You may need to refresh the page. https://github.com/scala/scala3/issues/21637
Here is the comparison of the standard while loop with the vectorised version.
And the function left in vexct over time (against regressions)
The two implementations are;
extension (vec: Array[Double])
inline def sum_loop =
var sum: Double = 0.0
var i: Int = 0
while i < vec.length do
sum = sum + vec(i)
i = i + 1
end while
sum
end sum3
inline def sum_vec: Double =
var i: Int = 0
var acc = DoubleVector.zero(Matrix.doubleSpecies)
val sp = Matrix.doubleSpecies
val l = sp.length()
while i < sp.loopBound(vec.length) do
acc = acc.add(DoubleVector.fromArray(Matrix.doubleSpecies, vec, i))
i += l
end while
var temp = acc.reduceLanes(VectorOperators.ADD)
// var temp = 0.0
while i < vec.length do
temp += vec(i)
i += 1
end while
temp
end sum
Conclusion
For small array sizes (n = 3) the implementatiosn are the same. This would be expected, as the arrays are too small to vectorise. The fact the means are so close, gives us some confidence, that the measurements are realistic.
There does appear to be significant advantage to the vectorised version (2-4x) depending on array size (100 - 100,000). I'm surprised as I was expecting the JVM to almost perfectly optimised on the while loop. Perhaps, it is not able to easily auto-vectorise scala. I would welcome opinon.