Resolume Forum

Hello all!

Last night I had a catastrophic failure at an event, synced with SMPTE.

I had two different machines with completely different specs and different windows versions (7 and 8.1) running Arena 5.0.2. The machines were turned on and Resolume starated in an interval of five minutes and playing back just one large clip, synced via SMPTE (DXV3 Normal Quality, about 75GB).

Exactly on the 7th hour of work, the machine I turned on first started massive drops of the SMPTE, about 80% of the frames were dropped. Thankfully, this was the backup machine, but I started diagnosing, what could be the problem - rising the audio buffer didn't do anything, the SMPTE input volume was spot-on (around -6dB) etc. By the time I had decided to restart Resolume, the main machine (turned on 5 minutes later) started doing the exact same thing.

So I quickly restarted Resolume on the backup, et voila, no SMPTE drops! So quickly go to backup, restart the main, return to main. But in the end about a minute of the show had video projection bordering on slideshow.

I've not had this problem with Arena 5.0.0.

Any ideas?

Yes, we're aware of this problem.

It occurs only on PCs, and only after an extended period of use. You had it after 7 hours, I've also personally had it after longer periods like 8-10 hours.

We've spent a lot of time on it, so far it's proven very difficult to debug. This is because it doesn't always appear, and when it does, it only appears after several hours. This makes it very, very difficult to see if the changes we make have any effect.

So far, a workaround is to toggle your audio input buffer up one setting and then back down. So if it's on 1024, knock it up to 2048 and then back down to 1024. Apparently this resets something, and will make the SMPTE run normally again.

This was the first thing I tried, but it didn't work for me, so I restarted Resolume. I changed the buffer size from 1024 to 2048, no effect, then back to 1024, no effect, then to 512, the card didn't agree with this buffer and it stopped completely, then back to 1024 - and the problem continued. So I opted for a software restart and it fixed it.

It's strange, but my colleague had no problems at a gig with over 13 hours of non-stop SMPTE, but with 5.0.0.

Hmmm. I wonder if this a new issue then.

Would be great if you could do some more troubleshooting. Ie, does it always fail exactly after 7 hours? Does it only fail on 5.0.2 or also on earlier versions? Does it also fail for your friend when he's using 5.0.2? Are you using a external sound card? What sound driver are you using for your main output and your SMPTE input?

In our test setups, we usually start the test at the beginning of the day and then go about our other work, checking back in every hour or when we go to get coffee.

Well, on the second thought I'm not sure my colleague didn't use a MacBook and because of that he didn't have issues.

As for me, I powered up the workstation at 2:00 PM and it started dropping at 9:00 PM. The backup PC was powered up 5 min earlier and started skipping 5 min earlier than the workstation.

The workstation was an AMD Phenom II X4 965 with an AsRock 970 Extreme4 main board and I was using the integrated audio on a freshly installed Windows 7 64 bit with all the latest drivers. The only software installed on the workstation was Arena 5.0.2. All drivers were up to date with no old orphaned ones - I even used the latest AMD Catalyst Crimson for the MB and video card (which is Radeon 7970). There was nothing to interfere or add external reason for the issue.

The backup machine was a Lenovo Y50 laptop with i7 CPU and Windows 8.1 64 bit. Old installation with various software and drivers installed. Also used the integrated audio. The issue was exactly the same as the other machine in exactly the same timeframe and with the same amount of dropped frames (by eye, not measured).

I'll try to replicate this next week in the office and also test with some higher end cards (I have some firewire interfaces around - MOTU 896 and Lexicon FW810s).

I hope this helps to shed some light on the issue...