Andreas Kloeckner <[email protected]> writes: > Tomasz Rybak <[email protected]> writes: >> I again got failure request trying to send email >> to the lists@ - so here I am sending it again. > > Can you forward one of these failures? > >> -------- Wiadomość przekazywana ---------- >>> I'd very much like to hear more test results from other GPUs for the >>> current code in git. >>> >> >> Loveland: >> There are still some errors is scan, but the one I was fighting >> does not appear on 13.1 drivers. I'll try test and play with it >> tomorrow. >> >> test_algorithm.py ................s.............F.F.F.....F...F. > > Those look somewhat similar to what I'm seeing on "Devastator" (a > Trinity/Northern Islands-based APU). Still not sure what's up there. > (btw--"Devastator"? "Loveland"? Who at AMD makes up these names? :)
Relatedly, I'm inclined to release fairly soon. I've recently set up a continuous integration server, and that tells me that, aside from these AMD APU hiccups, the code is in good shape across many (Linux) implementations and devices. (In particular, discrete AMD GPUs seem to be working fine.) Contributing to this is the fact that the current release apparently won't build any more on a number of systems--mine included. Probably due to some gcc changes. Andreas _______________________________________________ PyOpenCL mailing list [email protected] http://lists.tiker.net/listinfo/pyopencl
