dropouts/peaks. This is leading me to believe the root of the problem lies within C12 and not how we’re using it or in the way we setup our projects. I’m also currently out of ideas for further troubleshooting. The best next step is to try and report this problem directly to Steinberg and get their feedback. I’ll look into how to do that, In the meantime looks like I’ll need to stay on C11 for awhile longer.

Does anyone know how to add a Tag to this thread as an issue(bug) with C12? In hopes it will get the attention of Steinberg. Not sure how to do this since we’ve already started the thread, and I mistakenly did not include any Tags when I created it.

Any help will be greatly appreciated!!

I too can confirm spikes, peaks and lags in C12. I am struggling running even a small project and get lots of peaks which I don’t have in C11.

Also C12 seems to crash almost every time I quit it.

I have intel i5 10600K with 32gb ram (4x8gb)

C11 runs very smooth and has fast response. C12 feels laggy, and also seem to have some dropouts.

I hope they look into it.

I noticed something interesting. If I have record enabled on one of my VST-synths (VPS Avenger in this case) the peak jumps like crazy. Here I have two pictures, I took this picture while playback. In other I have just record enabled on VPS while playing back, and on other I have no record enabled and the difference in peak is huge while playing back.


4 Likes

Im seeing a similar issue. Ive been dealing with this issue for years, but had worked around it in c10, c11.

I saw mention of improvements in ASIO performance in the marketing for C12 and was hopeful this could finally offer me something, but alas… Ive again had to disable ASIO guard to stop pops/clicks, which was enabled by default in c12. CPU usage on average much higher than C11. I look forward to updates from steinberg on this.

3 Likes

The quoted box below shows an old post I made on this forum sometime ago in response to glitching in C11Pro:

Glitching seemed random in C11Pro, except it was definitely time-use related. Glitches are absent for 1st minute or so of play, then become more frequent as play rolls on.

Complete cure achieved by TURNING OFF HYPERTHREADING in BIOS setup (Lenovo Laptop Core i7). Thanks to all who suggested doing this on here, and on other DAW forums.

It now all makes sense to me. Hyperthreading allows background processes to hijack and use the spare capacity of a core that APPEARS to be idle. This is why glitching is often absent at the beginning of a session, then gets progressively worse as background process covertly slip into the core(s) you are using for your DAW. Turning off background processes in WIN 10 doesn’t work on its own , but with hyperthreading turned off too, the problem is solved.

Since turning OFF hyperthreading, I’m now running Latency down to 4 ms for recording with no glitches whatoever, whereas, before, 40 ms wouldn’t be enough to stop glitches, and the huge latency was unusable for recording anyhow.

So I experimented with hyperthreading /multicore states within my CBPro12 update this morning:
Running Groove Agent 5 on a drum loop (Beat Agent) and playing Halyon SE3 on a synth lead (Hammer something):
Experiment 1.

  1. ASIO selected = Yamaha’s ASIO THR11
  2. Latency settings: INput = 8.9 ms; Output = 31.5 ms
  3. Hyperthreading switched OFF in BIOS = 2 cores available.
  4. GA5 multicore = OFF
  5. Halyon se3 multicore = OFF
  6. RESULT = marked glitching with just the drum loop running - playing Halyon made it worse.

Experiment 2:

  1. ASIO selected = Yamaha’s ASIO THR11
  2. Latency settings: INput = 8.9 ms; Output = 31.5 ms
  3. Hyperthreading switched OFF in BIOS = 2 cores available.
  4. GA5 multicore = 2 CORES
  5. Halyon se3 multicore = OFF
  6. RESULT = NO glitching with just the drum loop running - playing Halyon was glitch-free

I then rebooted and switched hyperthreading = ON in BIOS (4 cores now available) and repeated Experiment 1 and 2 .

RESULT: Both GA 5 and Halyon were rendered UNPLAYABLE. It did not matter which core state is selected in either GA 5 or Halyon; i.e. OFF, 1, 2, ,3 or 4 cores selected.

I then repeated both experiments with ASIO4ALL selected (512 buffers = Input 13 ms; Output 20 ms) and it was aceptable with hyperthreading OFF but unplayable with hyperthreading ON.

