From Genome Analysis Wiki
→Why and how to perform a 2-step imputation?
mach1 -d sample.dat -p sample.ped -s chr20.snps -h chr20.hap --compact --greedy --autoFlip --errorMap par_infer.erate --crossoverMap par_infer.rec --mle --mldetails > mach.imp.log
In step1, one can use --greedy in combination with --states XX in MaCH versions 16.b and above. We have found that using 1/3 of the reference haplotypes (with 1/9 computational time) results in almost no power loss for the current HapMap and 1000G reference panels.
In step2, each individual is imputed independently and can therefore be split into as many as n (sample size) jobs for each chromosome for parallelism.