Emulator

Likely fixed in https://github.com/dolphin-emu/dolphin/pull/2580

Which Intel GPUs is this bug confirmed on?

What kind of performance hit should be expected if shadercache is disabled? I haven't tested the fix-PR myself (yet) but do you have any figures / educated guesses?

Also, I'm interested in the steps to reproduce the crashes, just so that this could be narrowed down more.

I have not had (not that I have played much with OpenGL, but occasional testing) the crashes on my HD 4600 /w latest drivers, and overall OGL performance seems good. Even better than D3D in some situations, SMG2 with 3D for example wes better with OGL, IIRC.

I'm not saying the crashes are not there, I'm just worried that Intel HD possibly gets a deal breaking performance hit with the fix, when the problem might be outdated drivers or outdated HW (HD 4000 vs 4600?), which might not concern every Intel HD user.

Run the same game twice in OpenGL on an Intel HD in windows, and that should trigger the crash.

Just tried with latest master 4.0-6729, with portable.txt .
Loaded New Super Mario Bros and Super Mario Galaxy 2 with default settings (OGL, of course). Exited dolphin, verified that User\ShaderCache has ogl-SMNP01-shaders.cache and ogl-SB4P01-shaders.cache
Started Dolphin, started the games. Clicked to title screens and a bit further, no crashes.
Again, I think this should be bisected more.

I'll test the fix-PR sometime soon, if I only have time, just to see what it does to performance.

If the only hit is that shader creation stuttering will occur every time when entering a new scene vs. only when entering it for the first time, then I guess disabling it is tolerable, as the short 1-4 second stuttering doesn't bother me much. If it really helps someone to run the game at all (on Linux or OSX, for example).

Disagree. If you can't reproduce it, compromising your experience for other people is not really fair. Can we compile a list of Intel HD cards and driver versions we know are working?

Fair enough :) Just noticed that I lied about having the latest drivers, my drivers were from earlier this year, 10.18.14.4156, and Intel has published not one but TWO newer versions in 2 months. Latest is .4222 , released just some days ago. Now that I heard about this issue, I might not want to update, since my current setup is working just fine with Dolphin, OGL or D3D.

But as a starter for this bisection, Win 8.1 with Intel HD 4600 and driver version 10.18.14.4156 works with OGL.

I became curious and couldn't resist updating to latest drivers 10.18.14.4222 . Repeated the test I did earlier (I did delete the shadercache files in between, just to be sure), no crashing. System is the same, Win 8.1, Intel 4670S.
So I'd say there is nothing (or maybe that's a bit overstated) wrong with Intel OGL drivers since March 2015, at least.

I haven't tested OGL on my Intel HD Graphics 2000 lappy for a while now. Will see if I can reproduce this

My apologies. I forgot that my HD Graphics 2000 does not support OpenGL 3, so it is not able to test for this issue. Sorry about the clutter. If it means anything, I tried loading Pacman World 2 multiple times with OpenGL (going past the "OpenGL is not supported" error lets you play the game, just no display) and launched into a level without any crashes occurring on both latest master (4.0-6734), and the disabled shader cache build.