FLiT icon indicating copy to clipboard operation
FLiT copied to clipboard

A project to quickly detect discrepancies in floating point computation across hardware, compilers, libraries and software.

Results 71 FLiT issues
Sort by recently updated
recently updated
newest added

For failure cases in the `runFlitTests()` function, the created output file is not deleted afterwards if the run fails to finish. This can be fixed with an RAII class (e.g....

bug
good first issue
c++

The documentation could use a flow chart of steps to 1. install 2. run tests 3. analyze results This is for documentation and ease of use.

enhancement
question
good first issue
documentation

Looking at using Docker to have images for 1. All components 2. Host DB component 3. Client run component 4. runall component The Client component could potentially have multiple versions...

enhancement
documentation

Currently, you can return a single string or floating-point result at the end of each test case. This is analogous to a unit testing framework that only allows a single...

enhancement
question
c++
documentation

Currently it is supported to add user source files under test to be compiled with the tests. However, not all build systems are simple enough to do that easily (e.g...

enhancement
make
documentation

In the paper submission, there is a timing plot that shows the reproducible points and is sorted by execution time. This functionality should be part of the repository, and hopefully...

bug
enhancement
python
documentation

Since changing the database structure, we want to introduce shmoo plots again. This time as part of the `flit analyze` command.

bug
documentation

While it is true that the ground truth information is stored in the `flit-config.toml` file, it could be that past runs were under a different ground truth. It may be...

bug
python
make
documentation

We cancel out the default optimization level by explicitly specifying the optimization level, since not all compilers use the same default optimization level (intel defaults to `-O2` whereas GCC and...

question
documentation

The -xHost flag seems like it should be set for intel compilations, where it takes advantage of the hardware where it is compiled. Should we do this for all compilations...

enhancement
question
python
make
documentation