asm.hpp: fix usage of _mm_prefetch

[oltolm]

I'd like to try again to fix usage of _mm_prefetch.

prefetch_exec has the following comment.

	// Try to prefetch to Level 2 cache since it's not split to data/code on most processors
	template <typename T>
	constexpr void prefetch_exec(T func)

I changed the argument from 2 to _MM_HINT_T1.

// MSVC
#define _MM_HINT_T1     2
// GCC/Clang
  _MM_HINT_T1 = 2,

As you can see there was no change. T1 is also correct:

T1 (temporal data with respect to first level cache misses)—prefetch data into level 2 cache and higher.

prefetch_read has the following comment

	// Try to prefetch to Level 1 cache
	constexpr void prefetch_read(const void* ptr)

I changed 3 to _MM_HINT_T0.

// MSVC
#define _MM_HINT_T0     1
// GCC/Clang
  _MM_HINT_T0 = 3,

As you can see there is no change on Clang/GCC, but there is a change on MSVC. I think that this is correct:

T0 (temporal data)—prefetch data into all levels of the cache hierarchy.

The way it is now it is prefetched into level 3 and higher on MSVC:

#define _MM_HINT_T2     3

T2 (temporal data with respect to second level cache misses)—prefetch data into level 3 cache and higher, or an implementation-specific choice.