Impact of sub-setting the data of the main Limousin beef cattle population on the estimates of across-country genetic correlations

Table 8 Computational requirements across scenarios based on single-core analyses

Scenario^a	Phenotypic records	Pedigree records	CPU (GHz)^b	RAM peak usage (GB)^c	REML rounds (number)	Average time per REML round (min)	Total time (days: hours)^d
ALL	3,115,598	3,431,742	4.0	2.39	1173	53	43:23
RND sample 15	533,816	706,717	4.0	0.49	1432	11	11:12
RND sample 20	521,077	692,205	3.7	0.48	1543	14	15:21
RND sample 2	556,100	729,778	3.7	0.51	1462	15	16:00
GSCB	506,080	654,841	4.0	0.46	1496	9	9:21
GSTOT	513,969	663,127	4.0	0.45	1530	9	10:06
HM	502,716	649,081	4.0	0.46	1675	9	11:06

^aALL, all data; RND, herds selected randomly; GSCB, herds selected based on genetic similarity considering common bulls; GSTOT, herds selected based on genetic similarity considering common bulls and common maternal grandsires; HM, herds selected based on harmonic mean of sire’s progeny size
^bCentral Processing Unit (CPU) frequency
^cRandom access memory (RAM) peak usage
^dTotal elapsed time

ISSN: 1297-9686