Ok found the issue, don't know how to solve it yet.
Thanks for the bug report..
Guys can you please test (and bug report) the snapshots, with 2 years between releases we really
shouldn't have this kind severe bugs in actual the release.
(10-31-2013 01:40 PM)dgud Wrote: [ -> ]Ok found the issue, don't know how to solve it yet.
Thanks for the bug report..
Guys can you please test (and bug report) the snapshots, with 2 years between releases we really
shouldn't have this kind severe bugs in actual the release.
I will do that. I wasn't sure Wings3d is stil being developed since the last new version was released in 2011, so I kept using 1.4.1.
I'd love to help with coding, but I'm afraid I don't know Erlang.
Thank you for all the work you've done and for looking into this problem!
(11-02-2013 11:16 PM)dgud Wrote: [ -> ]Can you test if this version helps?
https://www.dropbox.com/sh/54h7afpm5hvx45m/hcay8uZhgc
I can still crash Wings3d. Opencl was set both to 0 and 1 and I was able to crash it across multiple fresh starts of the app.
Perhaps I could obtain more information somehow? It's rather weird because no dump gets created, it's like it segfaults and just dies. Should I try to run it on a Windows install with Visual Studio installed to use a debugger to get more details than the error screen?
Dan, I'm not sure this report can help in some way, but I tried produce the crash reported and I noticed that it happened after Smooth computation has been done and when starting the Soft process (maybe about 10%) - in accord with Information line.
The PC I used have 4GB, Intel HD Graphics (i5), Windows 7 64Bits.
I needed smooth the initial cube 8 times and then call smooth (393,216 polys):
In the first stage, wer.erl took ~2,100,000MB of memory until smooth was done, falling to ~1,400,000 with a small increase until I got the crash.
It was generated a
erl_crash.dump in the Wings3D.exe directory (I do not use the
"Programs Files").
The header refers the crash as memory allocation problem:
Quote:=erl_crash_dump:0.2
Sun Nov 03 01:23:12 2013
Slogan: eheap_alloc: Cannot allocate 1898305688 bytes of memory (of type "heap").
System version: Erlang R16B02 (erts-5.10.3) [64-bit] [smp:4:4] [async-threads:10]
Compiled: Mon Sep 16 19:28:19 2013
basically the same problem we had talk about in the
past in the old forum.
Cannot allocate means that Windows do not want to give erlang more memory, not much I can do about
that, but that is strange since it's the 64b version.
The snapshot is much more stable for me..
Still crashing here.
Faulting module:- nvoglv64.dll (nVidia openGL driver)
The faulting module is now consistent, but when wings actually crashes is unpredictable.
ok guys you got to give me something more to work on here,
smoothing a cube almost always crashed after 7 times now it doesn't, until I ran out of memory.
What do you do, and in what display mode?
(11-03-2013 08:21 PM)dgud Wrote: [ -> ]ok guys you got to give me something more to work on here,
smoothing a cube almost always crashed after 7 times now it doesn't, until I ran out of memory.
What do you do, and in what display mode?
I just smooth a cube back and forth until Wings3d crashes. Let's say that subdivision level 0 is the original cube; I go from 0 to 5-7 (~96000 polygons) and then undo back to 0.
I do this sometimes to figure out what the smoothed objects looks like and then go back to the original object I'm working on. I don't need that many subdivisions when I'm actually using Wings3d, but this was useful for reproducing the bug.
(11-03-2013 10:15 PM)quadf Wrote: [ -> ]I do this sometimes to figure out what the smoothed objects looks like and then go back to the original object I'm working on.
by using the opencl option instead of truely smooth the object should work better for this case.
Just to know, how many memory your system has?
With the last snapshot I could smooth 9x until call smooth preview and crash ere.exe. But, I started with only 1.4GB used by other process and then it tried to break my 4GB limit and stoped to response when erl.exe was using 2.7GB.