ASM Community

X-Calibre is right.
X-Calibre is right.
X-Calibre is right.
X-Calibre is right.
X-Calibre is right.
X-Calibre is right.
X-Calibre is right.
X-Calibre is right.

...yeah, we hear you X-Calibre. :rolleyes:

Hi X-Calibre,

I don't understand why you said that we want to code a new
OS when we *just* want to send commands through the HAL
instead of doing it through DX.
Be sure that learning the command set of the HAL is probably
simpler and shorter than learning the whole DX.
In any case, why do you get so much angry about that? Even if
for us it was a total waste of time:

1) it would be OUR time wasted, not yours
2) it would be an enlightening experience that would make us
all then think along your lines: but we would then advice the
"newcomers" to go through this trial, since errors are the best
way to learn.

It's been since I was born that I've always worked this way,
and with great success. I don't know if you ever had a game
of yours distributed commercially in the whole world, but that
is encouraging for anybody to trust himself to not be an idiot,
believe me. I trusted not to be one even before having my
first commercial game published, but that's another story.

I do realize though that BogdanOntanu is working on his own
"OS". Well, throw shit on me, mate.. I've been doing it since the
start. I had my own "OS" on the Amiga, which partly, very little,
used functions of the original AmigaOS, and I implemented many
other useful (for me and my games) functions by myself. All in
all it became more an OS than a wrapper, day after day. And
I loved it and it worked great.. who coded games 100% in
AmigaOS instead had crap, slow, choppy games. I even wrote
my own programming language: do I deserve hell for that??

I have my own "game OS" both for Dos and Windows, under
the same .EXE, that loads and executes my own custom format
of executables (that I created not because I'm dumb or have
time to waste, but because I wasn't 100% happy with what the
PE has to offer, and wanted added functionality/versatility as
well as code crypting and more). Under Dos I have functions
written by myself that access the hardware directly, under
Windows it's more a wrapper to DX and in future hopefully to
the HAL, as well as many other functions I wrote by myself for
my games, functions which I find useful, otherwise I wouldn't
have written them, guess what!

What's so wild and wrong with this approach? Why do you get
so much angry at that?

We exchanged a couple of private msgs with "peace" intentions,
which I can only re-offer now and anytime, but please take
my serene and friendly view that here you're being a bit too
overheated about how we spend OUR own time. Nobody ever
forced you to team up with us.

By any means though, do not call us "stupid".

NOTHING PERSONAL, but please remember that often the most
presumptuos and ignorant people have a similar attitude to the
one you're showing, and don't even realize how much ignorant
and stupid THEY look.
If we write that Windows is not the best OS in the world (you
write the opposite instead), cope with it. Personally I think that
quality has never been what Microsoft has been known for, and
even if other hundreds of millions of people didn't think the same
thing, I would probably still think the same. Be clear, I'm always
open to change my opinion, but I have never seen in my life
crappier code than Microsoft's one.

I'm glad your opinion is the opposite: in a world ruled by Microsoft
you'll be a happy user and a happy programmer. I don't have any
problem with that.

In any case I'm not going to call you or your choice stupid, nor I
ever called you arrogant just because you said that Bill Gates (or
his company) are the best and most ingenious programmers in
the world. Personally I can only get a laugh at such a statement,
but NOT a laugh at you. I hope the difference is clear.
I can laugh also if my brother says that his motorbike is better
than mine, but I will still 100% respect him.

I hope you do the same. At the end, I hope with friendship, we
can compare the final results. Our games to the games of those
that think along your lines, etc.. Personally I'm not worried about
anything.

Greets,
Maverick

Not sure what to say after all that......


Every DX book will tell you not to, MS will tell you not to, this will not make you 133t, use at your own risk, may cause undefined behavior, I have not done this beyond a front/backbuffer for page-flipping.

Lock your surface only once, save the pointer to the surface. Use the pointer/handle to the surface and never Lock/Unlock again.

