BINARY INTEGER SQUARE ROOTS

[jj2007]

One more, using FastMath

Code Select Expand

282 ms for the FPU
608 ms for FastMath
X      Y fpu          Y FastMath
990.0  31.46427        31.46427
991.0  31.48015        31.48015
992.0  31.49603        31.49603
993.0  31.51190        31.51190
994.0  31.52777        31.52777
995.0  31.54362        31.54362
996.0  31.55947        31.55947
997.0  31.57531        31.57531
998.0  31.59114        31.59114
999.0  31.60696        31.60696
1000.0  31.62278        31.62278
The fpu is pretty fast
Twice as fast as the FastMath version, and a factor 33 faster than Ognil/Lingo's version.

[ognil]

NoCforMe,

I don't know what you're taking to calm your nerves, but maybe they need to increase your dosage because your nervous depression doesn't allow you to differentiate between hardware (something like NVMe SSD) and software (something like XMM).

JJ's "speed tests" are always manipulated depending on his current emotional state and are practically completely useless and unnecessary to anyone.

I'm not Lingo and I've already said that I don't care about optimizing code speed.

I suggest you stop here and go for a walk in nature for a while.

[jj2007]

JJ's "speed tests" are always manipulated

Prove it, or I'll call you a liar

Your code is hopelessly slow, but it's a nice intellectual exercise, so I liked it. And I don't care if it's AI or genuine ;-)

[NoCforMe]

your nervous depression doesn't allow you to differentiate between hardware (something like NVMe SSD) and software (something like XMM)

What???

That's provably, laughably false: XMM refers to a set of 128-bit registers (therefore nothing like software. (And technically they fall under the heading of "SSE".)

Your credibility here is slipping fast, man.

[ognil]

NoCforMe,

I agree with you that you, jj and zed are the greatest programmers here and everyone else on the forum (including me) should take your advice as law and bow down to the elegance and speed of your code.

Now please calm down and take a walk.