Add 256bit and 512bit vectorized hex decode

[benaadams]

Changes

• We already have a 128bit simd path; add 256bit and 512bit simd paths

Types of changes

What types of changes does your code introduce?

• Optimization

Testing

Requires testing

• Yes

If yes, did you write tests?

• Yes

[LukaszRozmej]

Tests & Benchmarks?

[benaadams]

Method	ByteLength	Mean	Error	StdDev	Median	Ratio	Alloc Ratio
Scalar	32	32.720 ns	0.8425 ns	1.2349 ns	32.698 ns	1.00	NA
Vector128	32	5.188 ns	0.2228 ns	0.3334 ns	5.134 ns	0.16	NA
Vector256	32	3.229 ns	0.0289 ns	0.0423 ns	3.234 ns	0.10	NA
Vector512	32	3.009 ns	0.1658 ns	0.2431 ns	3.133 ns	0.09	NA
Scalar	64	56.427 ns	0.3364 ns	0.5035 ns	56.424 ns	1.00	NA
Vector128	64	7.598 ns	0.0354 ns	0.0530 ns	7.599 ns	0.13	NA
Vector256	64	4.812 ns	0.0415 ns	0.0608 ns	4.807 ns	0.09	NA
Vector512	64	4.123 ns	0.1851 ns	0.2655 ns	4.134 ns	0.07	NA
Scalar	128	107.334 ns	0.4501 ns	0.6310 ns	107.531 ns	1.00	NA
Vector128	128	13.617 ns	0.0754 ns	0.1105 ns	13.625 ns	0.13	NA
Vector256	128	7.892 ns	0.1306 ns	0.1787 ns	7.840 ns	0.07	NA
Vector512	128	6.161 ns	0.1522 ns	0.2182 ns	6.157 ns	0.06	NA
Scalar	256	211.792 ns	0.6895 ns	1.0320 ns	211.823 ns	1.00	NA
Vector128	256	25.945 ns	0.0814 ns	0.1193 ns	25.958 ns	0.12	NA
Vector256	256	14.105 ns	0.0597 ns	0.0893 ns	14.090 ns	0.07	NA
Vector512	256	11.926 ns	0.0554 ns	0.0830 ns	11.913 ns	0.06	NA
Scalar	512	423.199 ns	2.1657 ns	3.2415 ns	422.230 ns	1.00	NA
Vector128	512	50.393 ns	0.4069 ns	0.5964 ns	50.362 ns	0.12	NA
Vector256	512	27.170 ns	0.1839 ns	0.2753 ns	27.170 ns	0.06	NA
Vector512	512	23.254 ns	0.3293 ns	0.4827 ns	22.974 ns	0.05	NA
Scalar	1024	837.884 ns	4.0460 ns	5.9305 ns	836.344 ns	1.00	NA
Vector128	1024	102.157 ns	0.3440 ns	0.5149 ns	102.196 ns	0.12	NA
Vector256	1024	50.981 ns	0.2298 ns	0.3368 ns	50.946 ns	0.06	NA
Vector512	1024	46.550 ns	0.0828 ns	0.1240 ns	46.521 ns	0.06	NA

[benaadams]

512 likely stronger on next version of Zen