I know of one commercial video game that shipped using this method, during development, it proved stable on a variety of cards. If you do not consider this a viable method, that's a-okay with me.

Hi,
I thought about it too.. but if I'm not mistaken it won't allow you
to render a 3D scene.. but just to draw some 2D gfx on screen.

Not really "complete".

Thx anyway
:)

The game in question used 2D and 3D simutanesly(sp?).

It also wrote data to the z-buffer. This was back in DX 6.0 days.


The pointer is good if you want to write data directly(not good for 3D). I have not used 3D, but using windowed mode, I still could use the handle to the surface to Blt to my front buffer.

I don't know about billyboy himself... but microsoft
do have some of the best programmers in the world.
Too bad they also have a lot of stupid f**ks that mess
up the whole thing with all the extra bloat included in
windows... and office... and all that. Now, let's not get
win9x into this discussion, it's a pile of sh1t that was
NOT written with stability in mind. It had a place to
fill, and it did this perfectly. It stayed too long, though.

The NT kernel is written for stability, and as long as your
drivers are stable, you wont have a kernel crash. Quite a fun
thing, microsoft written drivers are often more stable than
the drivers other manufacturers write. Like Creative... I've
had three BSODs under win2k, and two of them were because of
creative. A guy once called creative support to bitch about
his SMP system being *very* unstable with their crap products.
And the first thing the techie said? "WOW, you got a SMP system
to boot with our drivers?!". So don't blame microsoft for every
little fault. I don't like them and their bullying tactics,
but I know something good when I see it: windows 2000.

HAL and DDI access is ring0-only. Furthermore, DDI access is
not too easy to get your hands on unless you're writing a *video*
driver. Video drivers run in their own little "sandbox" and only
have access to a subset of the "API" that normal drivers have.
In fact, if you have imports from other (ring0) DLLs than what
you're supposed to (win32k.sys as far as I remember), your video
driver will not load.

So, what does this mean? You'll have to code "a few" KMDs,
possibly including a video "proxy" driver. You'll have to write
an interface between your ring3 part and the ring0 drivers, and
this will have to be written *very* well, otherwise you speed
will suck utterly. You'll also have to be *very* careful, as a
ring0 crash is a pretty mean thing. And worse under NT kernels
than 9x. Also, people will need Administrator privileges to run
your stuff... quite a bit annoying, really.


Take a look at some GNU software, and in particular, the linux
kernel. Oooooh boy ;).

As for no direct pointer access in DX8... your statements seem
to be at least partly guesswork, so I will take the liberty to
guess a bit as well :). In a 2D application, I can see no reason
why Lock should have been removed (although I remember something
about it being renamed...) However, it makes *perfect* sense that
a front/backbuffer lock is slow in 3D accelerated mode, since you
have to sync up with the accelerator hardware. You wouldn't want
the accelerator to overwrite the pretty pixels you have just doodled,
would you? ;-).

As for "proprietary texture formats to allow greater speed" - naw.
You still need to load your textures, and procedural textures are
popular enough that people like nvidia will *NOT* be stupid enough
to make this impossible/infeasible.


That is true. The problem is just that there's a lot of people who
don't seem to know why they use asm... who don't know it well...
who use it just to use it. I'm not pointing at you, and I won't
mention any names. But they're there, and there's lots of them.
Quite fine with me, as long as such people don't start attacking
other things they don't really know anything about ;).

CoffeeDrinker,

Bad bad bad. This can lead to some pretty nasty lockups. Might work
well on most machines, but... uh-oh. No go :). While you certainly
shouldn't lock/unlock per pixel, there's not much point in locking
only once. Lock screen before a screen update, unlock it when you're
done (before your AI code / whatever runs). As long as you do it this
way, Lock is sufficiently fast that a good old pentium can run a fullscreen
320x200x32bpp plasma synced to the screen refresh.

