| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 14
Comments:
<0> why you moving there
<0> better summers?
<1> i like to move to arizona
<1> i hate rain and thats all it ever does, here
<0> i prefer rain, makes coding more pleasant
<1> i would agree with you if our roof didn't leak
<0> our roof?
<1> i'll fix it one of these days, if it doesnt go to hell, first
<1> my roof
<0> who are you living with
<1> no one
<1> well, a roommate, which is no one to you
<0> and always will be with that attitude, hmmph
<1> lol
<1> i didnt mean anything by that, just wasnt meaning anyone in here
<1> which you would know
<0> np
<1> so, what language do you develop in?
<0> well unlike most people in here, i am one of those rare breeds of real programmers
<0> so pretty much visual basic
<1> oh, i thought you were about to say ***embly
<0> no im joking :)
<0> i know a bit of asm, x64,x86,z80
<1> cool
<0> predominantly code in C and C++
<1> i tried a bit of x86 and z80, and now im trying to teach myself arm9 asm
<1> well, not right now, but plan to
<0> once you know any ASM all you need is the opcode table for another set ;)
<1> true, for the most part
<1> anyway, im thinking of sticking with c, just because i want to learn os development on the way
<1> which won't be very pratical with c++, so i was told
<0> maybe the lowest of the low in an OS shouldn't be C++
<0> ideally you'd do it all in hand optimized ASM
<1> yeah
<0> C is usually faster than OOPd C++
<0> you just lose the "working benefits" of OOp
<1> yes, but never did OOP before, so im not sure what im losing
<1> and that's bugging me to hell
<1> but I never did*
<1> yeah, ill probably put os development aside and learn c++
<1> see if im missing anything
<0> OS development is a good project, just a big time sink
<0> you should be able to make windows xp fit on a floppy
<1> yeah, that would be a fun challenge
<0> one of those 120MB floppies ;)
<1> heh
<0> it'd be good having an ALL-asm OS though
<0> there probably already is one
<1> yeah
<0> http://www.dynatos.org/project/
<1> interesting
<2> interesting is the w3rd
<3> hrm
<3> that kinda seems like a bad idea
<3> you'd have to really tailor it to a specific chipset to be a good optimizing compiler
<3> and that would mean that by the time you were done you'd have a useless os
<3> err to beat
<3> not to be
<0> chipset?
<2> it has been said that chipset is Intel 440 BX
<0> do you mean x86 ?
<0> or some other "chipset"
<0> like the motherboard, graphics, whatever
<4> **** it I can't find it
<4> LOL
<4> I must have dream it then.
<0> lets code it then rambo
<4> SOAR+ROAM+CLOD
<0> got acronym?
<4> http://drtypo.free.fr/
<4> I don't think it's this one
<4> but I remember the page I saw is got black background, and pretty realistic rendering of planet and ground terrain
<0> that one doesnt seem too bad let me run it
<3> Polestar- I mean a specfic chipset. Like the you couldnt have it run the same on amd's and intels, or between 32 and 64 bit os's
<3> err sorry chips
<4> some fun project :
<4> http://aiplanet.sourceforge.net/aiplanet-old.html
<4> Artificial Planet v0.9
<4> AI.Planet is a virtual world for artificial intelligence. The world can be altered by adding water, land, suns, moons, rivers, and atmosphere. Plants, animals, and fish can be added to create a dynamic ecosystem. Clouds, wind, lightning, water currents, and weather naturally arise from the sun and other influences. You can explore your planet from outer space, by walking around, by tracking creatures, and by controlling a bot that interacts with the e
<0> nick: sure, but i'd stick to optimizing it for AMD
<0> since intel's newer chips are like the k8s
<3> but amd releases chips every few years
<0> pre conroe chips wouldnt be that hurt
<0> so... a new x86 or x64 isnt released every few years
<3> and each has a new instruction set that youd really want to use
<3> they add an instruction every few years
<0> a subset
<0> which only helps in certain operations typically, like 3d math
<3> but you would want to use the new features... a good optimizing compiler would beat you
<0> unlikely
<3> GCC wouldn't
<3> but a *real* one would
<3> hands down
<0> whats a real one
<2> i heard a real one was on irc.afternet.org
<0> because i havent seen one
<3> a commercial one
<0> like MSVC?
<3> no no
<5> ICC
<3> msvc is still just a toy when it comes to c/c++ compilers
<0> ok so list the real ones
<6> devcpp
<3> I wouldnt have a clue. I know they're out there because I've encountered them a few times
<3> but they're in the $10k range per license
<3> and thats just not something that really gathers my interest
<5> In terms of x86, ICC is the best bet
<5> and a manualy vectorized code path will beat most compilers which claim to do auto-vectorization (MMX, SSE, SSE2)
<3> so there you go.
<3> ICC looks decent
<0> a c compiler will never beat hand optimized ***embly, ever
<0> unless they add more to C for hints
<3> PolestaR - thats entirely untrue
<0> no.. it's not
<3> Yes, it is :)
<0> you can't describe everything in C that you can with ASM
<3> as the number of registers grows, humans won't be able to manage them as effectively as software can
<3> right there that gives compilers an advantage
<0> im talking about now
<0> not the future
<3> no you're not
<3> <0> a c compiler will never beat hand optimized ***embly, ever
<0> and then i just added my context
<3> you don't need to add to C for compilers to win
<0> what i said is true anyhow
<3> no, its not :)
<7> Dr^Nick: Unfortunately, ever compiler can beat my hand optimizing :(
<0> hand optimized ***embly will only be levelled by a compiler
<7> *every
<0> regardless of the amount of registers
<0> LEVELLED, not beaten
<3> PolestaR - so, I can already show you machines where thats not true
<0> no you can
<0> can't
<3> they're not x86 archs, but yes
<0> oh
<0> and show me them
<3> its entirely true.
<0> show me cases where a compiler will always beat hand ***embly on these magic archs
<3> here's a simple proof
<3> Register allocation is NP-complete
<3> Which means that optimal solutions for anything but small problems are entirely out of the scope of humans or computers
<3> what it comes down to is being able to do "really really really fast" checks. trust me, your computer is faster at this than you'll ever be.
<0> what isn't a small problem
<0> are we talking about 99% of the stuff which is coded these days?
<3> Are you familiar with NP-completeness?
<3> Basically I'm saying you're dead wrong, and the fact that its np-complete is the nail in the coffin.
<0> a little, im not sure why it's completely relevant to this
<3> http://en.wikipedia.org/wiki/NP-complete
<0> we're talking about writing code here not about solving all problems
<3> If you really don't understand this then you're missing something fundamental about code optimization
Return to
#gamedev
or
Go to some related
logs:
#politics
python iexplore httplib
#gentoo
#computers
#microsoft
#nhl
must activate windows already activated
#beginner
delphi PaintTo
#computers
|
|