Overloaded CPU

General discussion about Gig Performer
1

In one of my rackspaces I need to use 5 intances of Korg Triton extreme.
For the first time of my (short) use of GP, when playing using this specific rackspace I noticed a few cracks/noises randomly and I didn’t know why. Then, I noticed that the CPU usage was between 48 and 56%, much higher than my other rackspaces (typically below 30%).

So, do the noises occur due to the CPU usage?

Is the kork Triton Extreme vst a CPU “eater”? Any way to solve this?

What is the safest level (maximum CPU percentage of use) I should set to avoid this noises?

Greetings

Salomao

2

What OS (Mac or windows)? What audio interface? What sample rate and buffer size?
What kind of computer and how much RAM does it have?

You can see how much CPU is being used by bypassing four of them and see what a single one uses

3

Mac 10.14 (Mojave)
Mackbook Pro I7
16GB RAM
MOTU M6 audio interface
Sample rate and buffer size are GP default.

Each Triton extreme plugin seems to use about 13% of CPU.

That beeing said, is it normal to have noises if CPU usage goes beyond 50%?

4

To me it does not strike me as “abnormal”. There are some plugs ins (or particular sounds) that can eat up a lot of CPU.

But, there are often adjustments than can be made that do not compromise the sound, especially for live performance.

For example, there may be “quality” settings that affect CPU, which can be adjust. To the extent you might want the highest “quality” setting for recording, you probably do not need that for live performance. The “lowest” quality setting might work fine. You need to evaluate CPU use with and without different quality settings.

A lot of times effects are burning up CPU. Do you need them all? You need to evaluate CPU use with and without different effects.

Number of voices (polyphony). This can effect CPU. Frankly, in a live setting too much polyphony can cause more harm than good. Maybe look into adjusting that.

I hope this helps.

Jeff

4 Likes
5

That’s what I was thinking too.

6

Triton extreme let me choose between 64, 128 and 256 voices. 128 is the default. Would it be safe to choose the lowes option (64) with no compromisse of sound quality?

7

I would think so.

But, try it and listen how it sounds. Even if you notice some difference, I suspect there will be no difference in the context of playing with a band.

8

So I have the Triton Extreme for testing purposes, never really did anything with it beyond determining whether it worked.

I cannot find anywhere to change the number of voices (I have version 1.1) and from looking at the manual that may be something only available in the standalone version

However, I noticed that there is an option to change between Digital, Classic, and Studio and I observe that Studio and Classical are around 16% but Digital reduces the CPU to 6%

Didn’t really notice much difference in sound quality

screenshot_7947

1 Like
10

It’s there, in GLOBAL options:

image
image1434×977 129 KB

11

I don’t think you every told us the sample rate and buffer size you use.

12

Oh, ok.

13

Sample rate: 41.100 kh
Buffer: 512 sam,ples

14

Hello
Thank you for your feedback.
I mostly use combinations from Triton. I create my own combinations, layering the keyboard and then import them to GP. So, some rackspaces are just one instance of Triton Extreme.
About RAM: Unfortunately my MAC only has 16GB RAM and its a Mackbook model that doesnt allow upgrade, so I will have to stick with it.
Should I lower the sample rate to 480 or 448 samples? Wont it make difference in terms of global quality sound?

15

Lowering the sample rate reduces quality according to the Nyquist theorem.

But you are talking about sample buffers, right? I

17

How much latency do you have with 512 samples buffer?

18

You are right. I mean de buffer size (currently set at 512)

19

I really dont know. According to GP, I should have 11.6 ms

20

Have you tried it with 64 voices and lowered the settings David mentioned to “digital”?

That might do it without any perceptible negative effect.

21

512 buffer size is (IMO) too high for real-time performance. 256 or even 128 is better for real time.

Do you mean 44,100 kHz? 41100 is unlikely to be a valid sample rate

That should be plenty

22

Totally with you on this one … I started off using some awesome strings and ended up with Yamaha Motif XF strings section which uses about 75% less CPU and sounds just as good in a live situation… No need at all to use massive libraries for live gigs …

Here in Spain we play much varied styles of music, from 80’s pop/rock to latin style… Some latin songs I start off with an awesome ballad piano with all the bells and whistles (big fat sound with lovely reverb and pad, etc 40-50% CPU) but then I change it to a simpler piano ONCE the song changes in to merengue style with Piano tumbao technique… I don’t NEED the massive piano sound to play tumbaos … Look for what works GOOD ENOUGH in each moment.