Table 2.

Comparison of various photon migration Monte Carlo codes, according to the following parameters: RNG; the absorption and scatter coefficients ${\mu }_{\text{a}}$ and ${\mu }_{\text{s}}$, respectively; the size of the scoring grid; the acceleration platform; the photon throughput ${\mathit{T}}_{\text{acc}}$ and ${\mathit{T}}_{\text{ref}}$, expressed in photons/ms, for the accelerated and reference (CPU) platform, respectively; and the speed up achieved.

Implementation	RNG	${\mu }_{\text{a}}\left({\text{cm}}^{-1}\right)$	${\mu }_{\text{s}}\left({\text{cm}}^{-1}\right)$	Scoring grid	Platform	Threads	$T_{\text{acc}}$	$T_{\text{ref}}$	Speed-up
MC-HP	RAN3	0.0025	7.8	256 × 256	EC2 cloud	3,840	75,471	60	1258
MCX (atomic) (Ref. 16)	LL5	0.05	10.1	60 × 60 × 60	G92	512	9001	12	75
MCX (approx.) (Ref. 16)	LL5	0.05	10.1	60 × 60 × 60	G92	1792	3,8001	12	325
WMC-GPU (Ref. 15)	MWC	0	90	2001	8800GT	26,880	1,747	1.6	1080
GPU-MCML (Ref. 17)	MWC	0.015	708	100 × 100	GTX 480	13,440	384	0.4	870

¹Estimated.