If you have set up a SharePoint 2013 farm, for example for your development box, chances are you have wondered where all that good RAM went, and why performance is so slow when you start to add service applications to your farm.
A check in Process Monitor reveals that a fair chunk of RAM is used by a number of processes called Noderunner.exe (Microsoft® SharePoint® Search Component). These are the components you configured when creating your Search Service Topology. If you did this in PowerShell you can decide which components should run on what server (eg. Analytics Processing or Admin component).
In beta these processes had a memory leak, and the "cure" was to limit the space used by changing noderunner.exe.config. This however is *not* supported, and not recommended, so don't do this on production environment! If your components run out of memory they will start acting very weird and/or crash. I even have managed to kill my farm beyond repair by setting the memory limit too low, so consider yourself warned!
Another option for your dev farm is to change your performance level of your search service using
Set-SPEnterpriseSearchService -PerformanceLevel Reduced
This will reduce the maximum numbers of threads to the number of available processors. Note that this will only have effect on the server instance running the crawl component.
For more information on this topic read "Scale search for performance and availability in SharePoint Server 2013" on TechNet and the guide "Enterprise search architectures for SharePoint Server 2013".