Here is a list of compile switches, used to customize Radiance code
for specific machines and users:

-DALIGNT=(type)	Alignment type, machine-dependent.  Most RISC
		architectures align on 8-word boundaries (double).
		The default alignment type is int.

-DWFLUSH=(rays)	Override for number of rays before flush in rvu.

-DBSD		Operating system has a strong Berkeley flavor, which
		affects certain system calls, such as signal handling and
		resource tracking.

-DSMLMEM	The system has little RAM available, so size hash
		tables and the like accordingly.  Only allows for
		small scene descriptions (32,256 primitives).

-DSMLFLT	This setting tells Radiance to use short floats
		(32 bits) throughout, which saves lots of memory
		but can cause calculation inaccuracies in many
		cases.  Its use has been discontinued for this reason.

-DSHADCACHE=N	The array size to use for caching occluders.
		Setting this to 0 turns off this optimization.
		Values greater than 20 use over 4 Kbytes per
		light source, which can add up to a lot of
		memory in scenes with many sources.

-DAVGREFL=R	Set the average surface reflectance for the purpose
		of indirect irradiance calculations to R, where
		R is a real number in the range (0,1.0).  The
		default value is 0.50 (50%).

-DPMAP_OOC	Builds the out-of-core photon map. The photon data resides
		in a leaf file on disk with suffix ".leaf" and is loaded on
		demand. This is intended for very large photon maps (up to
		4.3 billion photons) particularly for annual simulations
		with rcontrib. This switch is disabled by default, which
		builds the in-core photon map used in previous releases;
		this resides entirely in main memory. These options are
		mutually exclusive, and the resulting photon map files are
		incompatible.
		
-DPMAP_PRIMARYPOS	
		Stores photon primary hitpoints for binning using the -f
		option with rcontrib if the function file requires them.
		This consumes an additional 3 floats and is disabled per
		default to reduce the size of the resulting photon maps.
		
-DPMAP_FLOAT_FLUX
		Stores photon flux as floats instead of more compact RGBE
		(the default). In terms of accuracy, the difference would
		only be noticeable with highly saturated colours.
		
-DPMAP_ROI	Enables the -api option to mkpmap. This constrains photon
		deposition only to the specified region of interest (defined
		as a bounding volume) for situations where it is known a
		priori that photon map lookups will occur outside this
		region. This is disabled per default as it can lead to
		biased results outside the ROI, or near its boundaries. Ziss
		option for ze eckspertz only!

-DPMAP_RUNAWAY_WARN
		Causes mkpmap to issue a warning if a photon reaches the
		maximum number of bounces (as specified with -apm) and is
		terminated.
				
-DPMAP_LOOKUP_WARN
		Causes rtrace et al to issue warnings when short photon
		lookups occur, i.e.  fewer photons are found than specified
		as lookup bandwidth with -ap.  This is generally not fatal,
		but can increase noise, and may indicate a higher bandwidth
		is needed. This is disabled per default as it can result in
		excessive console clutter.

-DPMAP_LOOKUP_REDO
		Causes rtrace et al to repeat short photon map
		lookups with a progressively larger radius until the
		target photon count specified with -ap is met. This switch
		has no effect in conjunction with the -am option, since this
		disables lookup radius adaptation.

-DPMAP_EPANECHNIKOV
		Causes rtrace et al to apply an Epanechnikov filter during 
		photon density estimates. This weights photons according to
		the inverse distance to the lookup point in order to reduce
		bias. The default behaviour does not perform filtering.