OK, using your averaging scheme, on a 32-bit KVM kernel, running at
idle, here are my quick results:
Original
48028 51419
46021 50065
44750 49231
Fastmix 2, 3 interations
95956 58313
97295 57599
97242 56942
Fastmix 2, 2 iterations
68998 41496
68940 41471
68619 41576
Fastmix 2, 2 iterations, unrolled
48725 39281
48313 38424
47349 37847
So with two iterations are at least no worse than before (and in fact
the deviation is less, which makes sense since we don't have the
twist_array memory accesses), and I can easily believe there will be
architectures/microarchitectures where it will be better.
I'll need to do a bit more looking to convince myself that 2
iterations is better from a mixing perspective, but this looks like it
might be a promising replacement for the 32-bit mixer.
- Ted
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
