Light ROM 4/Light ROM 4 - Disc 1.iso/text/maillist/1996/012296.doc/[e-mail] _Wed Jan 31 14:58:06 1996.msg

Internet Message Format | 1996-02-04 | 5KB

Received: from mail.webcom.com (mail.webcom.com [206.2.192.68]) by keeper.albany.net (8.7.1/8.7.1) with ESMTP id OAA27625 for <[e-mail]>; Wed, 31 Jan 1996 14:58:04 -0500 (EST)
Received: from localhost by mail.webcom.com with SMTP
    (1.37.109.15/16.2) id AA259838383; Wed, 31 Jan 1996 11:59:43 -0800
Date: Wed, 31 Jan 1996 11:59:43 -0800
Errors-To: [e-mail]
Message-Id: <[e-mail]>
Errors-To: [e-mail]
Reply-To: [e-mail]
Originator: [e-mail]
Sender: [e-mail]
Precedence: bulk
From: Joe <[e-mail]>
To: [e-mail]
Subject: Re: LWAMIGA4.0 is a pig
X-Listprocessor-Version: 6.0c -- ListProcessor by Anastasios Kotsikonas
Status: O
X-Status:

 > >What is the claim here -- that LW Amiga renders slower than 3.5?  What
 > >are the benchmarks and under what conditions do you see a slowdown?
 > >Raytracing, shadows, texture mapping, ?  I don't know of any SAS related
 > >problems with 4.0.
 > >    - Stuart Ferguson

 I have no benchmarks for raytracing, or shadows.  My test frames are just
 from the project I am currently working on.  Txture mapped polygones and
 picture background with med AA.

 First of all I find that my A2000/G-Force/040/33 with 16 meg 32bit
 FastRam is considerably slower rendering than a friends A4000 040 and his
 4.0 slowdown is in the regeon of 30% on my test frames while my slowdown
 is in the 200% range.  That doesn't help me but was interesting.

 Hey, I could spend a lot of money on an overpriced 4000 to only render
 30% slower.  Wrong.  Spending about the same or a bit more I can get an
 Intel setup and render 8 times faster!!

 My frames can vary in the amount of slowdown by a few factors I have
 tested.  One is the percentage of textured polys to picture mapped
 pbackground in the frame.  In 3.5, backgrounds are very fast so a frame
 with mosly background was fast. In 4.0 the background draws incredably
 slow, slower than a poly that is texture mapped and many that have
 procedurals and bumpmaps!. So in 4.0, those background "heavy" frames
 will show the most slowdown in comparison.

 Another factor that effects the benchmarks I have done is the AA threshold.
 At first I was using 8 which is way to much for the medAA I am using more
 and more now,Tthe benchmarks for the frames with 8 threshold and med
 was about 230% slowdown on my machine!  we didn't try that on the 4000.
 Raising the threshold to 20 which is what I finally decided to use for
 this job, brings the benchmarks for the same frames from 190 to 210% for an
 average of 200% slowdown.

 All of these benchmarks were done without Field rendering which I do use
 in all final renders for my logos.  I am assuming that since field
 rendering does some stuff twice, the slowdown will be even worse.
 Frankly I don't have the kind of time I would need to do a proper
 benchmark comparison.  I could tell the first time I pressed F9 that 4.0
 was a pig.  I really don't know from programing and compilers and don't
 really care why it is a pig.  I am like my clients,  I need to know:Can
 you do it? By when? and for how much? LW4.0 can do it but the time it
 takes to do it, costs to much.

 After all the bullshit NT has put us Amiga based prfesssionals though over
 this important "upgrade" in this last year, (when I took delivery of my
 flyer last Jan., I read the manuals through andd was quite excited about
 the LW 4.0 manual actually thinking that was what I had until I actually
 fired it up..... one more false advertizing lie by omission from NT) I
 must say that they have finally put me over the line.  I am going to Intel
 for rendering.  I want my money back or an Intel cross platform trade for
 the IntelLWSA.  I have the flyer in he 2000 so I will be using that lame 4.0
 over there but I have to render somewhere in a reasonable length of time.
 I find that IK is backwards compatable for the render of 4.0 scenes in 3.5,
 but are the cool new procedurals, Limited regeon bones and plug-ins?

 I really feel like an idiot.  I am sitting here for a year, waiting for
 the product that will let me do all this great stuff while I render double
 speed on two 040's ( I have set up an 3000/040). It boils down to this
 4.0 will render slower on two machines than 3.5 did on one machine!!!!!

 I would like to see some others' benchmarks for other types of scenes.
 Even if most people use 4000's and only get 30% slowdowns, that is not
 acceptable either for an upgrade that made rendering speed a priority in
 past upgrades. Also, it doesn't help me.

 Joe Freeman the Video Junky