>>Hoe to find if you have bugged Pentium chip ...
>>start the CALCULATOR-Program
>>try this: 4195835 / 3145727 * 3145727
>>The result should be = 4195835 (simple)
>>If you have a *ed-up Pentium chip the result will be: 4195579 !!!
>>(A slight difference of just 256 ... [who cares =)])
>That may just be a bug in Calculator, as I don't think it uses the FPU.
>(It works on a 386SX, so I'm pretty sure it doesn't, anyway.) Calculator
>is full of bugs : 3.11 - 3.1 = 0, for example.
What was posted above *does* sound like the Pentium FDIV bug. Perhaps the
Calculator detects an FPU and uses it if present? Heck, perhaps the Windows
API does this.
The secon bug (the 3.11-3.1=0 bug) is actually caused by a bug in the
Windows API. Something about it returning the wrong number of significant
digits or something. There's a work around for it too. The calculator that
comes with Win'95 uses such a work-around.
Incidently, has anyone heard anything about the other Pentium bug? The
multi-threading problem when using a write-back cache?
Carl Laurence Gonsalves - Computer Science, University of Waterloo
"I think not," said Rene Descartes, and vanished.