I’ve run into something (I think) is odd. I have a 6 core hyperthreaded Xeon E5-1650 v4 based machine and was asked to find optimal settings for local Unreal building. I run a “full build” test by deleting all the derived data cashes. Last time I did this test about a year ago, when loading up the editor or doing a “launch” from the editor I would usually see a large number of ShaderCompilerWorker threads get started up, usually up to the total number of threads the machine could handle (in this case 12).
Today I’m running the same test but I only see 2 ShaderCompilerWorker tasks and the build/launch takes much longer.
I am using VehicleDemo as the test project for this.
All the build configuration files on my machine were empty, so I tried adding the buildconfig like this
<?xml version="1.0" encoding="utf-8" ?>
<Configuration xmlns="https://www.unrealengine.com/BuildConfiguration">
<BuildConfiguration>
<ProcessorCountMultiplier>6</ProcessorCountMultiplier>
<MaxProcessorCount>6</MaxProcessorCount>
</BuildConfiguration>
</Configuration>
I know it is being read by unreal because I made a typo in an earlier version of the file and Unreal complained.
However even with this file it makes no difference, I still get just 2 shadercompilerworker threads.
Can someone tell me what I am doing wrong? Why does changing BuildConfiguration.xml have no effect?
Also, why did older unreal versions use all the cores while 4.19 seem to only use 2? Were the defaults changed?