Floating-Point Calculations May Be Incorrect on Alpha and MIPS

• Microsoft Windows NT operating system version 3.1
• Microsoft Windows NT Advanced Server version 3.1

SYMPTOMS

When you are using Calculator under Windows NT on the DEC Alpha computer or other RISC-based computer, you may experience floating-point errors in your calculations.

This problem does not occur on x86-based computers.

STATUS

Microsoft has confirmed this to be a problem in Windows NT version 3.1 and Windows NT Advanced Server version 3.1. We are researching this problem and will post new information here in the Microsoft Knowledge Base as it becomes available.

MORE INFORMATION

Steps to Reproduce Problem

1. In the Accessories group, open Calculator, choose Scientific from the View menu.

2. Find the sin of 90 (degrees in decimal mode); the result is 1 (correct).

3. Change to octal, binary, hexadecimal, or back to decimal mode; the results are all 0 (incorrect).

4. Find the sin of 90 again, then take the inv sin of the 1; the result is 89.99999914623--it should be 90.

5. Find the cos of 200 (gradients in decimal mode); the result is -1 (correct).

6. Take the inv cos of the -1; the result is 199.9999990514--it should be 200.