Hi All,

I exchanged some nice and friendly private msgs with
X-Calibre recently, but since the discussion is not
really between me and him, but rather between two
groups, I'd like to make my reply open, as well as to
read whatever he will want to add, in a open forum.

---

30 minutes ago I turned on my PC, booted WinME, I ran
Eudora and IE. Then I opened 4 IE sessions, on URL's of
the win32asm board. Then I disconnected from the Internet,
and this very action made my PC crash so bad that I had
to hold down the ATX power button for 4 seconds.
Then it was ScanDisk time on both my 60GB and 40GB HD's.
It took quite a bit to return to Windows, will all the
windows drawers lost (so I've to remember in which
drawers I was working, I do too many things at a time
and sometimes losing the opened drawers may get a problem),
and had to re-connect to the Internet and repeat the IE
part because History contained too many links (from last
late night).

Honestly I did more frequently this kind of "unjustified"
crash experiences (I mean, I wasn't running any "dangerous"
program, or any program that normally produces crashes..
I was even running just two programs!!) with Win98 and
Win95, so I don't blame WinME specifically (I'm even one
of the few that says it's the best Win9x).

I don't want to blame anybody, although I claim my right
to say that Windows is decently-thought, but bad-written,
bad-implemented and still buggy as hell.
That's my personal opinion about Windows and most of the
Microsoft products; I don't want to show any personal
aggression to anybody though, or to anybody's opinion.
It pisses me off a bit, though, if somebody (it happened
not on this board, but elsewhere) in response of what
just happened to my WindowsME writes something like:

<<
You stupid, change your mind and stop keeping opened
drawers when you turn off your computer. You know, it's
life, you can lose them anytime. Live with it, change
your mind, the OS is perfect.
>>

I mean, why the hell do they let you take advantage of
a feature if then it's lost like nothing? Or, anyway,
I know the differences in protection schemes implemented
in 9x vs NT, but why the hell should Windows9x or
anything else crash by itself when I run Internet Explorer?

If Microsoft got such a bad name, there must be a reason.

There will be who (because of past reputation) would deny
that Microsoft got great even in a (possible, but IMHO not
probable) future when Microsoft starts to write excellent
code. But there will be also who has always denied other,
really happened/happening things.

In any case I don't think that COMMERCE, MARKET, BIG
COMPANIES WARS.. should create hatred and bad relationships
among normal users or programmers.
Those big companies often speak motivated by money more
than intellectual honesty of opinions, so at least us that
do not gain anything from saying lyes, please let's be
objective (which still doesn't mean to agree on everything)
and in any case friendly.

We should remember that as soon as we get out from this
forum and maybe go into Usenet there will be some people
which will insult us *only* because we ever coded a line
of assembly in our life. They will say it's immoral, it's
a sacrilege, a mortal sin.. that we must forget ASAP that
we've seen the devil with our very eyes.

I think in a much more practical way: I want to use the
right tool for each job. For temperament I tend to create
and use my own tools, not always of course, but often. In
any case I reckon that if you want to be more competitive
than others, sometimes you've to pay a higher price (i.e.
no pain no gain). This may mean going to a lower-level,
thus coding in hand optimized asm some key routines or
maybe evaluating the benefits of skipping the limitations
imposed by DX 8 (e.g. in using videomem pointers directly)
and talking directly to the HAL instead, this may mean
many things.. but always with a better final result in
mind.

In any case I can't blame those who maybe code in asm
*just* for the pleasure of it. I may blame them only if
I was their manager and payed them to do a work in a
certain deadline: but I did voluntarily chose not to be,
and in any case I can't blame who I'm not paying.

