bulk add demo

[mratsim]

Demo for devcon cc @asn-d6

Compiled with GCC
Optimization level => no optimization: false | release: true | danger: true
Running on 11th Gen Intel(R) Core(TM) i9-11980HK @ 2.60GHz


⚠️ Cycles measurements are approximate and use the CPU nominal clock: Turbo-Boost and overclocking will skew them.
i.e. a 20% overclock will be about 20% off (assuming no dynamic frequency scaling)

Backend: BLST, mode: 64-bit
====================================================================================================================================

Scalar multiplication G1 (255-bit, constant-time)                             10273.058 ops/s        97342 ns/op       321460 cycles
Scalar multiplication G2 (255-bit, constant-time)                              4501.341 ops/s       222156 ns/op       733647 cycles
EC add G1 (constant-time)                                                   1923076.923 ops/s          520 ns/op         1717 cycles
EC add G2 (constant-time)                                                    688231.246 ops/s         1453 ns/op         4798 cycles
EC bulk add G1 - 10                                                          244259.893 ops/s         4094 ns/op        13512 cycles
EC bulk add G2 - 10                                                           94348.523 ops/s        10599 ns/op        34992 cycles
EC bulk add G1 - 100                                                          29748.624 ops/s        33615 ns/op       111005 cycles
EC bulk add G2 - 100                                                          12546.579 ops/s        79703 ns/op       263202 cycles
EC bulk add G1 - 1000                                                          3880.797 ops/s       257679 ns/op       850947 cycles
EC bulk add G2 - 1000                                                          1464.804 ops/s       682685 ns/op      2254472 cycles
EC bulk add G1 - 1024                                                          3541.152 ops/s       282394 ns/op       932569 cycles
EC bulk add G2 - 1024                                                          1449.244 ops/s       690015 ns/op      2278671 cycles
EC bulk add G1 - 2048                                                          1906.930 ops/s       524403 ns/op      1731767 cycles
SIGSEGV: Illegal storage access. (Attempt to read from nil?)

As mentioned, there is a crash in BLST currently if we go over 2048 G1 points or 1024 G2 points, unsure why but BLST batches addition 2048 by 2048 or 1024 by 1024 ( https://github.com/supranational/blst/blob/6382d67/src/bulk_addition.c#L148-L149 ) so the benchmark scales linearly to a million points