hi,

William F Pokorny <ano### [at] anonymousorg> wrote:
> On 5/17/20 3:38 AM, jr wrote:
> > fyi, the scene code crashes ("segfault") alpha.10064268 on one machine,
> > but not on another; ...
>
> IIRC, the build scripts end up with -march=native by default. Are you
> offering us the compilers 'native' translation or did you specify those
> architectures specifically on the two machines?

explicit spec.  Slackware package build scripts have a variable used to
set/append CFLAGS etc, I used the 'haswell' in both '-march' and '-mtune'.


regards, jr.

Post a reply to this message

On 5/17/20 4:43 AM, jr wrote:
> hi,
> 
> William F Pokorny <ano### [at] anonymousorg> wrote:
>> On 5/17/20 3:38 AM, jr wrote:
>>> fyi, the scene code crashes ("segfault") alpha.10064268 on one machine,
>>> but not on another; ...
>>
>> IIRC, the build scripts end up with -march=native by default. Are you
>> offering us the compilers 'native' translation or did you specify those
>> architectures specifically on the two machines?
> 
> explicit spec.  Slackware package build scripts have a variable used to
> set/append CFLAGS etc, I used the 'haswell' in both '-march' and '-mtune'.
> 

OK. This leads me to wonder what more, or less, these spawn than 
'native' given the compilers should do the right things with native for 
an optimized compile on the local machine.

In short, wondering if the hard coded architecture the problem?

Bill P.

Post a reply to this message

hi,

William F Pokorny <ano### [at] anonymousorg> wrote:
> ...
> In short, wondering if the hard coded architecture the problem?

very likely, although, as often, it's unclear whether "the fault" lies with the
compiler or the source.  guessing that some code (in spline, or text object) is
vulnerable wrt optimisation(s).  I will rebuild the alpha with 'core2' options,
sometime in the next few days; I expect the scene code will work then.


regards, jr.

Post a reply to this message

High!

On 16.05.20 00:47, jr wrote:
> hi,
> 
> weekend lockdown "advice".  :-)   (loop)

Kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...

...waiting for the bus to Kabul!

See you in Khyberspace!

Yadgar

Post a reply to this message

Hi(gh)!

On 17.05.20 09:38, jr wrote:
> hi,
> 
> William F Pokorny <ano### [at] anonymousorg> wrote:
>> On 5/15/20 6:47 PM, jr wrote:
>>> weekend lockdown "advice".  :-)   (loop)
>>>
>> Cool. :-)
> 
> thanks.  fyi, the scene code crashes ("segfault") alpha.10064268 on one machine,
> but not on another; the only difference being, afaict, the optimisation flags
> provided in the package build scripts.  using "-march=core2" works (gcc 4.8.2),
> using "-march=haswell" crashes (gcc 5.5.0).  :-)

POVing under the influence? I'm not so sure...

See you in Khyberspace!

Yadgar

Post a reply to this message

hi,

Yadgar wrote:
> High!
>
> On 16.05.20 00:47, jr wrote:
> > hi,
> >
> > weekend lockdown "advice".  :-)   (loop)
>
> Kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...
>
> ...waiting for the bus to Kabul!

:-)

in the late 70s, in West-Berlin, there was (still) talk of just such a service.
Berlin to Kabul, five or six weeks overland by bus.  (guessing the cost was
little more than the petrol money)


regards, jr.

Post a reply to this message

Op 18/05/2020 om 21:49 schreef jr:
> hi,
> 
> Yadgar wrote:
>> High!
>>
>> On 16.05.20 00:47, jr wrote:
>>> hi,
>>>
>>> weekend lockdown "advice".  :-)   (loop)
>>
>> Kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...kiff...
>>
>> ...waiting for the bus to Kabul!
> 
> :-)
> 
> in the late 70s, in West-Berlin, there was (still) talk of just such a service.
> Berlin to Kabul, five or six weeks overland by bus.  (guessing the cost was
> little more than the petrol money)
> 
> 
> regards, jr.
> 


https://en.wikipedia.org/wiki/Hippie_trail ;-)

-- 
Thomas

Post a reply to this message

hi,

Thomas de Groot <tho### [at] degrootorg> wrote:
> Op 18/05/2020 om 21:49 schreef jr:
> > Yadgar wrote:
> >> ...waiting for the bus to Kabul!
> > ...
> > Berlin to Kabul, five or six weeks overland by bus.  (guessing the cost was
> > little more than the petrol money)
>
> https://en.wikipedia.org/wiki/Hippie_trail ;-)

thanks, had not thought of/remembered that phrase.  (imo, Wikipedia is one of
very few examples of the interweb not having gone wrong)


regards, jr.

Post a reply to this message

Hi(gh)!

On 19.05.20 08:27, Thomas de Groot wrote:

 >
 > https://en.wikipedia.org/wiki/Hippie_trail ;-)
 >

It all started in 1957 with the Indiaman buses - see 
http://www.indiaman.101answers.com/ - even before Afghanistan officially 
opened its borders for tourists, which was only in 1959! Interestingly, 
otherwise than the "classic" hippie trail, the (initial) route of the 
Indiaman, left out Kurdistan, but instead fully ran though the Near East 
including Jerusalem (which after the Six Days War of 1967 probably was 
no longer feasible)... and they entered Afghanistan not, like 60s/70s 
hippies mostly did, via Islam Qal'a to Herat, but from Quetta via Spin 
Boldak to Kandahar! Back then, the Kandahar-Kabul road was not even 
metalled...

And what has this all to do with raytracing and 3D modeling? POVEarth, 
as it develops in the coming decades, surely will also provide a solid 
geographic background for the virtual rebirth of the hippie trail! Even 
more so as I plan to model the whole thing also in X3D... this would be 
real-time interactive!

See you in Khyberspace!

Yadgar

Post a reply to this message

> And what has this all to do with raytracing and 3D modeling? POVEarth,
> as it develops in the coming decades, surely will also provide a solid
> geographic background for the virtual rebirth of the hippie trail! Even
> more so as I plan to model the whole thing also in X3D... this would be
> real-time interactive!


[Unity] Procedural Planets (E01 the sphere)
https://www.youtube.com/watch?v=QN39W020LqU

Post a reply to this message