🚀 go-pugleaf

In article <Pine.3.87.9604121912.A6048-0100000@mbox1.ufsc.br>, 
Frank Hrebabetzky - CERTI       - CCE 2319561 <fh@mbox1.ufsc.br> writes:
> 
> On Fri, 12 Apr 1996, Doug Danforth wrote:
> 
>> Folks,
>>   Have you ever wondered how much time is consumed by a cycle
>> of an Oberon task loop?  I've run several small programs and
>> would like to report the following results for Windows 95 Oberon
>> V4:
>> 
>> One tick:  	1 millisecond.
>> TightLoop: 	10,000 cycles/tick ( 60 MHZ Pentium)
>>   "             30,000    "         150  "    "
>> Task:      	50 ticks
>> Others.Wait:	50 ticks
>> 
> 
> My measurements for System 3 / DOS on a 486DX, 33 MHz showed that the main
> event loop cycle frequency (corresponds to "Task" above) depends heavily on 
> the pointer position (message path):
> 
16> over a text viewer (not ~ document!):	1400 /s
> over the desktop background:		 480 /s
> over a single panel:			 220 /s
> over 3 overlapping panels:		 160 /s
> 

  here are the results on Linux 1.2.8, XFree 3.1.1., XServer Mach64,
Pentium 100 MHz, ETH Oberon V4-0.9. I tried several times:

minTicks always 5
aveTicks always 6    (well, one time was 7)

That is 6 millisecond per one cycle, 167 cycles per second.

Max ticks varied from 12 to 695, however, 695 happened just once when 
I was transferring a large file over the net at the same time.
When I just started / stopped the loop, maxTicks was between 
12 and 34. When I started the loop and shrunk Oberon to an icon,
maxticks went up to 300. Note however that average was always 6,
all the time. 

There was only one exception: when I started the loop and then 
clicked on a PopupElement (and held down the MM for a long while)
the average Ticks went up to 26.

Wojtek

In article <1996Apr13.131636.1@fermi.nsrl.rochester.edu>, 
skulski@fermi.nsrl.rochester.edu writes:
> In article <Pine.3.87.9604121912.A6048-0100000@mbox1.ufsc.br>, 
> Frank Hrebabetzky - CERTI       - CCE 2319561 <fh@mbox1.ufsc.br> writes:
>> 
>> On Fri, 12 Apr 1996, Doug Danforth wrote:
>> 
>>> Folks,
>>>   Have you ever wondered how much time is consumed by a cycle
>>> of an Oberon task loop?  I've run several small programs and
>>> would like to report the following results for Windows 95 Oberon
>>> V4:
>>> 
>>> One tick:  	1 millisecond.
>>> TightLoop: 	10,000 cycles/tick ( 60 MHZ Pentium)
>>>   "             30,000    "         150  "    "
>>> Task:      	50 ticks
>>> Others.Wait:	50 ticks
>>> 
>> 
>> My measurements for System 3 / DOS on a 486DX, 33 MHz showed that the main
>> event loop cycle frequency (corresponds to "Task" above) depends heavily on 
>> the pointer position (message path):
>> 
> 16> over a text viewer (not ~ document!):	1400 /s
>> over the desktop background:		 480 /s
>> over a single panel:			 220 /s
>> over 3 overlapping panels:		 160 /s
>> 
> 
>   here are the results on Linux 1.2.8, XFree 3.1.1., XServer Mach64,
> Pentium 100 MHz, ETH Oberon V4-0.9. I tried several times:
> 
> minTicks always 5
> aveTicks always 6    (well, one time was 7)
> 
> That is 6 millisecond per one cycle, 167 cycles per second.

This is not particularly impressive for a Pentium, is it? Here are some 
more details: 

when I start the loop, this is the CPU load as reported by "top" utility:
  PID USER     PRI  NI SIZE  RES SHRD STAT %CPU %MEM  TIME COMMAND
15056 root      17   0 5445 3076 1732 R    51.4 20.3  8:16 X :0
16033 skulski   18   0 4238  808 2692 S    46.3  5.3  5:09 /usr/local/Oberon/oberon
16100 skulski    8   0  102  316  336 R     1.7  2.0  0:01 top

These numbers do not seem to depend on whether Oberon runs full screen or 
as an icon. Seems half of the CPU is eaten by the X-server. Does this tell 
us something about the overhead incurred by running Oberon as an application
under another OS? Does this tell something about X-windows themselves?


Wojtek

In article <1996Apr13.133318.1@fermi.nsrl.rochester.edu>, 
skulski@fermi.nsrl.rochester.edu writes:
>> 
>>   here are the results on Linux 1.2.8, XFree 3.1.1., XServer Mach64,
>> Pentium 100 MHz, ETH Oberon V4-0.9. I tried several times:
>> 
>> minTicks always 5
>> aveTicks always 6    (well, one time was 7)
>> 
>> That is 6 millisecond per one cycle, 167 cycles per second.
> 

The above was under olvwm window manager. Under my favorite fvwm 
(Slackware default) minTicks=8 and aveTicks=9. That is, 30% speed 
difference by simply switching X-windows managers. 

Sorry for many postings. When switching managers I had to kill 
my remote sessions and thus lost connection to news.

Wojtek