Managing CPU Power

The following article was published in the October 2004 issue of EQ Magazine, and appears courtesy of EQ and Music Player Network.

http://www.eqmag.com/

September 2004

By Todd G. Tatnall
As soon as computer manufacturers ship faster models, software developers find ways to soak up all that power and speed. When a computer starts to have problems keeping up, I hear, “But I just bought the newest computer.” Granted, you can do a heck of a lot on the fastest Mac or PC. But you still want to be aware of a few key things to make sure you don’t run low on power.

YOUR ASSIGNMENT IS. . . .
Do your homework before buying your DAW, and you’ll get far before you have to kick into power conservation mode. Keep in mind that processor speed is only part of the power equation. Other factors include the speed of the system (front-side) bus, amount of RAM, and having a second hard drive dedicated to audio recording/playback.

Most recording applications have a “gas gauge” to tell you how hard your CPU is working, how much RAM you’re using, as well as other details. If you keep that performance meter open, and check it often, you’ll be able to take steps to conserve power before you have a problem.

BUFFER THE CPU SLAYER
Hardware buffers are important to understand and manage. DAW software and hardware use buffers to transport “packets” of audio in and out of the audio interface. A larger buffer is able to manage data throughput more efficiently and lighten the load on your CPU. If buffers are too small, the processor may not be able to keep up with all the tasks it’s being asked to perform. The trade-off with higher buffers is increased latency, or the delay you hear when monitoring playback from your software while recording live.

Locate the buffer setting in your DAW software and get used to adjusting it. As a rule of thumb, keep buffers low when tracking to reduce monitoring latency, and raise buffers when mixing to optimize CPU usage for things like plug-ins. You may be able to leave buffers higher if your software or hardware offers a low-latency monitoring mode.

PLUG-IN MANAGEMENT
Newer computers can run an absurd number of realtime plug-ins simultaneously. But certain effects still demand a fair amount of power. So even if you’ve got a speedy new CPU, it may still be necessary to manage plug-ins.

Most applications allow you to “bounce” a track to disk, which creates a new audio file with any inserted plug-ins rendered to the file. Then you can disable (or remove) the original audio track, replacing it with the new bounced file.

The latest versions of several DAW applications allow you to “freeze” a track. Much like bouncing, freeze automates the process of rendering a track to an audio file, then disabling the original track, into a simple step. Applications with this feature typically can “un-freeze” the track should you wish to make a change to the original track or its effects.

Most audio applications also offer aux sends and returns to allow several tracks to share one effect – just like when people mixed on analog consoles and had a rack of outboard reverbs. The process is pretty much the same. With hardware, each audio track has one or more “sends,” which split the audio on that channel and route it to another output, such as an outboard reverb. The output from that outboard device is sent back into an aux return and blended with the mix. Software applications have the same capability in their mixing windows. Typically sends can be assigned to a bus. An aux track whose input is that same “bus” has a plug-in inserted. The result is that several tracks busing through an aux track can use one plug-in, instead of each track requiring its own.

VIRTUALLY HUNGRY
Virtual instruments are probably the most power-hungry apps in modern DAWs. Sampler instruments capable of loading very large libraries such as MachFive, SampleTank, and Kontakt can demand a large number of processor cycles. Specialty instruments can do the same. Some applications can “freeze” instrument tracks as described above. If you can’t freeze a track with your application, it may be necessary to assign the output of your instrument track to a bus, assign the input of a new track to that bus, and record the instrument to the new track. By recording an instrument track as digital audio, you can disable the instrument and save some power.

Users often fail to realize that most virtual instruments can be multitimbral, with each channel having its own patch and receiving on its own MIDI channel. Opening two instances of an instrument when one can do the job really wastes resources.

Many virtual instruments offer power saving features. Polyphony settings allow you to raise or lower the number of notes an instrument can play at one time. Lowering polyphony to just what you need can reduce the amount of work your system needs to do. Some instruments offer a 32-bit mode, which saves CPU power at the cost of using more RAM. This can help with larger samples, if you have plenty of RAM.

The advanced tweaker can conserve power by customizing presets by removing layers or key ranges they’re not using, disabling unused filters and effects, and shortening long envelope releases (if they’re not being used) to prevent using up polyphony.

The best advice is to be aware of what your system can do, and to be ready to implement these steps before it’s too late. Don’t forget – watch those performance meters!

Knowledge Base