[Prev][Next][Index][Thread]

RE: executor & windows95



At 02:31 AM 3/8/96 -0800, you wrote:
>
>
>----------
>From: 	Scott Stegura[SMTP:scottuf@grove.ufl.edu]
>Sent: 	Tuesday, March 05, 1996 12:54 PM
>To: 	Mat Hostetter
>Cc: 	executor@ardi.com
>Subject: 	Re: executor & windows95
>
>Mat Hostetter wrote:
>> 
>> >>>>> "Scott" == Scott Stegura <scottuf@grove.ufl.edu> writes:
>> 
>>     Scott> Well, as an experiment I tried running Executor under
>>     Scott> MS-DOS Mode and I was surprised because it ran almost twice
>>     Scott> as fast!!!  The video was a lot faster and the CPU jumped
>>     Scott> from a 6 to a 12 on my 486DX2-66.  Well, I guess from now
>>     Scott> on I'll run directly under DOS instead of Win95.  I'm
>>     Scott> really surprised at how much of a drain Win95 is on my
>>     Scott> system.
>> 
>> That's interesting.
>> 
>> Video should be faster (if you've got UniVBE) because we can directly
>> access the linear frame buffer under DOS.  Still, I wouldn't think
>> that the CPU number should slow down.
>> 
>> Can you try:
>> 
>> executor -nosound -oldtimer
>> 
>> and see how fast it is under Win95?  Thanks!
>> 
>> -Mat
>
>The -oldtimer option helped speed things up some.  The -nosound option 
>didn't help though.  And I am running UniVBE 5.1a.  However, things 
>still weren't up to DOS speed so I think I'll stick with the hassle of 
>exiting windows to go to DOS mode for now.  Hopefully V3.00 (or maybe 
>V2.1) will be fully Win95 compatable.
>
>Scott Stegura
>
>I have the same problem you do with speed in '95. Well sorta...
>Sometimes I can get it to run as fast as DOS when I use a simple command
line like:
>
>executor -nosplash -memory 4M
>
>This command started working as fast as DOS around the time I changed my
serial mouse to a PS/2 port mouse. I am not certain though.
>
>I noticed when running in a small amount of RAM if you continue to run
executor, exit, then reload the numbers get less and less. So when I test
different command lines like Matt suggests I always reboot.  
>
>I took the liberty to use the command switches -nosound -oldtimer to see
what I came up with.  Below are my findings they are sorta interesting.
>=====================================================
>Specs:
>
>i486DX4100, 8M ram, Cirrus Logic (CL-GD5428) 1MB  VBE 1.2
>Using executor 1.99q8 and Univbe 5.1a
>2.0 VBE extentions, Linear framebuffer loacated at 14Mb
>Speedometer 3.23
>Standard command line: Executor -nosplash -memory 4M
>							
>		CPU	Graph	Disk C:	Math	Bench avg	Color avg
>Dos*	11.456	  7.701	3.707	26.614	17.375		1.674
>-nosound	12.627	  8.670	3.945	27.647	19.489		1.859
>-oldtimer	13.206	  9.382	3.827	29.395	21.067		1.972
>both**	14.108	10.328	3.915	34.067	22.637		2.165
>
>Win95***	11.456	7.613	4.160	26.250	17.108		1.678
>-nosound	  8.357	5.231	5.587	13.815	10.858		1.187
>-oldtimer	11.768	7.657	6.865	26.468	17.199		1.687
>both**	13.106	8.902	3.698	29.102	19.711		1.907
>
>*Used Smartdrv: smartdrv a b- c d 2048 0
>**Executor -nosplash -memory 4M -nosound -oldtimer
>***Only Explorer was loaded with Executor
>==================================================
>For some reason Win95 does not like -nosound.  The really odd thing is when
both -nosound and -oldtimer are in use, it is faster then each one
individualy. Especially when -nosound is slower then executor running with
just -nosplash -memory 4M.
>
>Looking at these results gives me a question.  Why isn't -oldtimer part of
the default settings when it makes the emulation run much faster?? What are
the pluses and minus of using -oldtimer?
>
>Well anyways I hope this gets to all of you cuzz this took forever :)
>
>Enjoy,
>
>
>Glenn R. Keyser	

I think that using -oldtimer will give you a less acurate timer so when you do
this you can't trust spedomter to give you accurate resaults.  Also when you
run a hardware testing program like spedomter in a DOS box under a multitaker
(windows, OS/2, deskview) then the program will only get to run some of the 
time so the ratings that you get are very far from accurate.  Too see what I
mean get a copy of speed200.exe which is a simple speed tester that runs on DOS 
this program will repeatadly test the machine for it's speed and when it's run 
on strait dos then the speed stays at the correct speed but when run in a
DOS box 
under a multitaker (windows, OS/2, deskview) then the speed changes to much
faster
than what you have to much slower than what you have and goes back and forth.

Basicily I don't trust a speed tester (including spedomter) when it's run in 
multitasking enviroment.





 


Visit my home page at:   http://www.inf.net/~hawk
e-mail: hawk@inf.net



Follow-Ups: