 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Pov thinks that a scene I have been rendering for around 30 days has actually
been rendering for about 12 million years.
Ian
Post a reply to this message
Attachments:
Download 'too long.png' (2 KB)
Preview of image 'too long.png'
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Le 10/03/2014 02:32, [GDS|Entropy] a écrit :
> Pov thinks that a scene I have been rendering for around 30 days has actually
> been rendering for about 12 million years.
>
> Ian
>
Could it be tied to the 49+ days rollover on windows ?
Is that 30 days, or something like 25+ days ?
For the technical minded: windows keeps the notion of time in a 32 bits
integer, as number of milliseconds. (actually they call it a DWORD, but
the API remains the same).
After 49.7 days, that counter rollover. (and that used to crash W95 &
W98... there was a patch for that, but the counter is still the same
size: GetTickCount() cannot change)
And after half that time, it might be interpreted as a negative value.
Of course, promoting the DWORD into a 64 bits value is to be interesting
if the receiver is signed.
2^62 is about 146 megayears, 11 megayears would be between 2^58 and
2^59... maybe some x1000 is applied to get a number of microseconds
before converting back to years.
--
Just because nobody complains does not mean all parachutes are perfect.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
It is between 25 and 30 when the issue manifested, as I remember it correctly
reporting a value of 28 days prior to this. :)
Thanks!
Ian
Le_Forgeron <lef### [at] free fr> wrote:
> Le 10/03/2014 02:32, [GDS|Entropy] a écrit :
> > Pov thinks that a scene I have been rendering for around 30 days has actually
> > been rendering for about 12 million years.
> >
> > Ian
> >
>
> Could it be tied to the 49+ days rollover on windows ?
>
> Is that 30 days, or something like 25+ days ?
>
>
> For the technical minded: windows keeps the notion of time in a 32 bits
> integer, as number of milliseconds. (actually they call it a DWORD, but
> the API remains the same).
>
> After 49.7 days, that counter rollover. (and that used to crash W95 &
> W98... there was a patch for that, but the counter is still the same
> size: GetTickCount() cannot change)
>
> And after half that time, it might be interpreted as a negative value.
>
> Of course, promoting the DWORD into a 64 bits value is to be interesting
> if the receiver is signed.
>
> 2^62 is about 146 megayears, 11 megayears would be between 2^58 and
> 2^59... maybe some x1000 is applied to get a number of microseconds
> before converting back to years.
>
> --
> Just because nobody complains does not mean all parachutes are perfect.
Post a reply to this message
|
|
| |
| |
|
|
From: Alain
Subject: Re: Render Reported Too long...[11.7 megayears]
Date: 10 Mar 2014 23:09:45
Message: <531e7e79@news.povray.org>
|
|
|
| |
| |
|
|
> Pov thinks that a scene I have been rendering for around 30 days has actually
> been rendering for about 12 million years.
>
> Ian
>
This has been reported some time back, maybe around 2 months ago. It
affect both 32 and 64 bits Windows versions. I don't know for the Mac or
the Linux versions.
It's a known isue that only rarely shows and is been investigated. You
can expect it to be corrected in version 3.7.1...
Alain
Post a reply to this message
|
|
| |
| |
|
|
From: Nicolás Alvarez
Subject: Re: Render Reported Too long...[11.7 megayears]
Date: 11 Mar 2014 10:10:07
Message: <531f193f$1@news.povray.org>
|
|
|
| |
| |
|
|
>> Pov thinks that a scene I have been rendering for around 30 days has
>> actually
>> been rendering for about 12 million years.
>
> This has been reported some time back, maybe around 2 months ago. It
> affect both 32 and 64 bits Windows versions. I don't know for the Mac or
> the Linux versions.
The Linux version doesn't have a GUI to show this bug in :)
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
