math errors

Some time ago there was a post about math errors in a plc and in a pc. I was talking to a co-worker and he was saying about how he has seen -0 and 0 on whole math calculations and was curious about that. I told him about math errors and that I would try to find info about it but I can't dig up the posts or links.

Thanks
Drewcrew6

drewcrew6 said:

Some time ago there was a post about math errors in a plc and in a pc. I was talking to a co-worker and he was saying about how he has seen -0 and 0 on whole math calculations and was curious about that. I told him about math errors and that I would try to find info about it but I can't dig up the posts or links.

Thanks
Drewcrew6

Click to expand...

What do you mean by math errors?
Are you talking about processor errors, such as divide by zero or overflow?
Or are you talking about cases where the PLC computes the wrong value.

There is a convention for computer floating-point math that most PLCs follow. It is IEEE 754. There are a lot of rules and conventions for that standard and for the standard integer math (called two's complement) that are not intuitive to a human doing the same calculations.

Actual math errors by computer processors are few and far between (the pentium fdiv bug for example) and could be caused by faulty hardware.


There are cases where -0 and 0 are valid answers BTW(Think about a value that maintains its sign but is small enough that it is rounded down (or up) to zero.

What I mean by math errors is for example taking 420 minus 420 and you get a positive or negative zero.

Drewcrew6