What's new

Cubase POLL: Disk Cache Spike with Audio Track

Do you get a disk cache spike when moving the playhead in new project with one audio track

  • Yes

    Votes: 12 92.3%
  • No

    Votes: 1 7.7%

  • Total voters
    13

TomaeusD

In Space
This is an issue I started noticing recently in Cubase 13 that according to the Steinberg forums goes as far back as Cubase 8, or even further to 2013.

Regardless of Cubase version, project location of SSD, or any audio settings changes, if you open a new empty project, add an audio track, and click anywhere in the playhead, the disk cache meter will overload briefly. So I wanted to set up a poll to see how common this occurrence is for Cubase users.

CubaseAudioTrackDiskSpike.gif

2024-04-12 12_16_52-Cubase Pro Project by Tom Dunkin - Untitled1.png
2024-04-12 12_15_35-Cubase Pro Project - Untitled1.png

I will share some of the forum posts I've seen already.


Please vote Yes if you get the spike, or No if you do not. It would also be helpful to post if you do and which version of Cubase you are using, your OS, and if you have noticed it previously, when you first noticed it.

Thank you!
 

Attachments

  • 2024-04-12 12_19_08-Cubase Pro Project by Tom Dunkin - Untitled1.png
    2024-04-12 12_19_08-Cubase Pro Project by Tom Dunkin - Untitled1.png
    8.7 KB · Views: 1
UPDATE: Skip what I wrote earlier, I didn't do the test right, I was wiggly the playhead, not letting go after moving it. I repeated the test on my main machines at home, both did the spike, results below.

==================================
I just reproduced your steps and got zero disk cache activity, the bar didn't even appear. I'll list my specs because I'm not on my home i9-13900 tower with a Babyface Pro at the moment, I'm actually at work on a tower I built to use here about 6 months ago, so I could still dabble with music at times (the corporate Dell thing my company provides was NOT going to cut it, lol):

PC:

  • Intel 10-core i5-12600KF (F means no video support)
  • MSI B760 DDR4 board, w/PCIe 4.0 buss
  • 32GB DDR4 3200 RAM
  • Samsung 980 Pro PCIe 4.0 NVMe C: drive
  • Older Samsung 970 Pro PCIe 3.0 NMVe drive for samples and Cubase projects (no spinning nor regular SSD drives)
  • RTX 2060 GPU
  • Win 11 Enterprise
  • *NOTE: I disabled core parking, all 10 cores/16 threads on this i5-12600KF run at 100%. I do that on my studio i9 tower and my 12th Gen laptop. All three had failed LatencyMon when cores parked. Once disabled, all three machines (including this light-weight office i5 tower I'm on now) passed. Does that affect drive access/spiking? Not sure on that one...
DAW:
  • Cubase 13 Pro, latest version
  • UAD Volt 2 with their ASIO drivers
  • I don't have a lot of 3rd-party plugins on this machine, just a few Plugin-Alliance bx-series FX, and Eventide reverb and for VI's I run SINE with a half dozen or so libraries such as BOI 1&2, MA0, Tallinn and some of the freebies, plus PianoVerse.
Thing is, you're just working with an audio file, not a VI, so all that doesnt really matter. One thing, when I imported the audio file to test, I did check the box to copy it to the project folder, and match the bit depth, which are not checked by default:

1712950907467.png

Hope something in here helps. If you have an NVMe slot on your MB, maybe spring for one for your project drive, and not use slower SATA SSDs. SSDs read in about 350-500mbps, whereas the older 3.0 NVMe's read in about 3,500mbps if you get the Samsung Pro's and update the firmware. If your board supports PCIe 4.0 NVMe slots, the Samsung Pro 980 and 990 drives read at about 7,000mbps. Just FYI.
 
Last edited:
I do get the spikes. Never noticed it before but if I add an audio track even to my template it will spike when jumping the cursor back and forth on the on the top bar of main window.


I'm wondering it has something to do with the ASIO Guard.
 
Hope something in here helps. If you have an NVMe slot on your MB, maybe spring for one for your project drive, and not use slower SATA SSDs. SSDs read in about 350-500mbps, whereas the older 3.0 NVMe's read in about 3,500mbps if you get the Samsung Pro's and update the firmware. If your board supports PCIe 4.0 NVMe slots, the Samsung Pro 980 and 990 drives read at about 7,000mbps. Just FYI.
Thanks for the info! I get the spike on a project running off of my Crucial P5 3.0 NVMe which gets up 3,400MB/s, and same with every other type of SSD I have. Also I have core parking disabled, and same audio import settings, although the spike happens before importing anything at all and it's just an empty audio track.

I do think this is a widespread and longstanding bug within Cubase but we'll see what other findings we get. Your post will be helpful as a sort of scientific control.
 
I do get the spikes. Never noticed it before but if I add an audio track even to my template it will spike when jumping the cursor back and forth on the on the top bar of main window.


I'm wondering it has something to do with the ASIO Guard.
Thanks José - I don't get any changes with ASIO Guard on or off. I hope I didn't ruin your Cubase experience with something you can't unsee haha.
 
Thanks José - I don't get any changes with ASIO Guard on or off. I hope I didn't ruin your Cubase experience with something you can't unsee haha.
Subliminally I've always noticed. But, now I can't unsee that. I think that if it's always been there then it hasn't effected me too much. I'll be okay but that is an odd thing. If I load up my template which already has blank audio tracks in it then I don't get the spike but if I add a new audio track, it won't stop spiking after that.
 
