For Soundeasy Digital Filter Users

[JMB]

I posted this over at the Soundeasy Board but have not yet had a response. I was hoping that someone here might be able to help.

I believe that there is a way to measure the filters we build on the CAD prior
to building the crossover in Soundeasy. I believe that I understand how to listen to the speakers with the crossover prior to building it (playing music through them, for example); for measurement, can we just set up the digital filter and then run EasyLab to measure the frequency response or are there specific concerns that need to be addressed or specific set up parameters to know about?

Thanks,
Jay

[john k...]

You can not run the digital filters and measure them with a single instance of SE. If (big if) your PC is very powerful and fast, and has two sound cards installed, you may be able to open a second instance of SE and use it with second sound card to measure. This depends on the ability of the MLS tool to run without interfering with the DSP tool. However, most likely you will need to run two instances of SE on two different PCs. One for playing the filters and one for measuring.

[dlr]

Originally Posted by john k...

You can not run the digital filters and measure them with a single instance of SE. If (big if) your PC is very powerful and fast, and has two sound cards installed, you may be able to open a second instance of SE and use it with second sound card to measure. This depends on the ability of the MLS tool to run without interfering with the DSP tool. However, most likely you will need to run two instances of SE on two different PCs. One for playing the filters and one for measuring.

I wonder whether or not multiple core systems could be used for this? In Windows with hyper threading on a single processor you can go into the task manager and set "Affinity" (I believe is the term) for any specific process so that it will run only in one thread. On a true multi-core system I would think that this would be possible so that you could have the two instances of SoundEasy running in separate cores. This also ought to be fast enough since each one runs at the speed of the individual core. I don't know much about what resources must still be shared, memory ought to be dedicated to each process, maybe the memory bus is shared and the video card at minimum.

You could go so far as to have two video cards, but that's supposed to be a tricky proposition and they both have to be PCI from what I know, you can't have AGP and PCI video in use simultaneously.

Dave

[DrDyna]

The number of video cards really shouldn't make a difference. I don't see a reason why a modern computer built in the last year or so couldn't run two instances without having to fiddle with thread affinity or priority. Having two sound cards is an easy enough task, my htpc has two sound devices for talking to multiple d/a converters. I'm not sure how all this works with their usb dongle, though.

Rather than use an entirely different pc, if 2 instances of the program interfere with each other, you could use VMware and load your second instance in a virtual machine.

[dlr]

Originally Posted by DrDyna

The number of video cards really shouldn't make a difference. I don't see a reason why a modern computer built in the last year or so couldn't run two instances without having to fiddle with thread affinity or priority.

I'm sure they can run two. The issue is CPU load. SE takes significant CPU time. If the task switching isn't fast enough, one or both are going to have a problem, the D/A and A/D are critical for real-time. The reason I made the suggestion is to isolate each instance within separate CPUs thus separating the loading.

Rather than use an entirely different pc, if 2 instances of the program interfere with each other, you could use VMware and load your second instance in a virtual machine.

The problem is the CPU load, a virtual machine doesn't solve that.

Dave

[john k...]

You can definitely run two instances of SE on one PC with two sound cards (and one dongle) and try to make measurements. I tired it years ago. The problem was the CPU load and timing issues. Sometimes I got a reasonable measurement, some times crap. Now that was with an older PC which is why I suggest that with a faster, newer PC it may work ok. But obviously, two PCs works which is what I do now.

[JMB]

Thanks all. Makes sense. I will try it with a second PC as I am under a little bit of a time crunch but if no one else has reported on this by the time I am done (I anticipate another few weeks to finalize the design and build of the speakers as I am doing 11 simultaneously for a media room), I will likely come back to this to try it and will report back.

Jay

[JMB]

I am guessing that the setup would be Soundcard output of the measuring computer to soundcard input of the filter computer and then soundcard out of the filtering computer to the amplifiers and on to the speakers. The input of the measurement computer, would of course be, from the mic and mic preamp. Is this correct?

[jerome69]

With the new technology of computer no problem to handle streams (perhaps not the I/O PCI BUS, not very young). With the old family Pentium 4 it was possible to acquire video streams.
The family Core i processor is very powerful. If the drivers of the sound card are ASIO, few latency can helps.
ASIO was created to use multiple sound card on a PC in realtime.
For ASIO see : Audio Stream Input/Output - Wikipedia, the free encyclopedia
I don't know if the software Soundeasy supports ASIO ?

You can test your PC with this : DPC Latency Checker
Some devices can cause latencies, this tool helps to find.

[DrDyna]

Originally Posted by dlr

The problem is the CPU load, a virtual machine doesn't solve that.

Dave

I meant it as a way to solve the problem of the two instances of the program fighting with each other, but if the only problem with running it twice is CPU load, I don't see where modern machines would have a problem with it.

I wouldn't try it on a 10 year old prescott P4, but does anyone still use one of those?