pouët.net

What makes me sad is that I tried some software codes with simple fx (plasma, rotozoomers, tunnel, vector balls, etc) to a friends PC which happens to be some short of Pentium3. And I noticed things are preety much slow for my liking. I also remember trying the same stuff in my father's Pentium2 to find out that plasmas at 640*480 runs at something like 15fps (iirc). But most times I didn't noticed because I have an Athlon. That's what makes me sad..

..I onced tried to run some software demos of people who made them in Athlon's/P4s on my slow P2 I bought for 80 euros at Germany to whine hard like a madman. I read a thread of a Yaphan demo where I was ranting like insane and there I was again, motivated to open this thread. And my slow on P2/P3 codes are integer ;P

And then I don't think I'll have the time to sit in my father's PC or get some old hardware (I want both P2 and P3 for testing) for cheap money to experiment. I won't have time or won't care much for that! But I have the slight suspicing that the future Optimii of 2008 might watch my demos in their oldskewl PCs and cry and piss me to death as I did back then to you dear koders. Karma?

What makes me happy though, is that I have the Athlon since almost 2 years and when I got it, it was still not state of the art. It seems that modern PCs are more than enough for cool stuff that everyone needs and we are not pushed anymore into a higher level. I remember that when I got the 486, it was outdated. When I got the Pentium it was outdated, the P2, the P3, etc. But it's the first time I have the Athlon and an outdated for today 3d card and that's enough (perhaps only the 3d cards would kill requirements in the future again). There is no more need for more CPU since a long. Like there is a stop in that evolution for the first time. Strange..

But that's not the point today. When I was in quickbasic, someone told me to try C. I did Atsou and I was expecting more from speed. I especially later tried C and SDL in Windows and thought that C would be fuckin fast, so fuckin fast that also with my optimized integer routines I would make fast enough things in even a P2/P3. But you have to stay in 320*240 or a bit more there, that's what I learned in strange agony there! Why isn't C as fast as I thought? If I had the time to make some tests in oldschool PCs I would be happier, but now I don't care and I won't to finish another software demo for my Athlon where everything will be fast enough. Sad :(

But why?

Have you thought of making a blog?
:)

I have a blog, but please don't think that as a blog but rather as a coder's discussion.

If you wonder why software demos are slow today, the answer would be: The fucking Windows XP

I really promise you, if you code your own OS (as I have, not yet finished - but runable) Alla software demos will go hell faster! but if your going to code your own OS, make sure to NOT code an multi threading one. I promise, Single threaded ones are alot better (or faster at least).

Oh, I forgot.. Code it in assembly if you want it to be that fast...

But, yes, ofcourse do faster computers make us not care that mush as before..

I hate PCs! Give me a modern architecture! :(

Yeah, everyone, buy Mac's. Bad configuration - shitty effecience in coding.

A GBA has a 16Mhz Processor!
Thik about how good our modern computers are, coz even a GBA can draw textured 3D-environments..

Before talking about "fast" code and optimizing, you have to meet two conditions:

- You need a target.
- You need be able to MEASURE that shit.

Then you can reach conclusions. You do not reach conclusions about optimizations before having done that.

Plus, if I may quote NN here:
<< optimizing performance synonymous isnt with speed of execution.
http://www.macdonalds.com also isn't. >>

xernobyl: like Power/Gx? Oh shit...

optimus: 64k (ram segment) is enough for everyone, right? :) there's plenty tasks for wich cpu is not yet enough. or bandwidth, or memory. and there will alywas be... =)

"If you wonder why software demos are slow today, the answer would be: The fucking Windows XP"

this is of course totally stupid. maybe it had some truth in it in the times when old windows ran on <=100mhz pc-s (although there were no win demos back then...) but today win consumes somewhere between 0 and 1% of your cpu time in the average. closer to 0 i guess.

the answer is that coders do not care (and/or are incapable to write good code, and/or they don't test their code on slower computers/gfxcards, etc). in fact, they didn't care that much back then either, but still there is a difference.

if you want speed-optimized code, try a limited machine.

yes.

try CTRL+ALT+DELETE => PERFORMANCE to look up how much does clean OS steals off the CPU powa

If it's demo programming instead of real code, and it runs fine on my average to decent computer, why should I care?

Fast PCs did not stop us caring, they only brought more headaches, cause optimizing for Z80 (or pick whatever else) and optimizing for PCs is a whole different thing.

And Optimus, it's not the language, compiler, OS or hardware that makes your code fast or not. It's something you have between your ears.

Just keep on using crap hardware: you will save money and, at the same time, you will not stop caring.
Eventually, you will install Linux on it to keep up with teh new intarweb technologies, and, after a while, you will stop using PCs at all.
Then you'll be saved.

Dixan wrote: .

Excellent.
Santo Subito.

perhaps you shouldnt measure modern technology by ancient metrics.modern pcs dont have quite the same bottlenecks as e.g. a c64 (duh), or even a 486. it's like e.g. optimising your inner loops for cpu cycles brilliantly, but then accessing memory in a completely messy way or copying large things on and off of the gfx card all the time. or making fpu code which generates shitloads of denormals and wondering why it crawls.
but hey, it's fun to watch people think their code is great because "its written in assembler" or similar. :)

oh, and you could just write that plasma as a pixelshader and get like 300 fps in high res. is that cheating? :)

fast pcs stopped me from caring about topics like this :)

I cannot speak for "us", but a fair amount of people currently cares even more about performance problems than ever before since single-threaded programs are not going to be executed much faster in five years than they are now: "The Free Lunch Is Over"
http://www.gotw.ca/publications/concurrency-ddj.htm

http://www.gotw.ca/publications/concurrency-ddj.htm
(perhaps this link works?)

Very true about the concurrency. More headaches! Ah the olden times of even something like a 386, where everything was oh so simple :)

if it werent for fast pcs optimus isnt capable of producing such heaps of text on pouet!!

yes, give him a good old 300 baud modem connection!!