So Graphics jobs was introduced as experimental long time ago. But its been very quiet about the feature from UT ever since. Its still in experimental, it still crashes the player both in DX11 and Vulkan.
Now when 6 and soon 8 cores are the norm for gaming PCs it would be nice if we could distribute the render load over multiple cores
6 and 8 cores still definitely aren’t the norm for your average gaming pc, but regardless, I share your concern about graphics jobs. Has anything changed about it in recent times?
Did you report bugs for said crashes?
Clicked thread expecting to type something about using Unity Connect… 
Even with only 4 cores 8 threads you benefit from graphics jobs. In the mainstream segment maybe 4 cores is still the norm, but VR and none casual gamers are moving toward Ryzen and Intel 8700k
But never the less its still interesting getting this feature for anyone with more than one logical processor?
Truth be told I don’t bother with that for experimental features, I use that time to report bugs on stable features 
But its easy to recreate, just check the Ghraphics Jobs, build game and it will crash almost instant. I test it once every new release I upgrade to
Oh I definitely agree that it’s still very useful for 4C/8T and even 4C/4T (which I’d say IS the common gaming CPU). Maybe VR, sure, but you have to realise they are by no means the average gamer, but high end and/or a niche. I’d say mainstream and even medium core gamers will still be on 4C, most games just don’t benefit from more so there isn’t the incentive yet
Odd, I use graphics jobs without crashes
We are getting there fast though the new Ryzens are crazy cheap, and I bet the 1700x will drop now
We use forward rendering and SteamVR, single pass rendering, soon we switch to single pass instanced when all our shader bugs are fixed (Though instanced only saves on batches and not set pass calls and for us it seems the set pass calls are by far the most expensive thing). I dont know if its a combo of those. But Im pretty sure I tried it in a vanilla project too. Maybe its becasue I’m on 16 threads?
Vanilla forward rendering for me, no issues yet. The only issue I’ve found is that if you stop remote profiling a remote dev build it crashes
Any performance gains?
I think I saw a speed boost but I haven’t compared in a long time, was worth turning on
I tried with 2017.4.1 and no crash for the short time I tried, thats nice!
But also no real perfomance gain either. I’m on a 8 core ryzen 1800x
Here with Gfx jobs enabled, watch the CPU frame time above graph
And here without gfx job
This part of the map culls pretty bad so its many set pass calls here
Tested jobs a bit more, and it seems it distributes work over more cores
Right side is with Graphics jobs. Since frame rate is locked to 90 fps in SteamVR GPU usage will not change. But the frame time for teh CPU is not changed either, isn’t that strange?
Do you test with builds or in the editor?
Also are you sure there isn’t anything else bottlenecking your frametime on the cpu?
Built, can’t really bench in editor, SteamVR has a built in tool which calculates the time between frames both on CPU and GPU. Our domain is really thight and only adds a few fractions of a millisecond. There is no change in CPU frame time between gfx job enabled and disabled
https://store.steampowered.com/hwsurvey/cpus/
4 physical cores is a big chunk of gaming rigs these days. No idea if they are hyperthreaded or not…
Isn’t that what I claimed?
