KVR Audio

buckman · Post by **buckman** » Wed Aug 04, 2004 9:10 am

Does USB devices eat processor power?

I have a 1.7Ghz laptop but want to use USB devices for my Midi keyboard (usb) Midisport 4x4, and my usb soundcard.

Will this compromise the laptops power at all?

Sepheritoh · Post by **Sepheritoh** » Wed Aug 04, 2004 9:28 am

I can give you only my experience. I'm sure that others will have something to say as well.

My previous computer was a 1.1 Ghz celeron P3. I have the evolution USB keyboard that used up about 20% CPU power without doing anything. When I installed the midi cable the usage went doen to about zero with almost no latency.

As soon as I added any synths with the USB keyboard the CPU usage became unreasonable and I experienced constant drop-outs.

This was an USB problem I tried a USB modem on that same machine and CPU usage shot up to 60% without doing anything. Using Sonar became a impossibility with the modem activated. My workaround that one was to have different user profiles in XP with USB activated and de-activated for music work.

My new computer is a 2.8Ghz P4 and the CPU usage is much lower from the USB (about 5%). I got rid of the USB modem so I do not know if that will make a difference on the new machine.

I do not know how true it is, but I was told that the problem with USB is that all data convertion and compression is done through software drivers, which obviously need CPU power. With PCI a reasonable amount of that processing and convertion of data is done on chip.

my 1/2 cent

Kim Lajoie · Post by **Kim Lajoie** » Wed Aug 04, 2004 11:24 am

Sepheritoh wrote:I do not know how true it is, but I was told that the problem with USB is that all data convertion and compression is done through software drivers, which obviously need CPU power.

That's what I've heard too.

To add to the unjustifiable claims, I've also heard that Firewire doesn't use any CPU for data transfer...

Forever,

Kim.

guido01 · Post by **guido01** » Wed Aug 04, 2004 8:32 pm

I've just noticed the last couplle of days that my Evolution MK-225C definitely uses up CPU cycles. I had a couple of projects I was working on that were pretty near to 100% CPU that suddenly stopped running - the difference being whether or not the keyboard was switched on. Back to MIDI I guess!

Trevor

kybernaut · Post by **kybernaut** » Wed Aug 04, 2004 8:55 pm

Independent from driver bugs etc. USB involves by design the CPU. IEEE 1394 does not.

--kybernaut

declassified · Post by **declassified** » Wed Aug 04, 2004 9:04 pm

kybernaut wrote:Independent from driver bugs etc. USB involves by design the CPU. IEEE 1394 does not.

So, if i used an USB audio/MIDI interface and connected the keyboard to its MIDI port, would that solve the problem?
Or do USB audio interfaces eat the CPU as well?

stefankuhn · Post by **stefankuhn** » Wed Aug 04, 2004 9:37 pm

timewastin wrote:
kybernaut wrote:Independent from driver bugs etc. USB involves by design the CPU. IEEE 1394 does not.
So, if i used an USB audio/MIDI interface and connected the keyboard to its MIDI port, would that solve the problem?
Or do USB audio interfaces eat the CPU as well?

Anything you stick up into your computer's USB ports will eat CPU, the more data is transferred the more CPU is utilized.

This is because the actual data transmission is handled by the CPU, there is no DMA (direct memory access) or any other neat hardware features that can automatically handle the transfer and communication protocol. At least to my knowledge.

Unlike Firewire (IEEE 1394) which is a real BUS system with a hardware driven protocol and DMA support.

(DMA = The CPU only tells some "DMA chip" to get the data from this device into that memory location, and the chip will take care of getting the data and putting it into memory, while the CPU is free for other things to do. Probably down in the pub drinking with friends, again.

)

Best,
Stefan

kybernaut · Post by **kybernaut** » Wed Aug 04, 2004 9:42 pm

I only wanted to point out that USB always needs some CPU cycles. It's like DMA vs. PIO mode for harddisks... PIO always involves the CPU like USB does. The percentages above are really high, though, and are rather due to bad drivers.

The Problems with USB MIDI interfaces seem pretty much gone nowadays (apparently not with directly attached USB keyboards...)

Concerning Audio interfaces, I would go the Firewire route whenever possible!

--kybernaut

[Edit]: Stephan Kuhn was a bit faster than me

AndreasE · Post by **AndreasE** » Thu Aug 05, 2004 8:33 am

Is the CPU load also increased when having attached to the PC other USB devices (Scanner, Printer, DSL modem, ...) but not simultaneously using them when playing midi/audio?

usb eats processor?