2024 updates - CPU: m7i / c7i / r7i

[szilard]

On r4.8xlarge's (Intel Xeon CPU E5-2686 v4 @ 2.30GHz) 16 physical cores (no HT) (current benchmark):

Tool	Time[s] 100K	Time[s] 1M	Time[s] 10M	AUC 1M	AUC 10M
h2o	11	12	60	0.762	0.776
xgboost	0.4	2.7	40	0.749	0.757
lightgbm	2.3	4.0	20	0.765	0.792
catboost	1.9	7.0	70	0.734	0.735

On m7i.8xlarge (Intel Xeon Platinum 8488C CPU @ 2.4GHz) 16 physical cores (no HT):

Tool	Time[s] 100K	Time[s] 1M	Time[s] 10M	AUC 1M	AUC 10M
h2o	8	8	26	0.762	0.776
xgboost	0.24	1.4	22	0.749	0.757
lightgbm	1.6	2.6	10	0.765	0.792
catboost	0.9	4.5	45	0.734	0.735

m7i, c7i, r7i have now all the same CPU: 3.2 GHz 4th Generation Intel Xeon Scalable processor (Sapphire Rapids 8488C)
r4 was new in 2017 (when this GBP-perf repo started, so kept for consistency)

Speedup is about 2x across the board (from 2017 to 2024 hardware, @ same number of cores that is 16, no HT).

Using all the cores of the newest CPUs (c7i.metal-48xl (Intel) and c7a.metal-48xl (AMD)), some further speedup is possible, see results here: #59