Charles Greathouse on Thu, 25 Jan 2024 00:02:44 +0100


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: make dobench failing recently for me


Rebuilding directly failed the bench again with these differences:
analyz-dyn.dif
+ [10, 127]

polyser-dyn.dif
+ [61, 4888]

trans-dyn.dif
! 541266542498557839619175749637426460610039831
! 541266542498557839619175749637426460610039830

I'm retuning and will report back on those results.

On Wed, Jan 24, 2024 at 2:56 PM Bill Allombert <Bill.Allombert@math.u-bordeaux.fr> wrote:
On Mon, Jan 22, 2024 at 01:04:12PM -0500, Charles Greathouse wrote:
> I'm not sure if this is a bug or local configuration issues, so I'll post
> here for now. While I'm able to build gp from the repo, for a while (weeks,
> at least) I haven't been able to pass the bench. The diffs show a
> precision/rounding issue in trans-dyn and heap/memory leak issues in
> analyz-dyn, number-dyn, and polyser-dyn.
>
> The diffs are below. I can include more information if desired; this is for
> the latest git version from a few days ago, 3056f0fd062, but I've had
> similar issues on all recent releases. I'm running Ubuntu on WSL 2 (which
> hasn't had troubles in the past). I can reproduce on at least one other
> machine from a clean build.

If this comes from tune, hopefully this is fixed in
8e4a7a355fa39bc991f8a730c429415a8b03ff03

Cheers,
Bill.