R
Roedy Green
I had a look at the code Jet generated for both the winning piotr
algorithm and my own lohi32. In both cases, Jet generated absolutely
perfect code. I compiled the piotr code into 4 CPU cycles and lohi32
into 3. You might wonder why piotr beat lohi32 so handily. I explain
that in the program listing posted at
http://mindprod.com/jgloss/benchmark.html
Hint. CPUs have drastically speeded up over the years, but RAM has
not. It is matter of hurry up and wait. Its all in the timing...
algorithm and my own lohi32. In both cases, Jet generated absolutely
perfect code. I compiled the piotr code into 4 CPU cycles and lohi32
into 3. You might wonder why piotr beat lohi32 so handily. I explain
that in the program listing posted at
http://mindprod.com/jgloss/benchmark.html
Hint. CPUs have drastically speeded up over the years, but RAM has
not. It is matter of hurry up and wait. Its all in the timing...