This is the Windows7 Ultimate Speedtest DOSBOX SVN ; AMD AB-5500 APU Quadcore 3.2GHz 4.00GB RAM ; Core= Full (Dynamic crashes) Cycles=Fixed 225687 : Time to generate the Patterns Image: 120 seconds ( acc Fractint info screen) ; Doesn´t need a comment ; Petta
Turns out it is ridiculously easy to run Fractint under pure DOS if your PC hardware lets you. I'm sure I knew this before but rediscovered it just now. It took me a whole 10 minutes to create an MSDOS-booting USB stick and run Fractint. The idea is to support the benchamarking effort. 1. Get yourself a spare USB stick. 2. Download Rufus (https://rufus.akeo.ie/). Does not need installation, just run, and turn your USB stick into a bootable drive. I installed MSDOS rather than Freedos 3. Copy Fractint distribution to usb stick. 4. Boot the stick. I did this by pressing <del> to get to my bios setup, and found special booting options under the "Save" menu on my AMD Bios. Your mileage will vary depending on your Bios. And how well it works will depend on VESA support of your video board. Vesainfo said my video had a 1600x1200x8 mode, but it didn't work. The 1280x1024 mode did. The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds. It's been a while since I installed DOSBox, and I'm not sure if it is set to run as fast as possible. Now to try the same thing on Iterated Dynamics and Manpwin on the same machine. After the holidays. Much more could be done. I'm sure it is possible to add more VESA video modes via software and to create a fractint.cfg.
I wrote:
The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds.
Yikes! I spoke way too soon. In the dosbox-SVN-Daum.conf file I changed "cycles=auto" to "cycles=max" and now the default Mandelbrot at 1280x1024 executes in 4.4 seconds, faster than under naked MSDOS on the same machine. I'd be curious if others get the same result. Not sure how DOSBox can be faster than DOS, but apparently it is, and by a large margin. Also, at the bottom of the *.conf file I have the following lines (which assume c:\fractint is where the fractint distribution is, alter to suit): [autoexec] # Lines in this section will be run at startup. # You can put your MOUNT lines here. mount c c:\fractint c: fractint On Mon, Dec 15, 2014 at 2:48 PM, Timothy Wegner <tim@tswegner.net> wrote:
Turns out it is ridiculously easy to run Fractint under pure DOS if your PC hardware lets you. I'm sure I knew this before but rediscovered it just now. It took me a whole 10 minutes to create an MSDOS-booting USB stick and run Fractint. The idea is to support the benchamarking effort.
1. Get yourself a spare USB stick. 2. Download Rufus (https://rufus.akeo.ie/). Does not need installation, just run, and turn your USB stick into a bootable drive. I installed MSDOS rather than Freedos 3. Copy Fractint distribution to usb stick. 4. Boot the stick. I did this by pressing <del> to get to my bios setup, and found special booting options under the "Save" menu on my AMD Bios. Your mileage will vary depending on your Bios. And how well it works will depend on VESA support of your video board.
Vesainfo said my video had a 1600x1200x8 mode, but it didn't work. The 1280x1024 mode did. The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds. It's been a while since I installed DOSBox, and I'm not sure if it is set to run as fast as possible.
Now to try the same thing on Iterated Dynamics and Manpwin on the same machine. After the holidays.
Much more could be done. I'm sure it is possible to add more VESA video modes via software and to create a fractint.cfg.
Tim you might want to take a look at this suite of benchmark in software for dos https://dosbenchmark.wordpress.com/ -David On 12/15/2014 12:19 PM, Timothy Wegner wrote:
I wrote:
The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds.
Yikes! I spoke way too soon. In the dosbox-SVN-Daum.conf file I changed "cycles=auto" to "cycles=max" and now the default Mandelbrot at 1280x1024 executes in 4.4 seconds, faster than under naked MSDOS on the same machine. I'd be curious if others get the same result. Not sure how DOSBox can be faster than DOS, but apparently it is, and by a large margin.
Also, at the bottom of the *.conf file I have the following lines (which assume c:\fractint is where the fractint distribution is, alter to suit):
[autoexec] # Lines in this section will be run at startup. # You can put your MOUNT lines here. mount c c:\fractint c: fractint
On Mon, Dec 15, 2014 at 2:48 PM, Timothy Wegner <tim@tswegner.net <mailto:tim@tswegner.net>> wrote:
Turns out it is ridiculously easy to run Fractint under pure DOS if your PC hardware lets you. I'm sure I knew this before but rediscovered it just now. It took me a whole 10 minutes to create an MSDOS-booting USB stick and run Fractint. The idea is to support the benchamarking effort.
1. Get yourself a spare USB stick. 2. Download Rufus (https://rufus.akeo.ie/). Does not need installation, just run, and turn your USB stick into a bootable drive. I installed MSDOS rather than Freedos 3. Copy Fractint distribution to usb stick. 4. Boot the stick. I did this by pressing <del> to get to my bios setup, and found special booting options under the "Save" menu on my AMD Bios. Your mileage will vary depending on your Bios. And how well it works will depend on VESA support of your video board.
