FLiT
FLiT copied to clipboard
PRUNERS
A project to quickly detect discrepancies in floating point computation across hardware, compilers, libraries and software.
Now that there is a testing framework, only a meager amount of the framework is actually tested. I would like to instill a code coverage metric. If gcc is used,...
Some of the python scripts have doctest unit tests inside, which is great. However, we should probably run then using `python3 -m doctest ` when we call `make check` on...
As seen in this blog post: https://plindenbaum.blogspot.com/2014/09/parallelizing-gnu-make-4-in-slurm.html for GNU Make version 4.0 and later, we can actually execute all Makefile recipes as slurm jobs. This is pretty cool. We could...
This is a followup to issue #91. But instead of having that issue remain for so long, it has been split up so that #91 can be closed. Bisect functionality...
The FLiT test framework has been updated, but the inputGen tool has been left behind. It needs to be updated with capabilities to handle the different return types (i.e. `long...
Even if the test framework does not directly support CUDA anymore, there are code bases that at least need to be compiled with the `nvcc` compiler, if they are used...
After finishing #163, we will also want the wrapping functionality to be part of `flit bisect`. I am imagining a flag called `--nvcc-wrap=` where `` is either an absolute path,...
The current, and I think desired default behavior of `flit bisect` is to look for files and functions that have **any** influence on variability according to the comparison function. However,...
It has been pointed out to me that python can be run in an optimization mode where asserts are skipped. The FLiT code base uses asserts for error checking and...
I just read the original paper introducing Delta Debugging. It is a very good read and surprising that it is not more well known. It can be coerced into doing...