Thanks for the info! I get the spike on a project running off of my Crucial P5 3.0 NVMe which gets up 3,400MB/s, and same with every other type of SSD I have. Also I have core parking disabled, and same audio import settings, although the spike happens before importing anything at all and it's just an empty audio track.

I do think this is a widespread and longstanding bug within Cubase but we'll see what other findings we get. Your post will be helpful as a sort of scientific control.
Oh man, you have every box checked! Rats, I was hoping a few of the things might bring you an answer, lol! I’m gonna run that test again on my laptop and my big tower tonight and see if it happens there, it certainly could.
 
Well OK, different story. I just tried this on my laptop (i7-12650H, 10 core/16 thread, 64GB RAM, Both NVMe's are PCIe 4.0), and I got the spike. Shot a simple video, did it 4 out of 5 times:


View attachment IMG_1630.mp4

Now I'm wondering if the work PC test earlier was bad data on my part... I was wiggling the playhead, not letting go of the mouse. I'm now figuring it's a yes on that.

However, after the laptop test, I did my Main tower, and after letting go of the mouse each time, BOOM, got the spikes. That rig is an i9-13900KF, 24 cores/32 Threads, 128GB DDR5 RAM, all 5 NVMe's are 4.0 980 Pros and I'm running an RTX 3080 12GB GPU.

I'm changing my vote to a YES. This is a Cubase featur... um... BUG.
 
If you select any other track which isn't audio, it should stop it from spiking - in other words, I think it only happens when you have an audio track selected.
 
Is it really a bug?
The asio guard is here to manage disk cache and prevent audio issue during playback of Audio tracks.
When you brutally change the playback position (with your mouse), I think its normal to have a spike, the cache should be clean a refresh with new position data.
The good question is, do you have disk spike during playback? and do you hear issue during playback?
 
Last edited:
Is it really a bug?
The asio guard is here to manage disk cache and prevent audio issue during playback of Audio tracks.
When you brutally change the playback position (with your mouse), I think its normal to have a spike, the cache should be clean a refresh with new position data.
The good question is, do you have disk spike during playback? and do you hear issue during playback?
That's exactly why I decided to start this thread and poll, because there is a lot of misinformation and confusion out there about this. According to Steinberg this is not normal, and people who keep finding this over the years are concerned, bring it to support forums, and are told that they need a brand new SSD or RAM after running through many troubleshooting steps. If this is a common benign spike (and not related to other issues that result in actual audio dropouts), it should be documented and clearly stated as such by Steinberg or otherwise fixed. Ultimately I'm trying to help alleviate stress and save some users money if this is the case.
 
Does it matter? Not to begrudge the op's intention and merit, but if it does interfere with playback, or with recording, then it is a problem, if not only a nuisance. Being a Nuendo user for years, and having the habbit of NOT having performance monitor active (only the tiny one on the status bar), there were times I observed it, other times I didn't, and eventually cared no more. Certainly not a jsutification to shell out for new ram or disks.
All this said, I do agree that, being as benign as I think it is to most users, Steinberg should issue some reassurement statement affirming this benign status, thus not relating it to more severe issues.
 
The Cubase 13 manual currently reads:

1713277702400.png

"The overload indicator to the right of the disk indicator lights up if the hard disk does not supply data fast enough.

If it lights up, use Disable Selected Tracks to reduce the number of tracks playing
back. If this does not help, you need a faster hard disk."
 
Yep, got that issue here. Additionally I always leave older versions of DAWs installed while testing newer ones - long story short, I compared, and Cubase 12 also exhibits the same behaviour.

I also get a pop/click and disk cache spike when enabling / disabling loop playback.
 
I do not get this issue (So far. I'm new to Cubase) but I'm finding the Audio Performance meter in Cubase to be a lot more random than the one in Studio One. The more plugins I use the more it seems to calm the meter down and bring it back to a more sensible level in Cubase. In Studio One the meter has a very small % register doing the same thing from the start.

Is the meter info in Cubase a bit wonky?

Cubase 13 with one instance of Dune 3 idle
cubase.png

Studio One 6.6 with one instance of Dune 3 idle
Studio one.png

Edit: Ah someone just explained the differences to me. with regards to Cubase meter compaired to Studio One "It takes the entire chain of audio computation into account."
 
Last edited:
Is the meter info in Cubase a bit wonky?
That's what I was wondering. Pulled up Performance Monitor (perfmon.exe), added all the Physical Disk Counters to the graph, and watched while performing the test rapidly for about 30 seconds. There was no appreciable disk activity that corresponded with the spike compared to similar repetitive activity like moving tracks up and down or rapidly drawing automation neither of which cause a disk cache spike. What disk activity I did observe was likely the undo cache being written to with respect to the associated actions. Best I can guess is that this is more graphical anomaly than anything.
 
That's what I was wondering. Pulled up Performance Monitor (perfmon.exe), added all the Physical Disk Counters to the graph, and watched while performing the test rapidly for about 30 seconds. There was no appreciable disk activity that corresponded with the spike compared to similar repetitive activity like moving tracks up and down or rapidly drawing automation neither of which cause a disk cache spike. What disk activity I did observe was likely the undo cache being written to with respect to the associated actions. Best I can guess is that this is more graphical anomaly than anything.
Thanks for saying also. I was getting a bit OCD with the meter. It seemed random.
 
Top Bottom