This paper was celebrated at PWLMTL#2 by Peter Zion, who also used it as
a springboard to discuss GPU architecture in general, as well as his
implementation based on the paper in the Fabric Engine
framework (fabricengine.com).
Rather than artificially separate garbage collection from other classes
of memory management, I propose moving garbage collection topics under
the more general label of memory management ("storage management" or
"memory allocation" might be reasonable alternatives).