Conclusion: Steinberg has an issue with hyperthreading /core handling that existed in CBPro11 and is even worse in CBPro12.
Groove Agent 5 works best with 2 cores selected with hyperthreading in bios = OFF.
And it remains playable/glitch-free if Halyon cores are on (2 cores) or OFF, when playing along with a GA5 drum loop on 2 cores.
And definitely, for my ThinkPad i7 at least, hyperthreading should be switched off when using either CB11, or CB12.

Sorry for long post!! :grinning:

7 Likes

I have Hyperthreading deactivated in BIOS (as well as c-states, and turbo mode) - Computer setting
is done for background services. I can confirm that hyperthreading is an issue for cubase.

As I have an i7 8700K I have 6 cores available (with hyperthreading it would be 12).

I yesterday played with the HALion setting for multicore. Tried to set it to 2/4/5/Off with no effect. I did not try it with Groove Agent 5.

br
Werner

1 Like

I have resigned myself to working in C11 until they get this CPU load issue sorted in 12. My levels in C11 are even and do not jump or randomly spike. C12 has tons of spikes, audible glitches and feels generally unstable under load.

So …yeah…count me as another person having CPU / ASIO issues with C12.

PC / 10700k
32gb RAM
2TB system drive m2 SSD
2TB project drive m2 SSD
Focusrite Rednet PCIe (Dante) as ASIO device (able to run 32 sample buffer in C11 all day long 3.2ms RTL)
Windows 10
Built-in graphics
I did no configuration changes when going from C11 to C12, and the system is specifically used for audio.

1 Like

Oh yes! the dreaded Hyperthreading + ASIO rabbit hole. I too was encouraged around the marketing regarding improved ASIO performance. And this may prove to be the case for some, but I remember seeing this topic of ASIO/Hyperthreading discussed hundreds of time over the years without much transparency or proven documentation. Seems like before, we’re left to figure out what works best for our particular use and system setup. Maybe this is the right/best way to move forward given all options. At times it feels like one step forward, two steps back.

Thank you bloodline1 for sharing your experiment and for documenting your experience. This is very helpful to us users!

In the meantime, I’ll continue to experiment with C12, but until we get some official comment on this issue, I’ll be using C11 for my real work.

Let’s all keep the discussion going!

1 Like

Same here. I thought it was the project. Rendered a lot of drums. But even removing plug-ins makes no real difference. I’m not running any VSTi. I’m using UAD for the heavy lifting. This project ran fine in C11. I tried boosting the ASIO Guard to high. No real difference. AMD Ryzen 9 5900X 12-Core Processor 3.70 GHz; 32 Gb RAM. Swap to C11 and I can run all my heavy duty plugins on the Mix bus. In C12, they cause the whole project to distort.

Can confirm random spikes and disk cache overloading… and a very high usage peak when certain vst instruments are record enabled. This doesn’t happen in C11.

It hasn’t made my system unusable… can still run C12Pro with largeish projects but it does cause random spikes and high cpu usage. Changing ASIO Guard to “High” hasn’t helped.

Specs -

Win 10
Intel i5 10600k 4.10 Ghz
MSI MPG Z490 Gaming Edge Wifi Motherboard
16 GB RAM

1 Like

OK. Another update and some potentially good news…?

I’ve been fiddling about with settings and at present I’m not getting any CPU spikes! However, the odd thing is that I’ve tried these same settings before and I had problems previously. Thinking back on it, the most notable thing I did was … Wait for it…

…I completely shut down the computer and restarted it a couple of times.

I rarely turn off my machine, so maybe this helped? I honestly can’t think of anything other than that that is different from before.

So to further test, I left C12 running over night with a simple composition looping. The piece is about 2 minutes long and uses six instances of Kontakt with some very basic automation of tracks and plugins. When I checked this morning, the performance meters had not registered a spike of any sort during the night.

Then I thought perhaps something like BackBlaze or TimeMachine had been running in the background earlier and that had caused the random spikes? So I ran playback while both of those processes ran. No spikes.

I then ran playback with all six Kontakt tracks record enabled. The meters changed somewhat to reflect this (especially realtime and peak), but no wild swings, no spikes.

