Closed romits800 closed 3 years ago
Hi Romy, I had a look but the test case is too large for effective debugging. Could you try to reduce it, perhaps by disabling code-bloating LLVM IR opts or removing parts of the original C function? For the latter, I recommend you using a tool like C-Reduce.
Hi Roberto, I tried to reduce the size by reducing the size of the actual function: I attach the following files together with a script that runs them: test3.5.tar.gz : The largest example, where the presolver seems to work as expected test4.tar.gz: The smallest example that demonstrates the same behavior as described in the initial post.
Let me know if the size is still large.
Thanks! Solved, I think :) Please check that this does not cause any regression in your experiments.
Optimizing function "fe25519_mul" does not give any solution, with the presolver proving that it is infeasible. After running Unison:
The presolver returns:
I am attaching the .mir file, generated by:
ed25519_ref10_mir.tar.gz