I've run CoreMark PRO on one of those free Ampere 4-core + 24 GB RAM instances, running Oracle Linux 8.4. TL;DR: that free instance performs very very well.

I've run CoreMark PRO on one of those free Ampere 4-core + 24 GB RAM instances, running Oracle Linux 8.4. TL;DR: that free instance performs very very well. WORKLOAD RESULTS TABLE MultiCore SingleCore Workload Name (iter/s) (iter/s) Scaling ----------------------------------------------- ---------- ---------- ---------- cjpeg-rose7-preset 625.00 172.41 3.63 core 6.86 1.72 3.99 linear_alg-mid-100x100-sp 595.24 154.80 3.85 loops-all-mid-10k-sp 29.96 7.80 3.84 nnet_test 28.09 8.45 3.32 parser-125k 190.48 50.00 3.81 radix2-big-64k 3289.47 848.90 3.87 sha-test 1000.00 277.78 3.60 zip-test 500.00 125.00 4.00 MARK RESULTS TABLE Mark Name MultiCore SingleCore Scaling ----------------------------------------------- ---------- ---------- ---------- CoreMark-PRO 23719.00 6305.03 3.76 For comparison, this is the result on my laptop: Intel Core i7-10850H @ 2.70 GHz, with 32 GB RAM and RHEL 8.4 x86_64: WORKLOAD RESULTS TABLE MultiCore SingleCore Workload Name (iter/s) (iter/s) Scaling ----------------------------------------------- ---------- ---------- ---------- cjpeg-rose7-preset 833.33 166.67 5.00 core 10.31 1.99 5.18 linear_alg-mid-100x100-sp 1063.83 236.97 4.49 loops-all-mid-10k-sp 34.29 8.55 4.01 nnet_test 36.10 8.64 4.18 parser-125k 111.11 23.81 4.67 radix2-big-64k 2890.17 716.85 4.03 sha-test 909.09 208.33 4.36 zip-test 666.67 200.00 3.33 MARK RESULTS TABLE Mark Name MultiCore SingleCore Scaling ----------------------------------------------- ---------- ---------- ---------- CoreMark-PRO 27058.61 6251.38 4.33 So that free Ampere instance performs slightly better than my i7-10850H on single-core, and just 10% slower on multi-core (and that's considering my i7-10850H has 6 cores, while the Ampere has only 4 cores!!!).