ASIO Guard 2. How does it work? Does it work?

Cubase
1

Howdy,

So I just want to know how ASIO Guard 2 works. I have just come into a few problems with CPU spiking and turned ASIO Guard off and the CPU spiking has completely stopped. This has confused me because why am I going to use it if it’s going to cause any problems.

I was under the impression that having Asio Guard 2 on allows me to play Sample Libraries and record without having noticeable latency with a higher buffer, which is great for having large track counts. Sorry if my understanding is incorrect.

Should I have ASIO Guard 2 turned on? I am having problems with it and I am about to venture into using VEPro 5 so is there anything I should know regarding ASIO Guard 2?

Thanks

Jono

2

Good question my friend… :slight_smile: I await the answers with bated breath, it would be nice if a tecky from Steiny appeared and explained it but that is highly unlikely…
If I turn ASIO guard of I get distortion.
I’m probably wrong but I get the feeling that ASIO guard is there to ward off some other problem that came about while they were adding some bloat…
best, Kevin

3

I have only started getting this problem when I put a new SSD in my machine as my Sample Drive. It’s a Samsung 850 Evo. I have installed the Samsung Magician software and have tried the Rapid Mode and it says it has sped the SSD up massively, however, regardless of whether Rapid Mode is disabled or Enabled, the CPU spiking happens when ASIO Guard 2 is enabled and does not when it’s disabled.

I wonder if the SSD is somehow messing things up (just thinking out loud).

Jono

4

Nah…SSD is a step forward for puters…there’s a lot wrong with CP8 since they released it too early to catch Xmas,
a lot of probs I had disappeared after an update so I’m waiting on the next one…
using Mixbus 3 now with no probs…might end up staying with it…
will for sure if the next update is a paid one…

5

I’ve had problems to. My custom template has problems with Asio guard 2. I’ve did some testing deleting VI. Hal se was causing my asio guard normal setting to constantly peak. So I have to keep my setting asio guard on high for my custom template

6

As I understand correct Asio guard is meant for playback tracks + recording tracks/playing VI at the same time. The tracks which are played back have different (bigger) buffers then the tracks being record (due to low latency monitoring), the purpose would be saving cpu cycles on the playback track as they can have bigger buffers.

I could be wrong but have seen this explanation around somewhere and makes sense to me.

7

Yes, that’s how I understand it.

When you hit record, arm a track, it removes it from asio guard and sets it to your tight latency setting, putting a heavy demand back on the cpu. When you have the track not armed it will be at a much larger buffer.

If we did the maths on paper this should mean. Say you had your latency at 128 no asio guard, you would only be able to have so many tracks and pluggins. With asio guard, all unarmed tracks will be at a higher buffer setting, say 512, even though your set at 128. This would mean you can have the amount of pluggins of 512 when set at 128 (minus the armed track).

Obviously those figures are for reference to explain and are likely not actual. I may also be wrong, completely, lol!

8

An armed track might even set itself tighter than your buffer setting, this I’m not so sure on!

9

Ok thanks. So can anyone tell me why I am getting massive CPU spikes when ASIO Guard 2 is enabled (at any setting) and when it is off, I don’t get any CPU spikes?

Baffled as to why this is happening.

10

Windows OS and CPU power management is the most common cause of this kind of behavior. Are you using the “High Performance” power scheme in the Windows Power Options control panel? If not, make sure it’s selected and then try again. Also make sure you’re not using the “steinberg power scheme” thing in Cubase’s Device Setup panel.

When ASIO Guard was first introduced (Cubase 7.0 I think?) it was really terrible, at least on my system. Significantly made things worse and caused all sorts of CPU spikes. 7.5 greatly improved the situation, and according to some 8.0 improved it noticeably further (although I’m kind of undecided on the 8.0 jump, based on my own experience. I don’t think it’s _worse_than 7.5 though).

11

Yeah, I have it set to High Performance and have everything set to never hibernating/sleeping/turning off.
Hard disk to Never turn off.
Sleep after Never.
USB Settings selective suspend setting - Disabled.
PCI Express - Link State Power Management - OFF
Processor power management - Core Parking min Cores - 100% Minimum processor State 100% Sys cooling Policy - Active. Maximum processor state - 100%

Steinberg’s Power Scheme isn’t turned on.

I actually disliked Cubase 7 immensely. It was unstable. 7.5, well, I was loving it but didn’t really use any Sample Libraries whilst using 7.5 so can’t really recall using ASIO Guard. C Pro 8 is definitely not worse than 7.5. C 7, for me, was a bad experience.

I notice many people on this Steinberg forum have talked about CPU spiking so I wonder what is going on?

12

Good times…

Been waiting for this to be sorted for quite some time.

13

Yeah that’s mad. My problem only happens with ASIO Guard enabled. When it is enabled and I record Arm Instrument Channels, it gets worse. It also gets worse if I lower the Asio Guard settings. So for now, I guess I’m going to have to turn it off. Like you, everything has been cleanly installed. Days of my life have been sucked away trying to figure this stuff out!

