The Pressure is On...

[Moose or Chuck]

Hello there. Dan, I just thought I'd let you know that I recently for and put in a blurb and link to your site, stating that a MIPS version is on the way. I just thought I'd let you know. I'm sure redesigning the display routines for (0,0) top left wouldn't be too hard. My reader's will be interested in playing Quake .<br>Hence the "The Pressure is On..." subject.

[Mark Rejhon]

This is not the biggie.<br><br>The biggie is converting Quake to use fixed-point-integer instead of floating-point mathematics.   For better framerate.  He has spent several days modifying 3,000 lines of computer code code, and still has thousands more to verify and modify.<br><br>Consider: One single typo or error, can cause things to go really bad in PocketQuake.  It is NOT a trivial programming task!<br><br>It will probably take Dan East one to about one-and-a-half weeks just to test and iron out all the bugs caused by the conversion to fixed-point.  <br><br>During such a monumental conversion, a programmer is usually unable to test the whole software program.   It's like trying to touchtype an essay blindfolded - you can't doublecheck for correct program operation until you test.<br><br>Usually, a programmer can test a program after just a few lines of code change, but this is a really major change - just to make PocketQuake playable on Casio devices (and also better framerate on iPaq).   Please be patient.  As a programmer, I know that it's not trivial.   The iPaq is about 8-12 times faster in floating point math than the Casio, and without converting these thousands of lines of codes to use the fixed integer math instead of floating point decimal math, it won't be playable on Casio.   Thankfully, it should be - once he's done doing the major converting!

[Mark (KILzt)]

A friend of mine, Tom, just showed me an old game he coded and then converted to the new math routines much like Dan. We had to re-compile for Windows98, but I don't think people will see the huge leaps expected by some.<br><br>Of course I don't even have to state the obvious differences a programmer is going to see in this (CPU, graphics cards etc.). However we did do some specific tests on an older system.<br><br>I think that at best you'll have a cool 6-10FPS to play with where normally scenes might have been 4.02FPS. Worse case is that you'll only get a 2FPS boost, although that's still quite a boost for PocketQuake.<br><br>We're all rooting for the best, but once you take into account the other elements then the chances are it'll be less than most hope. Personally and as an estimated guess I'd say a 3-5FPS boost is the most likely outcome for general play.<br><br>Casio owners will still be in trouble if it's so low, but then who knows. Personally I'd hope it's more and I'm wrong.

[Dan East]

I have to agree with what both Mark's said.  What I am doing now is a far greater task than porting Quake to Pocket PC.  Things are going well, and so far it appears that this will work.  Assuming that the following benchmark is correct, http://www.dalecoffing.com/General/VOBe ... chmark.htm , you will see that the iPaq is 12 times faster than a 150 mhz MIPS machine at processing floating point math.  If that is indeed correct, then I would say that with the existing floating point routines, a MIPS based machine would be lucky to get 1 fps.  Again, this fixedpoint conversion is IMO the only thing that will make Quake playable on the MIPS devices.  So it is much more of an issue than the display hardware's properties.<br><br>Dan East<br><br>Last modification: Dan East - 02/04/01 at 05:00:18

[Moose or Chuck]

No, I realise that. I had just heard, so far, that you've had it pretty much undercontrol and on target. I don't think I personally would have the will power to go through that much code to change it to Fixed Integer, so you've got to be real determined. And a few weeks is nothing. I could wait a month if need be without running up the walls. Good luck on line crunching. And you'll release the source simultaneously, right? Ok, that's what I thought. tchauLast modification: Moose Master - 02/04/01 at 05:19:47

[Mark Rejhon]

Mark - what systems were he testing on?   <br><br>The CPU speed between fixed point and floating point on the StrongARM is much, much, greater than the speed difference on a Pentium 200 Classic chip.<br><br>So, there will be a bigger boost for the iPaq for fixed-point conversions, than if you converted the same game for a Pentium Classic system.<br><br>One can always hope - I suspect that the framerate will roughly double.

[Dr. Phat]

Dan, i thought you said that you would not be porting to mips, and you would be leaving that up to other programmers. am i mistaken?

[Dan East]

Well, I wouldn't call it "porting" to mips. Basically all that needs to be done is to modify the drawing routines to dynamically adapt to the display properties, while currently I have them hardcoded for the iPaq.  This would take very little additional work.  I'm not planning on doing that before I release the source.  In fact, I may release the source before I finish the fixed point conversion, depending on how much longer it looks like it may take based on the next couple days.  At least someone else may be able to write the sound routines while I'm finishing up the math stuff.<br><br>Dan East