Vesainfo said my video had a 1600x1200x8 mode, but it didn't work. The 1280x1024 mode did. The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds. It's been a while since I installed DOSBox, and I'm not sure if it is set to run as fast as possible.
Now to try the same thing on Iterated Dynamics and Manpwin on the same machine. After the holidays.
Much more could be done. I'm sure it is possible to add more VESA video modes via software and to create a fractint.cfg.
_______________________________________________ Fractint mailing list Fractint@mailman.xmission.com https://mailman.xmission.com/cgi-bin/mailman/listinfo/fractint
-- -David David W Riccio (907) 780-6122 www.lemoncreekdigital.com (Check Website office hours, location, etc)
David wrote: Tim you might want to take a look at this suite of benchmark in software
Well it would be interesting to see if those benchmarks also show DOSBox is faster than native DOS. However the main interest in this context is how the speed of various fractal generators on various platforms and in various versions compare. For that the answer appears to be exactly what Rich Thomson said. The default Mandelbrot on both Iteratedynamics (a.k.a FractInt for Windows) and Manpwin is essentially instantaneous, way under a second on my machine. So a more ambitious fractal that takes much longer will be needed for a reasonable test, and I am sure the native Windows versions will end up being much faster exactly as Rich said. Iteratedynamics is a very faithful and certainly capable port of Fractint. The functionality is virtually identical, or at least very close, to the DOS Fractint. Mouse control of the zoom box has not been implemented, but the zoom box can be controlled with keystrokes, and of course video is done differently. Still, the result is amazing. The code base has been extensively refactored (I have no doubt very much for the better), so it is no longer possible (or at least not very easy) to capture Jonathan's changes to DOS/Linux versions subsequent to Rich's starting point. Mabe there are clues in the code as to which DOS version it is based on. Rich, do you know? The program announces itself as 20.99.8, but that is not any DOS version.
Point taken This is the generator that I am using most of the time ... a lot of Fractint formulas have been imported (but it can't directly read a .par file) --- Gnofract 4D http://gnofract4d.sourceforge.net/ -David On 12/15/2014 01:27 PM, Timothy Wegner wrote:
David wrote:
Tim you might want to take a look at this suite of benchmark in software for dos https://dosbenchmark.wordpress.com/
Well it would be interesting to see if those benchmarks also show DOSBox is faster than native DOS. However the main interest in this context is how the speed of various fractal generators on various platforms and in various versions compare. For that the answer appears to be exactly what Rich Thomson said. The default Mandelbrot on both Iteratedynamics (a.k.a FractInt for Windows) and Manpwin is essentially instantaneous, way under a second on my machine. So a more ambitious fractal that takes much longer will be needed for a reasonable test, and I am sure the native Windows versions will end up being much faster exactly as Rich said.
Iteratedynamics is a very faithful and certainly capable port of Fractint. The functionality is virtually identical, or at least very close, to the DOS Fractint. Mouse control of the zoom box has not been implemented, but the zoom box can be controlled with keystrokes, and of course video is done differently. Still, the result is amazing. The code base has been extensively refactored (I have no doubt very much for the better), so it is no longer possible (or at least not very easy) to capture Jonathan's changes to DOS/Linux versions subsequent to Rich's starting point. Mabe there are clues in the code as to which DOS version it is based on. Rich, do you know? The program announces itself as 20.99.8, but that is not any DOS version.
_______________________________________________ Fractint mailing list Fractint@mailman.xmission.com https://mailman.xmission.com/cgi-bin/mailman/listinfo/fractint
-- -David David W Riccio (907) 780-6122 www.lemoncreekdigital.com (Check Website office hours, location, etc)
Am 15.12.2014 23:27, schrieb Timothy Wegner:
teratedynamics is a very faithful and certainly capable port of Fractint. The functionality is virtually identical, or at least very close, to the DOS Fractint. Mouse control of the zoom box has not been implemented, but the zoom box can be controlled with keystrokes, and of course video is done differently. Still, the result is amazing. The code base has been extensively refactored (I have no doubt very much for the better), so it is no longer possible (or at least not very easy) to capture Jonathan's changes to DOS/Linux versions subsequent to Rich's starting point. Mabe there are clues in the code as to which DOS version it is based on. Rich, do you know? The program announces itself as 20.99.8, but that is not any DOS version.
Please read Paul´s statement about Iterated dynamics: / //The ported modified version of FractInt, which was really never finished for the Windows environment, does not work well with many Microsoft OS computers. I could never get it to work on Windows Vista, let alone Win-7. I just tell people to not even bother trying to use it, and they would be better off using Paul de Leeuw's ManpWIN program. // //// //// //Sincerely, // //P.N.L. // /This is my impression too. MFR_13 does not run - Buffer overrun and possible other issues./ /Richard promised to have a look into this. Waiting for a result./ ///Petta
Petta wrote (after quoting PNL): This is my impression too. MFR_13 does not run - Buffer overrun and
possible other issues.