I do understand your concern that you want to be helpful
and avoid that others make the mistakes that maybe you've
done in the past, and that's why I called "father" f0dder;
but I think that if I'm "mature" today it's because I did
my own errors and learnt from them, if I listened blindly
to my father and my mother today I would have almost no
personality. I'm not even sure I would be safer than I
am, also.
Moreover, in the specific, I have to add that I do not
agree with some opinions of you or f0dder, and I don't
think of myself less experienced than you (it sucks to
say that, but self-confidence is a bad disease.. often
necessary too, though.. without it I would have never
finished a project in my life).

Anyway, why can't all be friends and polite to each
other? We risk to end up living with the finger on the
trigger, always making flamewars, and not even remembering
why it all started.

Personally if I can help in a thread where I 100% disagree
I try to help anyway, we're all here to help each other
after all.

I also want to explicitly say (as I said in another
private post to X-Calibre) that I'm not a saint. Years
ago I did my good dose of flamewars, and I wasn't more
polite or less aggressive than anybody else in this
forum. I did really understand though that it didn't
bring to anything good (if you ask me I was really
reacting to their aggressivity, but at the end one
becomes like the others, without even realizing it..
and calls aggressive the others being even more
aggressive, etc..). I think there are other more civil
and friendly way that do more good to oneself and to
the community. I really think it.

Also, I think that so many hours "stolen" to coding
just to keep flamewars is the most agreeable "flag"
that we're all misusing our time.

Greets,
Maverick

A precisation: it seems that when I talk about stability problems I talk about WindowsME while X-Calibre or f0dder talk about WindowsNT/2000..

Well, I personally think that when we talk about Windows we should talk about the most common and available version of Windows around (be it 98?), and that it's not "fair" to resort to the most office-oriented or server-oriented versions just to prove its stability. We (me at least) were talking about games at least, where NT/2000 isn't a king really.

Also, I personally *do* think that, given its degree of stability, NT is slower than it could be and thus should be. Sure, you can tell me "go code your faster version then", but it's a joke.. because:

1) it wasn't me who denied the hardware programming documents of most chips, or other important information that killed everything that wasn't Microsoft (even today the Linux community can NOT have, neither under NDA, the hardware documentation for the "SoundBlaster Live!" EMU10K1 chipset, so all was guessed by reverse engineering, trial and error)

2) even if I won't write "QuickNT" I have the very right to point out that it's much slower than it could/should be. It's like if in a Formula1 grand prix the fastest runs at 1m:20s per lap, the slowest at 1m:25s, and suddenly Barrichello starts to run, with a perfect car, at 1m:30s. I have the very right to point out that he's going "slow", even if in the same car I couldn't go faster than 1m:50s. It's not that in that scenario you can tell me "go run yourself, let's see if you do better", I could very well protest that the possibility to run in Formula1 that were given to him hasn't been equally offered to me, and that ANYWAY he's going slower than he could/should.

3) the whole point is IMHO that Microsoft is NOT concerned with technology, but only with money and market. I.e. as long as the users don't stop buying their products, they will have no stimulus to improve or optimize them. Also, I do not think that the best programmers in the world are housed at Microsoft, I rather think that the best programmers in the world would never work for Microsoft; although, as in every thing, there are exceptions.

In any case, I've used the latest in NT technology: XP.. and I was sincerely disgusted by it. That's the thing more distant to a "realtime" OS I've ever seen.. let away that its wonderful "security" let anybody access anybody's else computer unless you install TWO patches (honestly it's ridicolous), and it serves the record industry instead of the user that purchases it, by deliberately reducing the quality of MP3's and by putting fingerprints everywhere.

When I have a PERSONAL computer, I want the computer to serve my interests, not the big industries' interests.

As for being the "most successful" company out there, I think that fair competition should be mentioned as well, and all the trials/proceedings in the Courts, against Microsoft, that have many reasons to exist than the ones that at the end will be carried out successfully.

