I reran the Table 9 benchmark on July 16, 2026. On our server "maple" which has two Intel Gold 6342 CPUs. Only 1 core was used. The benchmark is the Maple code in the file "maplecode". The new timings in the files maple16.out, maple17.out, maple18.out, and maple2019.out are reported below. They contain the output from Maple 16, Maple 17, Maple 18, and Maple 2019, respectively. Also, the files maple2015.out, maple2016.out and maple2026.out contain output from Maple 2015, 2016 and 2026, respectively. Maple 16 Maple 17 Maple 18 Maple 2019 n expand factor expand factor expand factor expand factor ----------------------------------------------------------------------- 5 0.207s 26.62 0.030s 9.41 0.032s 8.13s 0.035s 0.86s 6 0.448s 94.12 0.061s 27.47 0.051s 52.0s 0.054s 1.79s 7 0.806s 4.54m 0.049s 1.37m 0.057s 1.68m 0.064s 3.09s 8 1.02s 8.08m 0.050s 4.31m 0.060s 3.52m 0.067s 4.22s 9 1.11s 10.8m 0.052s 4.97m 0.061s 3.65m 0.069s 5.07s 10 1.03s 17.3m 0.052s 12.5m 0.061s 5.17m 0.069s 6.04s 11 1.14s 22.2m 0.050s 9.59m 0.060s 6.96m 0.070s 6.91s ----------------------------------------------------------------------- 12 1.32s 39.4m 1.20s 38.1m 0.992s 27.8m 1.27s 77.85s 13 1.22s 77.2m 1.22s 98.6m 1.04s 74.4m 2.52s 1.66m 14 1.24s 1.75h 1.32s 2.14h 1.16s 28.4m 1.49s 2.36m 15 1.26s 59.6m 1.35s 75.1m 1.67s* 2.20h 7.00s 2.46m *I subtracted the garbage collection time of 19.10s At n=12, POLY overflows so the SUM-OF-PROD representation is used Maple 16 was released in 2012. It uses the old SUM-OF-PROD representation. Maple 17 was released in 2013. It has the new POLY representation. Maple 18 was released in 2014. It has some improvements to POLY. Maple 2019 was released in 2019. It has the Monagan-Tuncer factorization algorithm. Prior to Maple 2019, Maple used Paul Wang's factorization algorithm. Michael Monagan