If, for example, I use a setting of: Sun heading = 44.8657°... does this calculation take longer(however slight, it all ads up, in my book) than a 'round' value of 45°?
I always make my TG numbers rounded, usually in incremental quarters if I can(0 - 0.25 - 0.5 - 0.75 - 1). I never usually see a render longer than 4-5 hours and am shocked and appalled by the reports of massive render times here sometimes.
I'd never use a colour value of '0.873649', for example, that would instantly be changed to a '0.875'.
So, the question is: If I were to not bother about this, over an entire scene, would the difference be apparent in computer/render time?
Just something that's bugged me for a while, not serious. :)
Hi dandelO,
There may be some numbers which end up being special cased by OS maths functions ( or maybe even some TG functions ) and handled a bit faster than other numbers, but with all the calculations TG2 does what you put in the UI doesn't necessarily stay in that form for long.
I personally wouldn't worry about it :-).
The long renders are no doubt due to other factors, quality settings etc.
Regards,
Jo
Quality settings make the largest difference on my side. I wouldn't think rounded settings would make a major difference against non-rounded. If there is one, it's very slim. As Jo said, other factors will determine render times. Incorporating lots of shaders, creating deep shadows, and water will all push render times to those times you see around here. Having render times around 24 hours is not unusual at all.
If anything its a very slim difference. As a computer treats the number 1.123456 the same as 1 (which would really be 1.000000 in the computers mind). Just because the decimals arn't displayed doesn't mean its not there.
From my youth spent programming in assembler. Whole numbers and fractional numbers are treated the same. Floating point numbers get handled just a little bit differently. So, say a whole number takes one clock cycle. A floating point value may take two or three cycles.
It's been nearly twenty years since I've done any serious programming, I've forgotten a lot. But anyway, the input numbers are most likely treated as floating point numbers. It's easier to write and call one routine to process them, than to write, test for type, and call a routine.
P.S.- Is there going to be a Terragen version for the Commodore 64? ;D
Quote from: njeneb on May 30, 2009, 08:55:25 AM
P.S.- Is there going to be a Terragen version for the Commodore 64? ;D
It's still rendering the test scene. ;D The problem is that you keep having to rewind the tape it is using for virtual memory.
Is the Sinclair Spectrum+ 48k version much further behind it?
Imagine working with all that processing power! :D
Hi,
Quote from: njeneb on May 30, 2009, 08:55:25 AM
P.S.- Is there going to be a Terragen version for the Commodore 64? ;D
I learnt to program on a C64. The Terragen 2 C64 port was finished several years ago, but it's still loading off the tape for its first run :-).
Next port is the Atari 400. Unfortunately it's taking a while as I don't have any storage for that machine and I have to key the program in each time...
Regards,
Jo
QuoteI have to key the program in each time
:D :D :D
My first experience was on a DEC PDP-11/45 (mini computer) for a flight simulator. It had no floating point processor so we had to write all the aerodynamics in assembler with what was called "B scaling". The instruction set was pretty basic: add, subtract, multiply, divide, and shift (all integer based) To make matters worse, the other division of the company (Singer-Link) used "Q" scaling. Don't remember now since I worked on both, but one used bit 0 as the decimal reference point and the other used bit 15.
Fun times!
You Guys are telling your ages. ;)
I'm in the habit of using 8 bit multiples, 64, 128, 256 etc pretty sure it makes no difference but like I said its a habit :)
richard
Quote from: cyphyr on May 30, 2009, 01:07:28 PM
I'm in the habit of using 8 bit multiples, 64, 128, 256 etc pretty sure it makes no difference but like I said its a habit :)
richard
I thought it was just me. :-[ :)
Quote from: cyphyr on May 30, 2009, 01:07:28 PM
I'm in the habit of using 8 bit multiples, 64, 128, 256 etc pretty sure it makes no difference but like I said its a habit :)
richard
I tend to do this too :)