Table 1.

Quality and speed of autopicking for the $\beta$-galactosidase benchmark. Comparing the CPU version with the GPU version using increasing levels of low-pass filtering yields progressively higher recalls at similar FDRs. The GPU version yields identical results to that of the CPU version, but at a much reduced computational costs. Filtering does not depend on GPU-acceleration, and will perform similarly using only CPUs.

DOI: http://dx.doi.org/10.7554/eLife.18722.011

Code	Filter	# picked	Recall	FDR	Time	Performance in
	(Å)	particles			(s/micrograph)	CPU core units
CPU	none	54,301	0.88	0.34	1,227	1
GPU	none	54,325	0.88	0.34	10	122
GPU	5	55,629	0.90	0.34	5.8	211
GPU	10	55,886	0.90	0.34	2.1	584
GPU	15	56,450	0.92	0.33	1.6	766
GPU	20	57,361	0.95	0.33	1.3	943