Latest build - "lagginess" in fullscreen mode?

RetroArch Linux devices
#1

This is a weird one, and it’s driving me nuts. Building from the current code in git and Retroarch keeps freezing at random intervals for up to a couple of seconds, then carrying on. Happens on more than one machine (tested on Arch Linux 32-bit and 64-bit), any core, no shaders or filters, only happens in fullscreen mode.

Running under X at all times, haven’t been able to try KMS mode as I hit a second issue where input isn’t working. I have opened a new issue on github:

https://github.com/libretro/RetroArch/issues/829

The Linux version of Retroarch has always been super-stable, not sure what’s going on right now. Any help would be appreciated!

#2

Hmm. Couldn’t reproduce here with latest git. Ubuntu 14.04 64-bit. kernel is 3.14.

Went back and forth from windowed to fullscreen a few times and let it run for awhile without any issues.

#3

Thanks for checking this out hunterk. I’m running LXDE / Openbox with no compositor of any kind. This is (in my experience) a “pure” X11 environment; I have only known issues with the frame rate and vsync creep in when using Gnome Shell or similar. Hence my confusion when Retroarch started misbehaving.

It’s possibly not a Retroarch problem, I’ll test further with some other emulators.

#4

Themaister commented on github:

“I’ve noticed this lately on Arch and Intel. Only in X. I believe this is a driver or X issue because it works just fine in KMS”

I am now sure it is NOT a Retroarch issue. I guess this is the risk you take when you run a bleeding-edge distro like Arch but I’ll do my best to find the actual cause. I have raised it on the Arch Linux forums for starters:

https://bbs.archlinux.org/viewtopic.php?pid=1439012

#5

Cool. Which IGP do you have?

EDIT: this could be related? http://www.phoronix.com/scan.php?page=n … px=MTc0Nzk

#6

The chip is GMA 3150 in a 3-4 year old Netbook. This behaviour was introduced in xf86-video-intel-2.99.912-1; with the previous package - xf86-video-intel-2.99.911-2-x86_64.pkg.tar.xz - the fault does not occur.

I think you might be right about the DRI 3 link, thanks for the information. I have included this in a follow-up post on the Arch Linux forum, but it hasn’t had any replies as yet.