Reply 40 of 95, by kruwi
So I was able to compile at least ... .
Applying these changes to the source code did not make any difference for me. Still unstable (though not with cvs-2007.1.17) !
Visit the end of the internet: www.groskreutz.de
So I was able to compile at least ... .
Applying these changes to the source code did not make any difference for me. Still unstable (though not with cvs-2007.1.17) !
Visit the end of the internet: www.groskreutz.de
Ok, keep that define outcommented for the moment.
In core_dyn_x86.cpp, there's a block after the includes, please replace
that one with the following:
#define CACHE_MAXSIZE (4096*2)
#define CACHE_PAGES (128*8)
#define CACHE_TOTAL (CACHE_PAGES*4096)
#define CACHE_BLOCKS (64*1024)
#define CACHE_ALIGN (16)
#define DYN_HASH_SHIFT (4)
#define DYN_PAGE_HASH (4096>>DYN_HASH_SHIFT)
#define DYN_LINKS (16)
This may not help, but I also run Windows 3.1 through DOSBox and use ykhwong's unofficial builds. His January 5 build and previous ones work fine with Win3.1. His latest January 21 build crashes Win3.1 upon exit. It's not that stable either. I'm back to Jan 5.
Robert
With these changes applied, windows finally works for me again.
But I do not find it very comfortable to manually change these lines every time ... .
Are there really no other users experiencing problems?
Visit the end of the internet: www.groskreutz.de
> but I do not find it very comfortable to manually change these lines every time ... .
Well maybe first you could try to find out what's working and what not.
Please get the latest cvs, and ONLY change the following line
#define CACHE_PAGES (512)
to some higher value (2048 instead of 512), if that works, try to lower
it again etc.
If it doesn't work even at high values, please see if raising the
CACHE_MAXSIZE value helps (4096*8 or so).
I will try out your suggestions, but I have to leave now ... .
I will not be able to post again until later in the evening
Visit the end of the internet: www.groskreutz.de
This may not help, but I also run Windows 3.1 through DOSBox and use ykhwong's unofficial builds. His January 5 build and previous ones work fine with Win3.1. His latest January 21 build crashes Win3.1 upon exit. It's not that stable either. I'm back to Jan 5.
As I said before, current and straight CVS works fine for me, no stability issues and no crash on exit. Keep in mind that ykhwong has many patches in his built and thus it may not be the most stable one at times.
But I do not find it very comfortable to manually change these lines every time
That's why they have you jump through loops, to narrow the problem down and find the reason and possible fix for this problem.
I am back home, finally ... .
I have downloaded the latest sources altered the line:
#define CACHE_PAGES (512)
to:
#define CACHE_PAGES (2048)
The instability remains though .
I'll report back later, applying the other changes suggested by wd.
Visit the end of the internet: www.groskreutz.de
I have just applied the second change to the sources, as has been suggested by wd.
The lines now look like this:
#define CACHE_MAXSIZE (4096*8)
#define CACHE_TOTAL (1024*1024*8)
#define CACHE_PAGES (2048)
#define CACHE_BLOCKS (64*1024)
#define CACHE_ALIGN (16)
#define DYN_HASH_SHIFT (4)
#define DYN_PAGE_HASH (4096>>DYN_HASH_SHIFT)
#define DYN_LINKS (16)
But: These changes did not solve the stability issues. The problems remain.
Visit the end of the internet: www.groskreutz.de
I think i found some bug that could cause this, it's no new/introduced bug
but seems to occur more often with the reduced number of code pages.
It should not occur if the CACHE_PAGES is large enough, whatever that
means in your case.
I can't reproduce the problem itself with windows, and don't have any
nice fix for this yet though.
I have tried to change the CASHE_PAGES to 4096 and finally to 8192 (!!!!). It did not change anything.
Visit the end of the internet: www.groskreutz.de
Additionally increasing MAX-CACHESIZE to 4096*8 does not help either.
Visit the end of the internet: www.groskreutz.de
For wd:
These are the results of my testing:
--------------------------------------------------
#define CACHE_TOTAL (1024*1024*4) in combination with:
#define CACHE_PAGES (1024): OK
#define CACHE_PAGES (512): unstable
-----------------------------------------------
#define CACHE_TOTAL (1024*1024*2) in combination with:
#define CACHE_PAGES (1024): OK
#define CACHE_PAGES (512): unstable
-------------------------------------------------
#define CACHE_TOTAL (1024*1024*1) in combination with:
#define CACHE_PAGES (1024): OK
#define CACHE_PAGES (512): OK
#define CACHE_PAGES (256): OK
#define CACHE_PAGES (128): OK
#define CACHE_PAGES (2048): OK
----------------------------------------------------
#define CACHE_TOTAL (1024*1024*8 ) in combination with:
#define CACHE_PAGES (1024): TOTALLY unstable
#define CACHE_PAGES (512): TOTALLY unstable
Visit the end of the internet: www.groskreutz.de
The latest CVS .66RC1 seems to work fine for me now. My installation of Windows 3.1 hasn't crashed yet, and the speed improvement in games like Duke3D are wonderful! I can now run it at 800x600 VESA with acceptable speed.
Robert
glad to hear that.
Water flows down the stream
How to ask questions the smart way!
Not working for me yet, though ...
I still have to adjust the source code in core_dyn_x86.cpp (to the values I have posted above:
#define CACHE_TOTAL (1024*1024*4) in combination with: […]
#define CACHE_TOTAL (1024*1024*4) in combination with:
#define CACHE_PAGES (1024): OK
#define CACHE_PAGES (512): unstable-----------------------------------------------
#define CACHE_TOTAL (1024*1024*2) in combination with:#define CACHE_PAGES (1024): OK
#define CACHE_PAGES (512): unstable-------------------------------------------------
#define CACHE_TOTAL (1024*1024*1) in combination with:#define CACHE_PAGES (1024): OK
#define CACHE_PAGES (512): OK
#define CACHE_PAGES (256): OK
#define CACHE_PAGES (128): OK
#define CACHE_PAGES (2048): OK----------------------------------------------------
#define CACHE_TOTAL (1024*1024*8 ) in combination with:#define CACHE_PAGES (1024): TOTALLY unstable
#define CACHE_PAGES (512): TOTALLY unstable
Visit the end of the internet: www.groskreutz.de
Another shot in the dark:
is Windows unstable in all possible color modes?
I'm asking because I tried the "CPU load reduction driver for Windows 3.x" that valnar posted about at CPU load reduction for Windows 3.1
The driver worked more or less for me but only when I switched from 1024x768 at 64k colors to 256 colors, my Windows became unstable. Apps crashing and such. Now I wondered if that is only because of the driver or the driver is just showing me that unstability that kruwi talks about.
FWIW, I only run in 256 colors within DOSBox and that CPU reduction driver works fine. I have not tried 64K.
I have also never tried out to use more than 256 colours for win 3.x in dosbox. I actually use a resolution of 1024x768x256 and sometimes also 800x600x256. Never tried out 64k modes, sorry.
Visit the end of the internet: www.groskreutz.de
Any ideas yet what could cause the problem?
Visit the end of the internet: www.groskreutz.de