drop asserts from benchmarks

[franziskuskiefer]

Also ensure nothing is optimised out (fixes weird p256 performance numbers).

[coveralls]

Pull Request Test Coverage Report for Build 3981397350

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 53.588%

---

Totals
Change from base Build 3959834694:	0.0%
Covered Lines:	30653
Relevant Lines:	57201

---

💛 - Coveralls

[duesee]

Since asserts are stripped in --release it's generally good we don't use them in benchmarks. So, from all the changes, I assume only this one -- moving r, s before the loop -- could be important? It doesn't seem to change much on my machine, though:

--------------------------------------------------------------------
Benchmark                          Time             CPU   Iterations
--------------------------------------------------------------------

--- franziskus/fixup-some-benchmarks, but with r,s before loop
P256_SHA256_ECDSA_Sign        328377 ns       328279 ns         2130
P256_SHA256_ECDSA_Verify      734751 ns       734636 ns          947
P256_ECDH                     320494 ns       320433 ns         2184
Openssl_P256_ECDSA_Sign        20389 ns        20383 ns        34030
Openssl_P256_ECDSA_Verify      54693 ns        54673 ns        12692
Openssl_P256_ECDH              38962 ns        38950 ns        17927

--- franziskus/fixup-some-benchmarks
P256_SHA256_ECDSA_Sign        325494 ns       325416 ns         2150
P256_SHA256_ECDSA_Verify      730341 ns       730129 ns          952
P256_ECDH                     318984 ns       318906 ns         2192
Openssl_P256_ECDSA_Sign        20162 ns        20154 ns        34424
Openssl_P256_ECDSA_Verify      54081 ns        54060 ns        12041
Openssl_P256_ECDH              38737 ns        38727 ns        18051

Do you see a difference on your machine?

[franziskuskiefer]

I don't see any changes in performance from this change either. It's dominated by the math.

[duesee]

Thanks! I left a question below, but in any case this is good to go!