zhongyijie
发贴: 36
积分: 30
|
于 2003-07-18 10:43
Description In the Java 2 platform and later, floating point calculations can now be done using the IEEE 754 extended-precision formats. These formats permit the exponent portion of intermediate results to be wider than the 32- or 64-bit size of the end result of the operation. In the original Java virtual machine specification, floating point operations were required to truncate intermediate results of calculations, which is now called strict mode. The strictfp keyword, added in the Java 2 SDK, forces floating point calculations in methods to operate in strict mode. All the methods in a class declared with the strictfp keyword will operate in strict mode.
The default mode on Windows for methods that do not use the strictfp keyword is to use the wider exponent on the operand stack. The wider exponent improves the performance of floating point operations. A wider exponent would not help performance on Solaris, so the default mode is the same as strict mode.
In most situations, default mode will give you the same result as strict mode. The result of a floating point operation in default mode will differ from the strict mode result only if the calculation overflows or underflows in strict mode in places it doesn't in default mode. Since most floating point calculations occur nowhere near underflow or overflow, default mode will most often yield the same result as strict mode.
Source: The Java Language Specification. Copyright (C) 1996 Sun Microsystems, Inc.
清楚了吗?
|