[SOLVED] Handbrake on 18.04 Neon totally broken?

[oshunluvr]

Anyone got handbrake working?

Here, the default installation crashes with "illegal instruction" and the stebbins ppa crashes with "segmentation fault"

dang, no encoding until?

[SpecialEd]

Seems to be working fine here. I use it all the time but due to previously mentioned, self-inflicted, weekend shenanigans, I just reinstalled it via Discover and it is working as I type.

[oshunluvr]

Yeah, I think I found out my video card was loose.I had to move because Florence so maybe that was it,Anyway, I'm trying again here...

[oshunluvr]

Yeah, I don't know what the issue is, but it doesn't seem to be handbrake.

[oshunluvr]

Well, I reseated my video card and ran memtest all night. This stopped my computer randomly shutting down, but Handbrake still won't run. The stable version seg faults and the daily build version gives an illegal instruction error. Both versions crash .2% into a conversion.

[oshunluvr]

Still don't know what the problem is. I've tried handbrake on 16.04 and 18.04 neon and Kubuntu no go. It must be something hardware related but I can't figure out what. Ran Memtest overnight with no errors, ran GPU stress and benchmark tests with no errors, that only leaves Mobo or CPU.

I've seen one other issues of late - if I have a browser open for awhile with lots of open tabs, the system will eventually (while fiddling with the browser) lock up for a bit, then reboot itself.

[dibl]

... the system will eventually (while fiddling with the browser) lock up for a bit, then reboot itself.

Sounds like thermal or memory -- I assume you've monitored and checked out both. If they're not suspected, then I would turn my attention next to the video card, and finally the mobo. BTW, a few years back, I went through a similar problem, convinced myself it had to be the video card, bought a replacement nice Nvidia card, only to learn the hard way that it really was the mobo after all. Not a fun experience.

[Snowhog]

It could possibly be the PSU as well.

[GreyGeek]

Mother Nature always gives the test before the lesson!

[oshunluvr]

PSU is nearly new - not that that's a guarantee - Mobo is even never but same caveat. I need to re-seat everything then really nail down what-when causes the problem, then I can start digging further. If I was to bet, I'd lean toward mobo, but we'll see. I've been monitoring the thermals since this started and nothing out of whack. The CPU is water cooled and the GPU fan speeds up as I would expect.

[oshunluvr]

Ok, turned out NOT to be a Handbrake problem.

I did some googling and after a few hints, that and the lack of any logging meant it was decidedly a hardware issue of some sort. So I began to attempt some hardware monitoring during the crashes.

The exact behavior I saw was I would launch Handbrake and begin a transcode. After 0.10-0.25% (a few seconds into the operation) program would segfault and quit OR the GUI would freeze for about a minute and then reboot.

Then I noticed one of the changes I had not brought over from 16.04 was using the Thermal monitor in my panel. I recalled that handbrake would make my CPU temps spike. So I installed the widget and saw the temps rise quickly before the freeze, but i never saw a thermal "panic." The GUI would freeze before that. One of the hints I uncovered was a reference to over-clocking and lack of voltage. I run my system "goosed" to 4.7GHz (stock "Turbo" is 4.4) but I had never needed to move the voltage off of auto on this mobo. So I went into BIOS and set the CPU back to stock speed and Handbrake transcoded without issue.

So I went back, bumped it to 4.6GHz and increased CPU voltage to a fixed 1.395 volts rather than auto and tried Handbrake again. This time it got to about 20% before freezing and rebooting. Finally, added a few % points to the RAM voltage and Handbrake transcoded three full length movies no problem.

I have an i7-6700K CPU so it's meant to overclock, but I guess it's aged enough that it needs more help than the BIOS was offering. Likely, the BIOS only allows a certain level of voltage increase when using the "auto" selection and my i7 and/or RAM needs a bit more than it did before. My previous CPU - a Q6600 with B0 stepping - went about 3-4 years at 3.5GHz (stock was 2.4) before I had to start slowing it down. It finally became unstable at anything above 3GHz but it stayed there for it's last 4-5 years of service. Probably if I upgraded the RAM and tweaked the other voltages a bit I could get this i7 stable near 5GHz, but it's probably not worth the headache. This i7 is so much faster than the Q6600 it really doesn't need much extra.