FWIW, my current settings are as follows:

  • ASIO DRIVER is using an Aggregate Device
  • 32 bit float
  • MULTIPROCESSING is ACTIVE
  • ASIO Guard is NORMAL
  • AUDIO PRIORITY is BOOSTED
  • DISK PRELOAD is 2 Seconds
  • RECORD LATENCY ADJUST is CHECKED
  • RECORD SHIFT is 0 SAMPLES
  • VOLUME AUTOMATION PRECISION is 8 Samples
  • BUFFER is 512

I am on an iMac Pro 10 core, 64gb Ram, Big Sur 11.6.4 CUBASE PRO 12.0.0 Build 205

I’m going to keep my fingers crossed moving forward.

Cheers all.

Thank you pkmusic and beanstalk40 for sharing your experiences! What is interesting is you mention setting ASIO-Guard to High did not have any benefit for you. On the initial testing on my system setting AG to High did make a difference. Maybe I’ll try running some different tests with various heavier types of workload. I’ll report back on this.

gesslr - I quess the old anecdotal help-desk advice to restart your device has some merit after all! I am curious to see if the problems return for you, please keep us updated. Full disclosure, after reading your post, I did try this myself. Unfortunately I wasn’t as lucky as you!

1 Like

As an experiment I tried playing with one of the demo projects that come with C12.
“Cubase 12 - Pop Demo By Austin Hull.cpr_MIXED”.

For me, the result is the same. When selecting one on the instrument tracks, then selecting a folder track, I get a dropout/click when ASIO Guard is set to Low of Normal. I don’t hear the dropouts with AG set to High (maybe it’s still there, but very faint and undetectable).

I would be very interested if some of the other users here could try this same test using the demo project. You can find it in your “Steinberg Download Assistant.” If you select “Cubase Pro 12” it’s at the bottom of the list.

I think having a standardized test subject like the demo project could be a great help for us all to compare our experiences using the same project material. Since the Demo projects where produced and provided by Steinberg, this seems like a good place to start.

Curious to know your thoughts on this …

1 Like

Shouldn’t the overload indicator by the audio performance meter come on whenever there is an audio dropout? Did it come on?

No, I get no indications of overloads in the performance meter. Watching the meters as I switch tracks I only see the adjustment from the tracks being armed and unarmed, but no red errors. Only an audible dropout/pop.

Sounds like that could be a bug in itself.

Yes, I agree. If proven true, let’s see what happens next.

I have been reading this thread with interest. I updated to Cubase 12 last Thursday, the whole process went fine, no problems. Thursday and Friday I played through 8 projects (songs) I am working on that were created with Cubase 11, they all played fine and I noticed that the performance meters seemed lower than in 11 and not one of the songs peaked the performance meters.

Saturday I recorded some new vocals for one song, everything went fine no issues. Sunday & Monday I worked on the same project editing the vocals, all good. Last thing Monday I played through the project a few times, all good, not once had the performance meters peaked, no audio drop outs. So the project is saved and the computer switched off.

Tuesday the computer is on all day (as it is most days, switch on in the morning, off last thing at night) early evening I open Cubase for the first time since the previous evening, I load the project I worked on the previous day, I play the project and the ASIO meter peaks and overloads, the audio drops out after about 5 seconds and continues to do it on and off through the track.

The project is exactly as it was the previous evening when it was played through numerous times with no issues. Nothing new has been done to the project. Nothing new has been installed on the system.

I try numerous things, changing the buffer size, ASIO guard but the project continues to peak the performance meter every time. I return the settings to the normal settings I have been working with.

I close the project and Cubase and go off to eat!

I come back about an hour later, I open Cubase I load the project and it plays perfectly no issues! I play it three or four more times and it plays perfectly, no performance meter peaks.

Today it has played perfectly with no issues.

I am at a loss to understand why for about an hour it refused to play without the performance meter peaking, then after closing the project and Cubase, waiting an hour and loading it again it then plays perfectly.

Cubase 12.0.0 macOS 12.2.1 Mac mini i7 3.2 16GB ram.

4 Likes

Hello SRN2016 Thanks for joining the thread! We have all been sharing our experiences with C12, and the random nature of the issue you describe can be the most frustrating and difficult to find the root cause (been there!).

First off, my thoughts for troubleshooting would be to see if any background processes where happening when you had the problem (like software updates, virus protection, etc). Then maybe check the integrity of all your system hardware, maybe something failing? I’m only guessing at this, because this type of random behavior is very difficult to pinpoint.

Also, at this point I would not rule out the possibility that there are issues with the software itself. Many here are reporting usual behavior and performance issues with the new C12. Please report back if you make any new discoveries regarding the problem.