Skip to content

[Test / Bug Fix] Fixed minimum iteration count in "tests/bench/bfloatfunc.pp" when in conformance mode

J. Gareth "Kit" Moretonrequested to merge
CuriousKit/optimisations:floatfunc-fix into main

Summary

This merge request increases the iteration count in tests/bench/bfloatfunc.pp from 1 to 8 when running conformance tests in order to prevent false failures from tests not being run (due to the inputs being selected via the iteration counter).

What is the current bug behavior?

tests/bench/bfloatfunc.pp fails when built from the tests/test/cg/tfloatfunc.pp conformance test.

What is the behavior after applying this patch?

tests/bench/bfloatfunc.pp will now pass when built from the tests/test/cg/tfloatfunc.pp conformance test (but give nonsensical durations).

Relevant logs and/or screenshots

Before (all results from index 1 onwards on each test are uninitialised):

...
                  Min (single-precision) - FAIL - index 1; expected  1.048576000E+06 got  0.000000000E+00
                  Max (single-precision) - FAIL - index 1; expected  1.048577000E+06 got  0.000000000E+00
         Implicit Min (single-precision) - FAIL - index 1; expected  1.048576000E+06 got  0.000000000E+00
         Implicit Max (single-precision) - FAIL - index 1; expected  1.048577000E+06 got  0.000000000E+00
...

After:

...
                  Min (single-precision) - Pass - average iteration duration: 0.000 ns
                  Max (single-precision) - Pass - average iteration duration: 0.000 ns
         Implicit Min (single-precision) - Pass - average iteration duration: 0.000 ns
         Implicit Max (single-precision) - Pass - average iteration duration: 0.000 ns```
...

Merge request reports