14:44 <gdaiss[m]> hkaiser: do you know when the hpx hipcc tests last worked/succeeded? with my PR, the build itself works again and the tests hence run again -- half of them fail with timeout errors though. Was this an issue before as well?

14:55 <hkaiser> mdiers[m]: could you try this patch, please: https://gist.github.com/hkaiser/8e3001377921456c383866c073e75bd3

14:56 <mdiers[m]> hkaiser: oh you are fast, yes I am testing

15:18 <mdiers[m]> <hkaiser> "mdiers: could you try this patch..." <- The first try with the changes didn't bring any improvement.

15:18 <mdiers[m]> I meant earlier also a yield() instead of the wait_any(), is probably already too late for me.

15:18 <mdiers[m]> With the yield() it ran through now, but then got stuck at the end when cleaning up.

17:31 <hkaiser> mdiers[m]: there is another change for wait_any, I will create a PR we can use to play with things

17:52 <mdiers[m]> <hkaiser> "mdiers: there is another..." <- ok, then let me know when you have created it

18:17 <HHN93> I am getting an illegal instruction error when running a HPX program

18:18 <HHN93> Is it an generic error or does it point to a specific issue?

18:31 <srinivasyadav18[> may be if the code is compiled in one machine (architecture) and executed on different

18:36 <HHN93> ok will try compiling again

20:28 <HHN93> does something go horribly wrong if you call a hpx algorithm from within the lib?

20:28 <HHN93> something like an illegal instruction error?

20:29 <srinivasyadav18[> what do you mean by calling hpx algorithm within lib ? can you elaborate it please ?

20:30 <HHN93> my hpx::fold_left function calls hpx::reduce

20:32 <srinivasyadav18[> illegal instruction error usually happen when the generated executable file is not run/executed on the same machine architecture that it has been compiled on (or not same dependecies loaded)

20:34 <HHN93> I changed the hpx::reduce to my implementation of reduce and it works

20:34 <HHN93> I compiled it the same way

20:34 <srinivasyadav18[> thats strange

20:35 <srinivasyadav18[> can you share more information about the steps that you have performed (log file containing steps and outputs)

20:36 <HHN93> I don't think I have observed a hpx algorithm call another hpx algo either, they always call std:: versions of them

20:36 <HHN93> `can you share more information about the steps that you have performed (log file containing steps and outputs)`

20:36 <HHN93> sure, will try

20:55 <srinivasyadav18[> <HHN93> "I don't think I have observed..." <- There are many places where one hpx algorithm calls other hpx algo, there is nothing wrong in that I belive. Its just another function-like call

22:57 <gnikunj[m]> HHN93: Illegal instruction error is not HPX specific error. When building with —march=native and running on a different architecture, you may observe such errors. This is because compilers use architecture optimized instructions which may not be available on another architectural.

22:59 <gnikunj[m]> For HPX calling into internal HPX algo implementations, go into GDB and step through functions. It’s a nice exercise ;)