Board index » delphi » Re: 64 bit (?)

2003-09-04 06:11:23 AM
delphi208
In borland.public.delphi.non-technical, "John Kaster (Borland)"
<XXXX@XXXXX.COM>writes <3f566233$XXXX@XXXXX.COM>...
 

Delphi compiler, be it Win32, Linux, or .NET is probably 10% where 90%
is the ide/vcl/etc. Delphi (the total package) costs anywhere from
$100 to $10000 with most people paying on average $1000.
If the compiler is 10% of the time investment, then I think it's
reasonable to expect a standalone compiler to cost 10% of the cost.
100 bucks.
Say you get 10,000 sales of the standalone compiler, which is easily
10% of the overall Delphi sales. That makes it a million dollar
product. Now depending on how much work is remaining to complete the
project, it may or may not be worth the investment. Say you need 1
month developer time to finish off the project. At 100,000 a year for
this said developer, that would be 10,000 to finish off the project.
That is a 100x return on the investment.
Now if that developer must be Danny Thorpe, then you must consider that
to finish off the compiler project he will be ignoring other projects
and pushing back timetables. This obviously adds to the investment
costs in other non-tangible ways.
I'm not convinced it is really a user expectation issue as John
proposes. I am more likely to believe that it is a "Danny Thorpe is only
one man" issue and the other projects he's working on expect to return
much more than this measly $900,000.
That's my perspective anyway.
Oh, and regarding the De{*word*81}. If it were a Java Byte Code compiler,
wouldn't you theoretically be able to use a standard java de{*word*81}?
Maybe not, but it seems reasonable.
--
Jason Southwell
President and CEO
Arcana Technologies
www.arcanatech.com
 

--
Jason Southwell
President and CEO
Arcana Technologies
www.arcanatech.com
 

"Captain Jake" <johnj[nospam]@comcast.net>writes
-Danny
 

"Captain Jake" <johnj[nospam]@comcast.net>writes
"droogies" refers to "friends" Even the milk bars in the film were a
Russian cultural reference.
-Danny
 

"Danny Thorpe" <XXXX@XXXXX.COM>writes
;) I guess my reasoning is that I don't know if I would trust the Wintel
arrangement, seeing such big bucks are on the table, meaning I don't think
anything MS does is going to be optimized for my chipset. I would also add that
the compiler should be done, otherwise they wouldn't know if they got most
of the bugs knocked out of the chip... These leaves the frameworks (couple
more hours right ;) )..
By the way, I am a big AMD fan, have been since the 386DX 40. Currently
typing on a 1.7Ghz Athlon XP, and is the 7th AMD chip I have personally owned.
I usually recommend AMD over anything Intel has for the performance along,
and they happend to me less expensive most if not all the time.. If it
weren't for AMD, and Intel had their way, we'd still be upping Mhz by 10Mhz
at a time, and we'd probably be at about 400Mhz by now..
thanks for the reply..
Curt..
 

"David Erbas-White" <XXXX@XXXXX.COM>writes
product with fast compile speed. There was no database support, no
application framework, no de{*word*81}, no external linking (objs) and only a
rudimentary editor.
Support at that time consisted of 3 people on the phones. One of them was
also the product manager, but he only worked part time while finishing
school. There was no in-house development staff at all. (TP was licensed
from Anders Hejlsberg, who ran his own software business in Denmark. Anders
did not become a Borland employee until 1989 or so)
Turbo Pascal 1.0 was never a corporate strategic product. Turbo Pascal 1.0
was successful for the same reasons the IBM PC was an unexpected hit - it
appealed to the hobbyist and individual coder, and was completely ignored at
the corporate level.
By Turbo Pascal 3.0, the price had risen to $99, and by Turbo Pascal 4.0, to
about $150 I think.
$49 products are still ignored today at the corporate level. As ridiculous
as it sounds, the number of zeros in the price does imply "substance" or
"heft" in many people's minds. The low price of Turbo Pascal 1.0 prevented
it from being taken seriously by the majority of professional coders of the
day.
-Danny
 

