mirror of
https://github.com/yuzu-emu/yuzu-emu.github.io.git
synced 2025-06-29 22:34:17 +00:00
Fix innacuracy about NaN values. (#258)
* Fix innacuracy about NaN values. There was an error in the description of cases that yield NaN values. The erroneous data was scrapped. * fixed minor error
This commit is contained in:
parent
28cafc54ed
commit
3d5ff36494
|
@ -172,7 +172,7 @@ On any digital system, all numbers are represented in a binary base in order to
|
|||
Naturally, there is a limit to the amount of numbers and the precision with which they can be represented, not to mention that a number can be expressed in many ways by just declaring it as a different type (e.g. integer, [floating-point](https://en.wikipedia.org/wiki/Floating-point_arithmetic) number, etc.).
|
||||
Likewise, there are also many mathematical elements that aren't numbers per se, but special cases that result from mathematical operations.
|
||||
Think, for example, about the square root of a negative number: although the result of this operation is a valid [complex number](https://en.wikipedia.org/wiki/Complex_number), it's not a defined type (i.e. a number that the computer understands), therefore, it becomes *something else*.
|
||||
There are many other operations which yield similar results: division by zero, multiplying by infinity, or even more exotic things like dividing zero by zero, zero by infinity, or multiplying zero by infinity.
|
||||
Or think about dividing any number by zero: this mathematical operation is undefined, and thus it yields a similar result.
|
||||
|
||||
These types of indeterminations are defined as a special floating-point value called [`NaN`](https://en.wikipedia.org/wiki/NaN) - which stands for "Not a Number".
|
||||
The architecture of the CPU of the Nintendo Switch (`ARM`) handles these NaN values differently from the architecture used by any computer CPU (`AMD64`).
|
||||
|
|
Loading…
Reference in a new issue