Accelerate additional cross platform hardware intrinsics#61649

Merged

tannergooding merged 12 commits intodotnet:mainfrom tannergooding:xplat-hwintrin

Jan 4, 2022

+3,188-866

Commits on Jan 3, 2022

Updating Vector64/128/256.IsHardwareAccelerated to be treated as a constant and return true where supported
tannergooding
committed
Accelerate the CmpOpAll intrinsics
tannergooding
committed
Accelerate the CmpOpAny intrinsics
tannergooding
committed
Accelerate the ConverToDouble/Int32/Int64/Single/UInt32/UInt64 intrinsics
tannergooding
committed
Applying formatting patch
tannergooding
committed
Fixing ConvertToInt32 and ConvertToSingle to use the right intrinsic
tannergooding
committed
Fixing some issues and assert types are correct
tannergooding
committed
Updating ConvertToDouble and ConvertToSingle to have correct vectorized versions on x86/x64
tannergooding
committed
Ensure Vector<T>.ConvertToDouble/Single are accelerated
tannergooding
committed
Swap operands and invert immediate so the constant can be contained on blend
tannergooding
committed
Restrict ConvertToDouble(Vector128<UInt64>) tests to inputs no more than long.MaxValue
tannergooding
committed
Ensure that we create a long/ulong rather than a uint
tannergooding
committed