home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
EnigmA Amiga Run 1998 July
/
EnigmA AMIGA RUN 29 (1998)(G.R. Edizioni)(IT)[!][issue 1998-07 & 08].iso
/
recent
/
iib122.lha
/
IIB
/
Threads
/
IFA_config
< prev
next >
Wrap
Internet Message Format
|
1998-02-10
|
7KB
Date: Tue, 3 Feb 1998 13:04:56 -0400
From: Scott Sutherland <ssutherl@GAMMAMETRICS.COM>
Subject: [IML] AMIGA: ACUP differences in .config (5.0 to 5.1)
Hello All:
Well, to further determine why my IM 5.1 is 2.5 X SLOWER than my IM 5.0 on
the test project, I looked at the .config files for each and tried to
isolate functions that might affect speed. Here are the settings:
PARAMETER IM50 IM51 DESCR
----------- ------- ------- -------------------------------
OLDA T F use "old" anti-aliasing in trace
USAA T T use AA chip set if available
S256 T T run 640x480 imagine in 256 color
RTGS F F show real-time grey scale image...
EDLE 30 10 Anti-aliasing threshhold
AARC 255 20 Anti-aliasing rays per pixel limit
RSDP 8 8 "Resolve Depth" (multiple reflection)
OCTD 6 6 # levels in oct-tree (max)
SMOD 0 0 imagine screen width (0=640)
OTRL -1 -1 max oct-tree RAM in KBytes (or -1)
OTFL 40 40 min free RAM after oct-tree (Kb) (or -1)
SSLE 45 45 Soft Shadow Light Elements
USCG F T flag-imagine uses cybergraphics.library
graphics routines
These are the major items of note. Basically, as you can see, the major
differences
that stand out are the EDLE value (is a higher number more or less
anti-aliasing?
I forget), the anti-aliasing rays per pixel limit (does this have any
effect on
trace times?), the use of OLD anti-aliasing in my IM50.config, and NOT using
the cybergraphics.library graphics routines in IM50.
I will change these tonight to see what happens. Any comments/suggestions in
the mean time would be helpful.
Scott Sutherland
----------------------------------------
From: "Ash R. J. Wyllie" <ashw@LR.NET>
>EDLE 30 10 Anti-aliasing threshhold
That is the line that makes all the difference. raise the threshold and you
will get a huge increase in speed. EDLE determines when more rays should be
tried in order to do antialiasing.
I have noticed a significant speed increase in quickrenders with my 040/40.
-ash
(Where is John Galt when you need him?)
----------------------------------------
From: Nick Curcio <nik2@FEIST.COM>
Using old AA in your Im5.0 will make the rendering drastically faster. The
higher EDLE value (less AA) makes it faster, too. The number of rays/pixel has
no effect if you are using the old AA. So, using old AA and the higher EDLE
value will make your Im5.0 run a lot faster...as you have seen for yourself. ;)
nik
----------------------------------------
From: Eric van Velthooven <eric@STACK.NL>
> PARAMETER IM50 IM51 DESCR
> ----------- ------- ------- -------------------------------
> OLDA T F use "old" anti-aliasing in trace
From: Clae & Deborah Tanett <cd@ACCUTEK.COM>
in trace
I'll bet you kazillions, this is the setting responsible...
old AA seems to be a lot faster than the fancy new AA they
out into v5.x of imagine.
> EDLE 30 10 Anti-aliasing threshhold
> AARC 255 20 Anti-aliasing rays per pixel limit
From: Clae & Deborah Tanett <cd@ACCUTEK.COM>
pixel limit
I think these settings influence new AA
Eric
----------------------------------------
From: Clae & Deborah Tanett <cd@ACCUTEK.COM>
Hi Scott, Eric
Clae here
This talk of the OLDA and how it sends down the rendering times made me
relook at things in mine. I have the OLDA set to T (I just used my old
Imagine.config - didn't want to mess around redoing the colors in the
prefs... not to mention adding all my custom buttons). Back to track... I
heard you discussing the OLDA slowing down the rendering times sooo I set
mine to F in the OLDA... the 060 blasted through it. Just as fast with F as
T. (my system is an A4000/Mk II 060 w 80megs ram:)
Just a note.
Martin also asked if I saw a big rendering difference... "no."
Hey... what can I say... "That's the 060."
The object I rendered was a 19,200 figure (no textures - in setup stages
yet), screen: 640x400 24-bit; 30 seconds from go to palette color/clean up
end.
Clae F Tanett
cd@accutek.com
----------------------------------------
From: Scott Sutherland <ssutherl@GAMMAMETRICS.COM>
Hello All:
Well, I answered my own question when I had to go home for lunch today
to get my briefcase. I tested the effects of the various differences
between my IFA 5.0 Config file and the default 5.1 Config file. I post
the results here for general interest and to pose a question.
PARAMETER IM50 IM51 DESCR
----------- ------- ------- -------------------------------
OLDA T F use "old" anti-aliasing in trace
EDLE 30 10 Anti-aliasing threshhold
AARC 255 20 Anti-aliasing rays per pixel limit
USCG F T flag-imagine uses cybergraphics.library
graphics routines
Here are the results:
My original 5.0 rendering time was 60 seconds (frame 13).
My original 5.1 rendering time was 153 seconds.
The following changes to the 5.1 config resulted in the following rendering
time
changes:
Changing EDLE from 10 to 30 reduced the rendering time to 77 seconds.
Changing from the new AA routine to the OLD one reduced the rendering time
to 53
seconds.
Changing AARC and USCG had no effect on the rendering time. So, final
results:
IM 5.0: 60 sec.
IM 5.1: 53 sec.
A similar change to the other posted times.
I was NOT surprised at the effect of changing EDLE, but I was very surprised
at the magnitude of the effect of changing from the NEW AA routine to the
old one (~ 30% reduction in rendering time).
BTW, I have created my own project (objects, etc.) and ~ the same settings
for
a comparison of IFA and IFW. My original 3000 '030 with IFA 5.0
was 40X slower than my IFW on a P120 using this project. Now the difference
is ~ 1.5X slower with my PPC(200 MHz)/060(50 MHz)/60 ns RAM/IFA 5.1. Quite
a difference. Can't wait for the PPC IFA!
Thanks.
Scott Sutherland