I have been meaning to learn how to use VEPro 5 (haven’t had the time yet) as I just bought it in a sale and I wonder whether (on a single machine) this helps with CPU load as the Multi Processing is done in VEPro 5 instead of in Cubase.

I know it must be difficult for a company to release a product to work with every different type of PC component combination but I’m starting to wish Steinberg would release a recommended hardware list so when I build a machine, I could just buy components that work 100%. Perhaps Win 10 will work well with C Pro 8 or maybe I should just go back to Win 7? More wasted days…

14

Jono does it spike on a empty project? With no vst channels

15

Are you guys getting real CPU spikes, as in, when you look at the real cpu usage from the OS system monitor?

Or, is this what you’re seeing in the CuBase VST Performance monitor?

Have you run the disk/bus interrupt latency checks recommended by Steinberg?

Is it ‘sounding bad or dropping out’ or is the VST Performance meter just bothering you?

Unless you’re getting bad audio, drop outs, etc…ignore the VST performance monitor.

If you’re getting clicks, pops, dropouts, etc…enlarge the main ASIO buffer size bit by bit until they go away.

For me…I need larger buffers if I disable ASIO Guard (256k for starters) if I plan to run several VSTi tracks with VST effects applied.

If things sound and work fine…simply ignore that VST Performance Monitor. It’s NOT showing your CPU activity, nor does it necessarily mean anything is wrong or abnormal.

With ASIO Guard enabled, I find I can drop back to much smaller buffers on the audio device settings…128k and smaller, and can run quite a bit of streaming VSTi stuff before I have to enlarge the buffers, or start ‘instant rendering’ tracks into an all audio format.

I’m mostly using a quite old Delta 1010 interface over PCI.

A USB2 Tascam 1200 is working fine as well, with even lower latency, but it also pushes the CPU usage up, as well as the ASIO Guard activity up a bit (nothing major though…everything works just fine).

From what I can tell on my ancient Phenom II system, the VST Performance meter is simply showing buffer activity…armed VSTi tracks for recording or monitoring makes that little line spike and idle. Even on this ancient computer that all the experts say is garbage compared even the lowliest Intel atoms…with some decent sized projects, the CPU isn’t even breaking a sweat. It still works and sounds fine, no problems with 40+ VSTi tracks going…so I ignore it. Seems normal to me that if ASIO Guard is buffering and correcting timing, that the buffers are going to fill up and empty on a regular basis. That’s what ASIO Guard is for…to buffer, recalculate, and tighten up the timing on whatever tracks are ‘armed’ as you play or record into them…with all the stuff that is simply ‘playing back’.

Notice…if you disarm everything (no record or monitor buttons), ASIO guard isn’t being asked to do anything, and the VST Performance monitor doesn’t spike and idle anymore.

For the poster who says there is distortion without ASIO Guard. What size are your buffers? If you turn the guard off, then you’re probably going to need to increase the main ASIO buffer size. Much of this will have to do with your audio interface and how its drivers are implemented.

On this really ancient hardware, only problems I’ve had with CuBase 8 were due to some bad SSD drives that just don’t want to cooperate with streaming data, and they also perform horribly when hosting system pagefiles. Tried the drives on more than one system (both intel and amd, with AHCI and IDE drivers), with more than one DAW, and yep…disk loss issues. Got all system page files, and anything that needs to ‘stream’ off of those particular drives and have had great performance since. I have no clue why the PNY drives hate streaming or hosting memory swap files. They bench mark well, and pass the latency checks…but for some reason the DAWs and streaming VSTi stuff I tried to use them with don’t always get the data they ask for in the right time or position they expect it to come.

Try tapping alt-ctrl-del, then start the task manager. Watch your CPU from there. In my case, the performance monitor in CuBase is NOT showing CPU usage…it’s showing ASIO Guard buffering activity. With ASIO Guard totally disabled, the meter shows a little higher average activity, and the spiking activity goes away.

In either case…with the Guard on or off…I can tweak the buffers and get very solid performance. I choose not to tap F11 and be annoyed by that meter. As long as things sound good enough to get an accurate mix and translation through my monitors, and of course ‘render’ without flaw…even the occasional crackle or pop is no big deal (given the age and limited power of my old computer).

16

Oh my god that video again… We’ve already explained this, i even commented on it… He has 3 tracks armed for record, when he disarms them it goes away. That is exactly what asio guard is! Those 3 tracks awaiting to receive midi data and to play the audio and compute all the information, IN REAL TIME, is too much demand for his PC.

I believe, as mentioned though, some people get wrong meter readings, possibly due to graphics card/drivers so unless your getting drop outs, your not actually maxing out. Apparently Nvidia but that’s not the problems I was getting when testing, I have Nvidia.

17

Notice…if you disarm everything (no record or monitor buttons), ASIO guard isn’t being asked to do anything, and the VST Performance monitor doesn’t spike and idle anymore.

according to Steinberg, it’s the other way round.

