Add ARM version of adler32

[brianpopow]

Prerequisites

• I have written a descriptive pull-request title
• I have verified that there are no overlapping pull-requests open
• I have verified that I am following the existing coding patterns and practice as demonstrated in the repository. These follow strict Stylecop rules 👮.
• I have provided test coverage for my change (where applicable)

Description

I was experimenting a bit with ARM hardware intrinsics and decided to port the adler32 version of chromium to ImageSharp.

BenchmarkDotNet=v0.13.1, OS=ubuntu 20.04
Unknown processor
.NET SDK=5.0.402
[Host] : .NET 5.0.11 (5.0.1121.47308), Arm64 RyuJIT
DefaultJob : .NET 5.0.11 (5.0.1121.47308), Arm64 RyuJIT

Method	Mean	Error	StdDev
AdlerARM	1.082 us	0.0123 us	0.0109 us
AdlerScalar	3.823 us	0.0165 us	0.0146 us

CPU info:

Architecture:        aarch64
CPU op-mode(s):      32-bit, 64-bit
Byte Order:          Little Endian
CPU(s):              6
On-line CPU(s) list: 0-5
Thread(s) per core:  1
Core(s) per socket:  3
Socket(s):           2
Vendor ID:           ARM
Model:               4
Model name:          Cortex-A53
Stepping:            r0p4
CPU max MHz:         1896.0000
CPU min MHz:         100.0000
BogoMIPS:            48.00
Flags:               fp asimd evtstrm aes pmull sha1 sha2 crc32

Benchmark was run with the following data:

int N = 4096;
byte[] data = new byte[N];
new Random(1).NextBytes(data);

There is still a bug, adler results for N 4096: fixed

Arm: 1010560139
Scalar: 1010560140

[gfoidl]

For .NET 7 cross-platform intrinsics will be available, so that code can be written on behalf of Vector128<T> and not via Sse2 / AdvSimd.

I know that it's still a long time until is .NET 7 is available and adopted, but is this something that should be focused on now already?

[brianpopow]

For .NET 7 cross-platform intrinsics will be available, so that code can be written on behalf of Vector128<T> and not via Sse2 / AdvSimd.

I know that it's still a long time until is .NET 7 is available and adopted, but is this something that should be focused on now already?

That's interesting, I was not aware of this. How would that work? SSE has intrinsics which ARM does not have and vice versa.

I think its to early to early to consider this now. Anything ARM related is now only pretty much experimental and will not land in the main branch anytime soon.

[gfoidl]

How would that work? SSE has intrinsics which ARM does not have and vice versa

Only the "common subset" is exposes as xplat-intrinsics. The journey started in dotnet/runtime#49397 but more operations got added so far -- I can't find a up-to-date summary of all operations 😢.
For the special cases one needs to fall back to the current intrinsics style (Sse.IsSupported / AdvSimd.IsSupported).

I think its to early to early to consider this now

This is also my expectation. Just wanted to point to it (as a reminder that something is on the way).

[brianpopow]

This is also my expectation. Just wanted to point to it (as a reminder that something is on the way).

Thanks for sharing, that's very interesting to know.