I don't understand why some users are so fond to Microsoft. They gave us the crap Dos and the crap Windows, etc.. I personally think that everybody who has seen a lot of OS's and computers in their life can easily agree that Microsoft did more harm to the computer world than bringing benefits. Microsoft was always the last one to jump on the vagon.. be it Internet, be it multimedia, be it multitasking, be it anything.. When you have such a capital you CANNOT ruin it, but that doesn't make you a good manager anyway, even less a good software developer.
If Microsoft never existed on the face of the Earth I'm really sure that the world today would have had better computers. I don't pretend to convince everybody, nor I want to despise or insult who thinks the very opposite, I just think all of this is so clear and straightforward that I cannot stop from saying it.

Greets,
Maverick

Okay, a few thoughts here... 9x was never written with stability
in mind. It was written as a migration path to NT technology
(redundancy, huh? New Technology Technology.) It served it's
purpose sort of okay. Legacy stuff ran well most of the time,
and you had a more or less complete subset of the win32 API.

Microsoft's biggest flaw was to carry on and carry on and carry on.
98. 98 second edition (most stable 9x to date). WinMe (YUCK!).
The mixed 16/32bit architecture of 9x is not good in the long run,
not good for stability, not good for performance.

Microsoft cares a lot about money and market share, yes. That's
why they continued 9x. Why they continue adding craploads of extra
stuff in Office. But they do care about technology as well... in
their NT kernels. There's not much use in trying to beef up 9x
anymore, it would be like stuffing a porche engine in a lada.
Or a jet engine, perhaps. The framework just isn't stable enough.

XP... humm. Not many improvements from 2k kernelwise (NT5.1 vs NT5.0).
Made the UI ugly (fortunately that can be turned off). Add a lot
of useless crap. New security holes... guess this could only be
expected (if not justified).

Realtime? They never claimedn or aimed at being realtime. However,
and win2000 fulfills this (and XP as well, I guess), they aimed
at stability and security. Yes, IIS leaks and reeks, let's not
talk about this. I'm talking about kernel security and user privileges
on the local machine. This seems to be pretty solid on 2k/XP.


If you use microsoft software for ripping/encoding, yes. But nobody
stops you from using EAC (or cdex) and LAME - which you really should,
if you care the least about quality :).

I don't know where we would have been today without microsoft. They
have certainly done a lot of bad things, crushed small innovative
companies, et cetera. But at least today we have an OS that runs on
a multitude of hardware configurations and offer good performance on
most hardware... might have happened without MS, or perhaps we'd still
be in a DOS like environment with poor support for anything. OS/2 could
have been good (even though the first implementations had major flaws).

Dunno abot BeOS or the nextstep stuff or QNX. Perhaps it would be fine
if it had better hardware support - I don't know enough kernel/performance
statistics, and I don't just buy the hype without looking a bit into it.
Linux just doesn't cut it for desktop use, for a multitude of reasons.

I wouldn't like to see something *perfect*. But I know this just isn't
going to happen. It wouldn't have happened if microsoft hadn't been
here, there would just have been some other bullies. The world works
like this. We have to make the most of what we got... at least win2000
is a *decent* operating system, if not everything I would want it to be.

If you want something good? Take the win2k kernel, re-implement the
core win32 api more efficiently, and throw away everything else. That
would be a *very* good building stone for a high-performance and highly
stable operating system.

And I should have added, "and Fodder will say its 'Bad bad bad'":grin:

I guess I should also stop modifying esp.

It's just my style to program in a way not considered safe.
If you don't want to take the risk, use the safe way.

-----------------

Now as far as windows being #1. It is the only OS that has made it simple for clue-less people to use AOL. There is a lot of market in the lowest common denominator. Linux and FreeBSD will have to greatly improve their install process to catch up with Windows. The install process alone is enough for me not to want to use them.

CoffeeDrinker, modifying esp is fine, if you know what you're doing.
Like allocating a temporary buffer by "sub esp, 256". Nothign wrong
with that really.

But the constant locking of a directdraw surface... it doesn't really
serve a purpose (locking/unlocking once per scene update is *not*
a performance problem), while it does add problems (constantly
holding the win16mutex... icky icky. Not good if a taskswitch happens).