Asio Guard is working on the disarmed track. Everything that is disarmed is going to a larger buffer setting than the one you have set. This means with Asio Guard on, one should be able to set their buffer setting to a tighter setting overall (which I can tell you know) resulting in the low latency improvements associating to Asio guard…

Kind of like Asio Guard is doing the opposite to achieve the improvement. Larger buffers for unarmed so we can set a tighter buffer as default. Not a tighter buffer for armed so we set a larger buffer as default.

At least by the description on that link.

Both would achieve the same thing but would be controlled via the settings differently.

Possibly it implemented the second way, like you see it, would be a much better way to implement it. I suppose it’s down to the code.

18

No. No Spiking at all.

19

Yeah I have been getting Audio Drop outs. Asio Guard has been unusable and if there were no Drop outs I would just ignore the VST meter. If you can’t hear something then it isn’t there right? haha!

Anyway, I have had enough and thought “Well my computer can’t get any worse” so I have just upgraded to Win 10 so I’m going to see what that’s like and if it’s any worse than what I was experiencing I shall just install my Clone of Win 8.1 and have another look at this. Or install Win 7 again but I’d rather not go back if I can help it. Thanks for everyone’s help and I shall let you know what C Pro 8 is like with Windows 10 (and yes, I know Steinberg have recommended not using it yet)…

20

Ouch…terribly frustrating I know. So many things can and does go wrong. I’m not in any way trying to deny that there may well be problems with CuBase itself. I just know that QC on hardware is all over the map these days. I’ve had to send back quite a number of motherboards, memory chips, and so forth over the years.

I totally agree with a comment made earlier in the thread that it would be nice to see a listing of recommended hardware by model numbers tested and known to work well together.

It took me months to figure out my more recent CuBase 8 + PNY SSD drives issue. I still got a lot of work done while I was trouble-shooting by forcing myself to ignore the glitches (the pure audio tracks, as well as any internal mix downs and/or instant renders from VSTi were fine). In my case I was kind of lucky that the problem popped up with other DAWs and even some non DAW applications, so I had a very strong indicator to rule out the DAW as the soul problem.

I don’t know if this will be helpful at all (you may well have tried every bit of it dozens of times already) but for what it’s worth:

  1. I’ve found that I should disable any ‘tune up’ software I have installed before working with a DAW or video streaming project. I.E. I have iolo System Mechanic Pro. When it comes to working with my DAW and vsti plugins, it’s smoother/more stable if I disable all the real time memory management, power profile, disk optimization, and cpu tuning features that package offers before using my DAW.

  2. I’ve run into situations before where virus protection engines cause problems. Isolate the DAW machine from the web long enough to try things without your virus protection engine running.

  3. I’ve had fancy keyboard and mice drivers cause problems before (usually the stuff that comes with user mappable gaming style gear, programmable keys, really high resolutions, high polling rates, and robust KRO abilities, etc. Sometimes they do some weird stuff…like sharing and resetting system timers, etc.).

  4. Power profiles that allow any part of the system to sleep, suspend, or throttle back have caused issues for me before.

  5. I’ve had cases where moving audio interfaces, drives, and cards to different ports/connectors, or even swapping out cables fixed issues. Moving some stuff around calls for reseating the connections, and can force the system to reinitialize system drivers and reevaluate how it assigns interrupts.

  6. Almost every system I’ve ever owned gets profiles built for CPU overclocking, undervolting where possible, and gunning for the best bus and memory speeds/timings I can muster. I don’t always get it right (nor are the default bios settings always stable). As crazy as it sounds…I’ve resolved quite a lot of issues in the past by simply writing down my BIOS settings, reseating the memory sticks in different slots, clearing the cmos, and starting fresh with default settings and slowly re-optimizing one bit at a time, along with all the corresponding memory and burn in testing that goes with it.

  7. I’ve had cases where everything seemed great on the hardware end, only to discover that the power supply was not within tolerance and keeping a steady voltage under load.

  8. I’ve had problems before trying to aggregate multiple audio devices with things like ASIO4ALL. I eventually got it working well, but it did take quite a bit of trial and error to get it stable enough to use.

  9. I’ve owned some hardware, even pro line, way over-specced bits that just doesn’t cut it for some reason or another (I.E. My PNY SSD scenario). Too many memory sticks have failed over time to count. Have also had more than a fair share of audio cards that simply have something wrong with them. Finding the problem part or driver can drive a person crazy, particularly if he doesn’t have training in diagnostics software, or another systems to mix, match and swap out different components.

  10. If I had a nickle for every audio card or breakout box that I’ve had to replace or recap to get all the pops and clicks out, I could get myself a new DAW machine today! I.E. My Delta 1010 got on in age…and even with no digital signal at all, it was known to send a snap or pop to the amp at times. The worst of it went away with a new shielded cable, and some fresh capacitors sorted out the rest.

If you haven’t already, I’d definitely write up a report of your issue that is as detailed as possible and file a support ticket. Run dxdiag from the start button or a command line and include the log it produces with your report. Zip up and include all the dmp files in:
%USERPROFILE%\My Documents\Steinberg\CrashDumps