TABLE 2.

Computing time of GVCHAP using the Mesabi supercomputer.

	GVCHAP using one node with 62 Gb memory				Multi-node processing (MNP)
Number of individuals (n)	7549	7549	15,098	15,098	37,745a
Number of SNPs (m)	82,128	328,512	328,512	657,024	328,512
SNP Ag and Dg	0.12 h	0.71 h	0.70 h	1.32 h	2.67 h
Haplotype Agh	1.31 h	7.88 h	14.31 h	22.11 h	≈3 hb
Time per iteration	19 sc	37 sc	82 sc	62 sc	22–29 minc (one node)

^aHaplotype genotypes of this dataset could not be processed by a single node with 62 Gb memory limit and was used as an example using the MNP method. The haplotype genotypes were divided into 75 files each with 200 blocks for most of the 75 files that were processed using 75 different nodes. ^bEach ${\mathbf{\text{W}}}_{\mathrm{\alpha }h}^i{\mathbf{\text{W}}}_{\mathrm{\alpha }h}^{i^{\prime }}$ file required 0.831 ± 0.248 h to process, and the summation of the 75 ${\mathbf{\text{W}}}_{\mathrm{\alpha }h}^i{\mathbf{\text{W}}}_{\mathrm{\alpha }h}^{i^{\prime }}$ matrices and the calculation of ${\mathbf{\text{A}}}_{\text{gh}}={\mathbf{\text{W}}}_{\mathrm{\alpha }h}{\mathbf{\text{W}}}_{\mathrm{\alpha }h}^{\prime }/{\text{k}}_{\mathrm{\alpha }h}$ took about 1.66 h to complete. Adding 0.831 and 1.66 h indicates the calculation of A_gh approximately required 3 h. ^cThe computing time per iteration is affected by the number of jobs running on the Mesabi supercomputer.