>>>Me: Somebody is really confused here. But then, the good part:
However, one ISV used Java to run complex mathematical equations for calculating the value of options in the currency market and was shocked at the results.
Cats Software, a Palo Alto, Calif.-based risk-management software vendor, found that Java performed only 4 percent to 10 percent slower than Visual C++ when running Partial Differential Equations.
"These results are staggering," said Jim Kleckner, chief technology officer at Cats. "I would have expected Java to perform about 100 percent slower than Visual C++."
Hmm, Reg better inform these guys on the fraudulence of Java math. On the other hand, if they know PDE's they probably know a little numerical analysis too, and might not be totally impressed with Reggie's analysis.
Some of those CATS guys used to work with me in financial product development at BofA. We had some connections there because CATS was one of our suppliers. So clearly, they would have to be more people with no understanding of "high end finacial math", along with my poor benighted self, in Reg's view, and completely beneath him.
Seriously, it's like I kept trying to tell Reg: 'It's the floating point processor, not the language, stupid.' There's a limit on how much the software procedure calling mechanism could be expected to slow down the hardware-based calculation engine under any reasonably intelligent programming scenario.
I must admit I found it alarming that somebody thought Java would be "100% slower." Now that would be a blow. :-|
Cheers,
Chaz |
