I really recommend looking into the full paper:
https://www.hertzbleed.com/hertzbleed.pdfThey point to the paper "Luca De Feo, Nadia El Mrabet, Aymeric Genêt, Novak Kalud̄erović, Natacha Linard de Guertechin, Simon Pontié, and Élise Tasso. SIKE channels. Cryptology ePrint Archive, Report 2022/054, 2022."
https://eprint.iacr.org/2022/054 which describes the main problem: a zero-value attack on the crypto algorithm. It is possible to construct input data for the algorithm SIKE where it can get on a sequence of all zeros in its calculation. They do describe in section 5 countermeasures.
Only because the selected crypto algorithm generates a sequence of calculations with zeros is there a big difference in CPU calculations, which lead to the CPU changing frequency. I don't see for now a general problem. Nonetheless software developers should be aware of this type of attack.