Timings for a Masm64 SDK test

[hutch--]

The first was a single pass, the next was 1000 iterations. Just under 1000 gig in just over 600 ms, probably fast enough. 

[hutch--]

Here is a tweaked version, I use the Agner Fog StrLen version and unrolled the scan loop by 4 and its about 30% faster. On my old Haswell, I keep getting 438 ms.

[jj2007]

Very good

In case you need more inspiration