Piano World Home Page Forums Our Most Popular Forums Digital Pianos - Electronic Pianos - Synths & Keyboards WHY cpu varies from one day to the next?

Hello all I just bought the Embertone Walker 1955 Concert D. I personally love combining Close and Main mike positions for a sound I've always hoped to find. Problem being I'm getting audio glitching especially using sustain pedal. My only way around it is use one mike position or lower voices, sample layers and go to 44.1 instead of 48. I find that some days it will run perfect with 2 mike positions and on 48 but most of the time it doesn't. Anyone know what causes this variation of playback? Thanks MooganDavid

Audio streaming problem.
1: Driver issues, ASIO or ASIO4ALL depending on what interface you're using
2: Hardware issues, not enough RAM. If disk streaming, your HDD's not fast enough, in other words, your hardware has a small bandwidth to stream multiple samples (audio files).

What hardware and software are you using?

Walker 1955 Concert D my computer is I7 16 gb ram and SSD and my controller is yamaha cp33 and I'm using a Babyface Pro interface. Like today it ran perfect the whole time I was playing but yesterday it was glitching like crazy so I had to switch it to 44.1 and only use 1 mike position.

I think David B has luck running some VIs with a DAW, allocating different mics to different channels. Perhaps some DAWs do better allocating CPU cores. I don't know if this will help Walker 1955 which seems to push CPUs hard.

CakeWalk is free and Reaper has a long free trial (plus is cheap).

Stop wineing Moogan! Sorry, bad joke. A nice ultra-sweet Concord grape wine is just the thing for… oh, I don’t know when it would be good!

Two unrelated comments.


First, why are you running 48? If your samples are native to 44.1 (the old school standard for CDs) then playing at 48 only creates more work for your CPU, with no audio improvement. Unless you’re doing video syncing, you might just leave 44.1 as your normal setting.

Second, your computer has other things to do besides music. You can easily eliminate some of this by:

1). rebooting your computer before you play. This turns off any programs that still might be active from previous sessions.
2). Turn off your networking (WiFi, Ethernet). You have all kinds of programs running in the background that might decide at random times to go search for and download updates, etc. They will suck up CPU cycles that you might want to use for music.

I saw no indication that Moogan has been drinking.

Sometimes keeping these piano vst's running properly makes you say "I need a drink" Clean for 15 years can't do that!

You don't need 48kHz samples. With current DACs there is no real benefit. 44.1kHz captures what you need within the Shannon-Nyquist limit.

I think one day you use the sustain pedal less intensively, another day more intensively. At least, such was my understabdung when I tried to tame this unprofessionally written software, before removing it entirely from my SSD.

understanding (sorry for the typo). This piano plays ideally if the sustain pedal is not used. A serious pedal work is impossible however few mics you apply. At least, with my Dell Vostro 15 5000 i5, 16 GB RAM, fast SSD, this was impossible.

More pedal work means that the VST has to play many more samples at the same time. Now, if you use 2 mics at the same time, the amount of samples to load will be 2x. If you use 3 mics, the amount of samples will be 3x and so on... So, more mics + keeping the sustain pedal pressed for more time => too many samples to load at the same time => audio glitches. And if the os is doing something in the background (updates, file indexing, antivirus scan, etc.) this is even worse.

A simple logic for maintaining "greedy" objects, like audio samples, in a resource-limited system, is forgetting the least important objects (such as samples loaded before all others, or samples whose current loudness is neglectable), rather than waiting for audio glitches and forcing the user to look for the culprit.

This logic is successfully followed in more advanced piano plugins, such as Ravenscroft 275.

If I remember well, Embertone Walker 1955 Concert D is based on Kontakt player, which was developed in last century. I suspect that Kontakt does not allow to fully implement the above logic. To some extent, the authors of Pearl Concert Grand (also based on Kontakt) did this for supporting re-pedaling, but their re-pedaling is not as good as in Ravenscroft 275 or Garritan CFX.

You seem to put on the same level re-pedaling which deals with how the library is scripted (the responsability of the library maker) and the ressource handling (the responsibility of the engine maker).

On Kontakt, you can define a voice limit, then avoid drop-outs even if your CPU is limited. But the threshold doesn’t appear to be dynamic (depending of the CPU usage). In practice, I had only troubles with Bechstein Digital Grand and have to put the voices limited at 150 or something like that.

One thing which improves quite a lot the audio glitches is to remove all noises, i.e. pedal noise, hammer noise and such. Those are not needed in general, even undesirable in most cases, and generate extra load to both SSD and CPU. It is not a matter of lowering volume, you need to disable them (which, admitedly, some times is setting volume at -infinitum, as in VSL Synchron Pianos)

Also, minimizing any graphic interface helps to reduce both GPU and CPU loads.