| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
It is not (yet) a fatal error if floattest fails, but you will be
warned, and told which test fails to run.
|
|
|
|
|
|
|
|
|
|
|
| |
Testing equality of floating point numbers (f == g or f != g) is
usually Considered Harmful because floats and doubles are only
approximations. Granted, some numbers can be represented precisely -
which is what I use in this test suite. All intermediate results can
be precisely represented as well. The same tests are used for both
single and double precision numbers.
Tested on i386, amd64 and sparc64. ARM tests to follow shortly.
|
| |
|
| |
|
| |
|
|
...a simple utility to check that your toolchain can do single and
double precision math properly. I've heard trouble reports from two
embedded user where the problem was traced to a buggy toolchain.
|