Table 4 Average runtime in minutes (s.e.) of the Gauss–Seidel solver with and without randomizing the order of markers for updating marker effects, with increasing numbers of SNPs and environments, based on 10 replicates of scenarios 4 (4311 SNPs) and 5 (42,034 SNPs)
From: A new approach fits multivariate genomic prediction models efficiently
Randomized | Number of SNPs | Number of environments | PEGS | THGS | UV-THGS |
|---|---|---|---|---|---|
Yes | 4311 | 10 | 0.2 (0) | 0.2 (0) | 0.1 (0) |
Yes | 4311 | 50 | 3.5 (0.4) | 3.5 (0.4) | 0.6 (0) |
Yes | 4311 | 100 | 14.4 (2) | 14.4 (1.8) | 1.1 (0) |
Yes | 4311 | 200 | 80.5 (10.1) | 79.2 (11) | 2.3 (0.1) |
Yes | 4311 | 400 | 459.3 (55.1) | 448 (58) | 4.3 (0.1) |
No | 4311 | 10 | 5.5 (1) | 5.4 (0.9) | 1.9 (0.2) |
No | 4311 | 50 | 44.9 (7) | 44.6 (6.9) | 9.3 (1.1) |
No | 4311 | 100 | 120.9 (10.1) | 123.7 (9.9) | 20 (1.8) |
No | 4311 | 200 | 361.1 (48.9) | 364.6 (44.4) | 39.3 (2.8) |
No | 4311 | 400 | 1261.8 (115.8) | 1261.7 (107.9) | 74.1 (8.3) |
Yes | 42,034 | 10 | 0.8 (0.1) | 0.8 (0) | 1.2 (0.1) |
Yes | 42,034 | 50 | 9.9 (0.4) | 12.5 (1.3) | 5.7 (0.4) |
Yes | 42,034 | 100 | 36.4 (1.4) | 29.2 (2.7) | 11.3 (0.6) |
Yes | 42,034 | 200 | 123.2 (17.1) | 119.7 (10.1) | 22.5 (2) |
Yes | 400 | 730 (64.4) | 802.2 (118.2) | 46.4 (4.1) | |
No | 42,034 | 10 | 64a (14.7) | 64.2a (16) | 14.5 (5.1) |
No | 42,034 | 50 | 540.2a (38.3) | 536a (26.8) | 106.5 (63.2) |
No | 42,034 | 100 | 1109.6a (71.5) | 1148.1a (109.3) | 181.4 (40.6) |
No | 42,034 | 200 | 3057.3a (292.7) | 3001.2a (259) | 310.3 (114.8) |