Add fallback for ARM Windows fp16 detection.

[edgchen1]

Background: On a Windows ARM system, I observed that cpuinfo_has_arm_fp16_arith() started to return false after upgrading to a more recent cpuinfo version.

In #333, the initialization of cpuinfo_isa.fp16arith was updated to use IsProcessorFeaturePresent(PF_ARM_V82_FP16_INSTRUCTIONS_AVAILABLE). I suspect that this is not supported on older Windows versions.

This change adds a fallback path to set cpuinfo_isa.fp16arith the old way.

cpuinfo/src/arm/windows/init.c

Lines 205 to 208 in d3a86a8

	// Assume that Dot Product support implies FP16
	// arithmetics and RDM support. ARM manuals don't
	// guarantee that, but it holds in practice.
	cpuinfo_isa.fp16arith = dotprod;

[edgchen1]

Hi @tonybaloney, would you mind taking a look at this one?

[tonybaloney]

What chip is it?

[edgchen1]

What chip is it?

This was observed on an Azure Standard D16pds v5 machine with an Ampere Altra processor.

[fbarchard]

Although this is true in practice, if the detection for FP16 is available and used, and it says false, we should respect that?
There have been recent A75 without FP16. They also dont have dot product.
Can you clarify a case for when this is necessary?

[edgchen1]

Although this is true in practice, if the detection for FP16 is available and used, and it says false, we should respect that?

Yes, I agree that if detection is available, we should respect it. However, I am not sure that it is actually available. IsProcessorFeaturePresent() returns 0 both when the feature is actually not detected to be present and also when detection is not available.

Can you clarify a case for when this is necessary?

The behavior I observed was that, on a CI build system with an Ampere Altra processor, cpuinfo_has_arm_fp16_arith() used to return true (prior to the use of IsProcessorFeaturePresent(PF_ARM_V82_FP16_INSTRUCTIONS_AVAILABLE) in src/arm/windows/init.c) and then started to return false. On that same system, I was able to run a test program with NEON FP16 intrinsics successfully, so it appears that FP16 instructions are actually available.