REW Beta Release REW is a bit glitchy this time around

[Matthew J Poes]

hi Jon,

I am using a Motu 828x with my BSWA mic on a Windows 10 laptop, HDMI out, ASIO4ALL. I have the newest beta. Last night i was doing some dynamic and distortion testing of tweeters and ran into a bunch of problems.

First, it kept running out of memory. 4M sweeps seem to be only able to be done one at a time based on memory usage. I couldn’t get 4M to work last night and gave up.

When using longer sweeps and an acoustic timing reference it wasn’t hearing the timing reference. To get it to record the sweep I had to use no timing reference. Because I was connecting both the 828 and an HDMI, I couldn’t use a loopback as timing reference. Switching to the 828 as both mic and source with a loopback did seem to fix it (i had a timing reference at least) but the 4M length still didn’t work.

Stepped sine distortion test didn’t work. It would just sit there and do nothing. I never managed to get it to work no matter what settings I chose. I tried restarting but that didn’t help.

IMD distortion test didn’t work. Tone was generated but the RTA wouldn’t record anything. It locked up and I had to restart.

Measuring past 22khz doesn’t work unless you use the trick you posted. I know you already know this, but I missed that last night.

Using shorter sweeps but taking a bunch of measurements causes it to run out of memory. I’ve done this exact procedure before and never had this problem. The way it works is that I use the longest sweep possible for the first handful of measurements because the distortion at those volumes are so close to the noise floor. Then as the sweep increases in volume such that the harmonica are above the noise floor, I lower the sweep length to reduce strain on the driver.

How it behaves when it runs out of memory is less than ideal. I have to close and restart the program. It doesn’t let me do anything else. When I close it, it does let me save it. If it’s possible to make it simpler to increase memory (from within the program) and also make it so that when it runs out of memory it just discards the sweep so I can at least go in and label my files and save them.

 

[John Mulcahy]

Which installer did you use, Matthew? The Windows 64-bit installer now has an option to specify a higher memory allocation when REW is run, though that option is also available through manually editing the roomeqwizard.vmoptions file in the installation directory.

Java only allows applications to be run with a fixed memory allocation, there is no ability to dynamically alter it.

You can reduce REW's memory use per measurement by setting IR oversampler to "None" in the Impulse response calculation section of the Analysis preferences.

It is certainly possible to make multiple 4M sweep measurements, I have just run a series of eight 4M sweep measurements in a row without any issues, but I assigned 2G of memory to REW when I ran the installer. I'll remove the option for 2M and 4M sweeps for installations that don't have more than 1G of memory allocated.

I haven't encountered any problems with stepped sine or IMD measurements. What settings were you using in each case?

Edit: The series of eight 4M sweeps was actually made with a 1G memory allocation

 

[John Mulcahy]

Could you also go back through the REW log files and send me any with errors captured in them, Matthew?

 

[Matthew J Poes]

Hi John,

I had actually installed the 32 bit version. I've installed the 64 bit version and set it up for 2 gigs of memory. That fixed the long sweep problem, but other issues persist.

How do I find the log files?

 

[Matthew J Poes]

never mind, found them.

I switched to using the UMM-6 today for a project, so the logs should reflect that. I think it looks like these logs include the 32 bit version and the 64 bit version, and one is from yesterday.

Thanks for any help John.

 

[John Mulcahy]

Logs 6 and 9, which have errors captured, are the error that occurs with Check levels on the measure dialog requiring a custom range. I've already fixed that for the next beta. Would result in no output when trying to do the level check. A few of the logs show reset requests from the ASIO driver and a couple show the driver not providing any audio input data, which would stop things - would normally expect a message to be shown for that. If there isn't a requirement to use ASIO4All (e.g. only need to access front channels on HDMI) worth trying to run without it.

Can't see anything relating to stepped sine or IMD. What settings were you using when those were not working?

 

[Matthew J Poes]

Logs 6 and 9, which have errors captured, are the error that occurs with Check levels on the measure dialog requiring a custom range. I've already fixed that for the next beta. Would result in no output when trying to do the level check. A few of the logs show reset requests from the ASIO driver and a couple show the driver not providing any audio input data, which would stop things - would normally expect a message to be shown for that. If there isn't a requirement to use ASIO4All (e.g. only need to access front channels on HDMI) worth trying to run without it.

Can't see anything relating to stepped sine or IMD. What settings were you using when those were not working?

Hi John,

I will try the Step and IMD again tonight and record a video of what I do. I’ll load that and any log files to share.

This is for some projects so using ASIO is probably needed. I can do the java driver too, but it’s not a great solution. Plus this was all working fine before I updated.

I also found that on a few occasions the program would disappear. We can call it crashing, but it appeared to still be running in the background. I just couldn’t see it anymore. It happened when I hit measure. Instead of starting it just crashed. To fix it I had to open task manager and force close the program. Restarting the program fixed it. This didn’t use to happen.

Temporarily I can go back to a stable version, I just wanted to share what was happening for me.

 

[John Mulcahy]

You haven't relocated to the Bermuda triangle, have you? If the window disappears please try and capture the log file. If the Java runtime has run out of memory behaviour is unpredictable, however.

 

[Matthew J Poes]

You haven't relocated to the Bermuda triangle, have you? If the window disappears please try and capture the log file. If the Java runtime has run out of memory behaviour is unpredictable, however.

Will do. It did it 3-4 times last night, so I’m not sure why it wasn’t captured. If it is memory related, I don’t know. I have 12 gigs of ram, and I set this up to use 2 gigs, but it’s still happened.

I’ll try one more time tonight and if this still shows problems will revert to a stable version. I’ll send you any logs and videos tonight.

I am still in Chicago and it’s the same laptop I’ve had the last year.

 

[John Mulcahy]

It would be great if anyone else seeing behaviour like this could chime in. I'm using the latest builds pretty much all day every day and haven't seen anything similar, but there may be some combination of factors I'm not encountering.