I have 16 available cores each capable of 4.5 GHz but I get only 1,186,383.53 Keys/s when running your code. When I check utilization in top or htop I see there are only about 4 cores utilized at 100% but even then I would have expected higher rate. Anything I'm missing ?

I'm using secp256k1 from iceland2k14. When I manually set core for instance to "15" it utilizes 15 cores, but I would have expected that the cpu_count enumeration had worked. Seems it didn't, despite that ... With 15 cores I can achieve 2003121.42 Keys/s that is 2 MKey/s. You said you're achieving 600 MKey/s, how do you do that with CPU only? Wouldn't it be better to utilize GPU for such things?