FreeBSD has plenty grunt for server work. Linux... well, it sorta
works, and at least apache has less security flaws than IIS. But I'd
still use BSD for a server. (I run linux right now - couldn't find any
decent encrypting filesystem for BSD :( ).

Thing is, linux just doesn't cut it for desktop use. Kernel is a mess.
Distros are messed up. Thing are (VERY) poorly documented, if they're
documented at all (somehow I don't think "read the source" is
documentation). XFree86 has very sucky drivers, and many X apps has
their entirely own standards. Ok, Gnome and KDE are being more widely
used, but still.

It takes a whole damn lot more than a "nifty" install to bring these
systems to the end users. And thanks goodness for that... neither are
ready, and their users/developers don't have mentality that would allow
it for "real" desktop use anyway.

fodder, let's just say I'll do that bad thing with DirectX, just because I can. Someday I'll do it the right way.

You expanded quite well with what is wrong with xfree86 and Linux. They can't make a nice installer, cause they are both a mess. How can you have a messy installer that makes a nice OS? Entropy doesn't work that way. A messy installer makes for an even messier OS.

CoffeeDrinker, a messy installer but a nice product can certainly
be possible. I find the windows installer (.msi files) quite messy,
but the products can be quite fine.

Some of the linux distros have "cute graphical installers" - take a
look at mandrake, for instance. You need to answer a lot more
questions, but you get more flexibility that way. Still doesn't change
the fact that linux is a messy junkheap though ;).

As for doing something just because you can - why do it if it's a
worse way to do it?

Just because I can.

I don't do everything with a logical purpose in mind, sometimes, "Just because I can", is enough.

I've written my code so it is easy to change to the proper way.

Well... I am only in the initial stages of the DynatOS IA-32 Core Kernel... but I will inform you guys of anything that I find interesting... there is always some way to do something better without taking too much away from capatibility or ease-of-use.

[ sigh ] At least we can code in asm rather smooth, so it seems that Microsoft is full of hatred for asm-coders and want to banish 'em all. Knowledge is too dangerous.



And end your days in a jail for copyright violence :grin:.

Quite the conspiracy theorist, huh? =). Remember who developed masm...
Microsoft. Who upgraded their assembler to support 3dnow, SSE, and whatnot?
While Borland has stopped developing tasm? Well, microsoft added new
instruction set support. No, masm is not longer a marketed product... probably
because the demand isn't there. But they include it for free in the DDK...
so who do you have to thank? The thing with microsoft, they give people
what the majority want... something fairly easy to use with pretty looks.
If you look around, I don't think you can deny this. I'm not saying this
is a particularly good thing, but it's just the way it is. But yet they still
supply us with lowlevel tools and enough information to program both end-user
applications, games and device drivers. For free (well, ok, certain things like
the IFS kit cost a lot, but I guess there's quite some proprietary things in
there...)


Why would I? As long as I don't try to sell a product based on their technology.
And if reverse engineering is needed to accomplish the goal... it might end up
under "fair use".

I meant .Net and CLR.

I am not sure is it possible to sell VC without assembler.

AFAIK, they also update TASM with new versions of Delphi.

Because they didn't expect money for it...

I haven't to thank Microsoft, because I don't use MASM :). I have to thank Tomasz Grysztar.

Not only. Also they force people to think they want it.

Most of the last versions of M$ software is overbloated. Most of people don't need hundreds of function provided by it. But they are caught by wheel of Microsoft... khmm... sorry wheel of Sansara...

GREAT THANKS! I'm so happy that they gave me a chance to write something for their OS. It's normal that OS-author provides information. And: may be it is Microsoft who must be happy that I code for this OS :grin:.

I heard according DCMA (or ...? ) it is illegal. Of course, it is a problem of US-ppl, but AFAIK Europe is also about to copy some strange American laws.