I defer to you and the other folks here who spend serious hours with fractal explorations and advanced use of Fractint on how how well other software does or does not match Fractint. I have personally never used Fractint as extensively as you folks have. My use of Fractint over the years has consisted of debugging, testing new features, and regression testing, mostly a long time ago. So I don't have a clear picture of what differences crept in (or didn't) to Iteratedynamics. Re some comments proposing different DOSBox settings. Yes, cycles=fixed 205170 and core=dynamic seem to help. I also realized I was trusting the execution time that shows on the tab screen. Better to run a slow fractal and time with a stopwatch. Of course under DOSBox the times Fractint reports can't be trusted.
If I had to guess on why DOSBox is faster than DOS on the same machine I would guess that the DOSBox knows more about the graphics system than DOS does and can emulate the video better than DOS can utilize the native graphics hardware ... I would guess that the graphics card in the machine would make a lot of difference. I came across someone that has generated code that lets you program the GPU directly on modern Graphic cards (move your code to run on that processor) One of the example was creating a Mand. Set on the graphics processor ... see: https://developer.nvidia.com/how-to-cuda-python -David On 12/15/2014 12:19 PM, Timothy Wegner wrote:
I wrote:
The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds.
Yikes! I spoke way too soon. In the dosbox-SVN-Daum.conf file I changed "cycles=auto" to "cycles=max" and now the default Mandelbrot at 1280x1024 executes in 4.4 seconds, faster than under naked MSDOS on the same machine. I'd be curious if others get the same result. Not sure how DOSBox can be faster than DOS, but apparently it is, and by a large margin.
Also, at the bottom of the *.conf file I have the following lines (which assume c:\fractint is where the fractint distribution is, alter to suit):
[autoexec] # Lines in this section will be run at startup. # You can put your MOUNT lines here. mount c c:\fractint c: fractint
On Mon, Dec 15, 2014 at 2:48 PM, Timothy Wegner <tim@tswegner.net <mailto:tim@tswegner.net>> wrote:
Turns out it is ridiculously easy to run Fractint under pure DOS if your PC hardware lets you. I'm sure I knew this before but rediscovered it just now. It took me a whole 10 minutes to create an MSDOS-booting USB stick and run Fractint. The idea is to support the benchamarking effort.
1. Get yourself a spare USB stick. 2. Download Rufus (https://rufus.akeo.ie/). Does not need installation, just run, and turn your USB stick into a bootable drive. I installed MSDOS rather than Freedos 3. Copy Fractint distribution to usb stick. 4. Boot the stick. I did this by pressing <del> to get to my bios setup, and found special booting options under the "Save" menu on my AMD Bios. Your mileage will vary depending on your Bios. And how well it works will depend on VESA support of your video board.
Vesainfo said my video had a 1600x1200x8 mode, but it didn't work. The 1280x1024 mode did. The default mandelbrot took 10.4 seconds at 1280x1024. Under Windows 8.1 and DOSBox, the same fractal was 40 seconds. It's been a while since I installed DOSBox, and I'm not sure if it is set to run as fast as possible.
Now to try the same thing on Iterated Dynamics and Manpwin on the same machine. After the holidays.
Much more could be done. I'm sure it is possible to add more VESA video modes via software and to create a fractint.cfg.
_______________________________________________ Fractint mailing list Fractint@mailman.xmission.com https://mailman.xmission.com/cgi-bin/mailman/listinfo/fractint
-- -David David W Riccio (907) 780-6122 www.lemoncreekdigital.com (Check Website office hours, location, etc)
Am 15.12.2014 22:57, schrieb David W Riccio:
If I had to guess on why DOSBox is faster than DOS on the same machine I would guess that the DOSBox knows more about the graphics system than DOS does and can emulate the video better than DOS can utilize the native graphics hardware ... I would guess that the graphics card in the machine would make a lot of difference. Read my messga Windowsi Patterns Speedtest qte Cycles=fixed 205170 core=dynamic The Patterns image that yesterday in the "cycles=max" setting needed about 14 minutes, is now finished in 2 minutes 15 seconds, unqte the default mandel needs 7 seconds at 1280/1024 The cycles value has been found during calculation and watching the CPU performance on the sidebar cpu meter. If core is set to dynamic 2 cores are used above a certain speed. The XP dos is still much faster the Patterns image is generated between 65 and 90 seconds, pending other processes ; Petta
Am 15.12.2014 21:48, schrieb Timothy Wegner:
2. Download Rufus (https://rufus.akeo.ie/). Does not need installation, just run, and turn your USB stick into a bootable drive. I installed MSDOS rather than Freedos Can´t get fractint to work properly as the keyboard layout (northern european) is wrong. The TILDE and the BACKSLASH keys are missing. They can be called using the ASCII Code (126 and 92) but still fractint.cfg is not read. Any idea how to get the correct keyboard layout? Thanks for any assistance Petta
participants (3)
-
David W Riccio -
Multifrac＠t-online.de -
Timothy Wegner