Joe Rowell

Results 32 comments of Joe Rowell

Your input matrix has quite large entries. Version 4.0.4 was a bit before my time, so I'm not sure how it handled precision, but I suspect this is the problem....

OK my intuition was really wrong. The input lattice: - Terminates with ```fplll -a bkz -b 20 -f mpfr -p 53 < matrix_input.txt``` (so with the same precision as double)...

Hi Manos, When I originally looked into this issue at the last FpLLL days I ended up chasing my own tail a bit... I think the problem is to do...

Does this not assume access to the strategizer in fplll? I don't think fplll ships with those bits (i.e set_mdc.py is probably necessary, although I guess we could fix 64...

Hi Manos, I'm not requested to review this, but I have a few comments. None of these comments are on correctness (I haven't looked that deeply), but I noticed a...

I think the gist of it (c.f [this](https://gcc.gnu.org/pipermail/gcc-bugs/2019-December/683885.html)) is that its an old warning for ABI compatibility. I think you can safely ignore it unless you're passing those templated types...

I didn't actually try OPENBLAS_NUM_THREADS=2: I would guess that this would be fine, provided that I only use 18 threads otherwise (so that the total "in use" core count sums...

Happy to purge the Sagemath stuff from here: the suggested idea seems much better (mea culpa!) I can't replicate the bug on my Debian box, but I'll trust from the...

(Just to make sure I understand the state of this issue) is your argument: - that the macros could be placed elsewhere during building, or - that we shouldn't have...