You need to draw every unit (in visible part of map), you need to pathfind for every
moving unit, if it is a computer's army it should make decision for each unit, also calculate
trajectories for bullets, missles, etc. More units - more calculations. Also map itself is vivid in sense
you build/destroy buildings, eliminate trees and so on (against static maps in IWD) .
it just due some stupid stubbornness.
All those rules are just dice rolls - no? (and random function is quite fast, or do you think they
use some gaussian distribution routine?)
And btw how 3d Unreal relates to 2d IWD?
a LOT of units - again pathfinding calculations.
complicated than it.
 

messages, That is why I put that "smile" in previous letter ;)
 

Dennis Landi writes:
marketplace if you can not think beyond Itanium 2 or AMD64. You don't
see AMD64 processors making major in-roads into the desktop, do you?
They (Intel) don't have any reason to lower the price yet when they can
charge an arm and a leg from customers who rely on Intel's "quality" (I
put it quotes because while I do believe Intel CPU's are of a higher
quality than AMD CPU's, I wouldn't necessarily use it as a deciding
factor in choosing one technology over another-- but the point with
that statement is-- some people DO choose Intel for that very reason)
and pour all that money back into R&D for their next major rev of the
Itanium processor.
I imagine that it'll basically play out like this-- AMD will succeed at
gaining momentum with AMD64 and will make progress with the desktop
marketplace and PC makers. Intel will see this and react-- now how
they react is somewhat irrelevant, but still interesting to ponder
nonetheless; they will either a) push forward with the Itanium,
lowering the prices in an aggressive bid to displace AMD's market
penetration with AMD64-based CPU's or b) introduce their own (but
incompatible) 64-bit extensions for x86 processors. Again, with either
one, expect Intel to bet the farm on it, and IMHO, expect them to come
out winning just as they have with the Pentium 4.
Of course this narrow vision issue also applies to Borland. It's
unfortunate that Borland has taken a step back it seems from being
innovative and instead just reacting to what everyone else does. Sure,
not making a bet or a gamble keeps Borlands shareholders happy and
maybe even makes them profitable for right now, but a day of reckoning
will eventually come upon them, and the mistakes of the past will
eventually catch up to them (especially if .NET flounders for whatever
reason).
Will
 

Jason Southwell writes:
smoldering and eventually toasting the motherboard. The Intel CPU,
under the same circumstances as the AMD CPU, simply lowered it is clock
speed to keep from overheating, and ultimately locked up IIRC (rather
than toast the hardware and render the whole system a waste).
AMD processors aren't that great, quality wise, IMHO, and there's a
reason they're cheaper.
Will
 

Obviously, you don't see difference, but i do. Example - both maps made from
tiles, but RA's tiles is pretty standard and selected from 30..40 predefined
tiles. In IWD nearly each tile is unique (RAM!!!). Example. In RA each unit
consists from few pixels and has few animation frames, in IWD one unit is
pretty big and contain lot of frames. Dig game resources, you will find that
count of sound effects used in RA and IWD very differ.
AR>You need to draw every unit (in visible part of map), you need to
AR>pathfind for every moving unit, if it is a computer's army it should
AR>make decision for each unit, also calculate trajectories for
AR>bullets, missles, etc. More units - more calculations. Also map
AR>itself is vivid in sense you build/destroy buildings, eliminate
AR>trees and so on (against static maps in IWD) .
As i remember RA's units move throuhg map cells. Please try to implement
pathfinging on the cell map and on the not-cell surface with random
obstacles. Just to have understanding what you talking about.
AR>communicate with you.
You want that to you were respectful? Act also in relation to anothers.
Bill Gates, the "anonymous" programmer from BlackIsle, or your opponent in
conference - it does not matter.
You playing games. This means that you may waste your time on another things
instead of games.
AR>also I am quite bored with the game itself and finishing it just due
AR>some stupid stubbornness. All those rules are just dice rolls - no?
AR>(and random function is
No. Just open AD&D player's handbook...
AR>quite fast, or do you think they use some gaussian distribution
AR>routine?)
Your representations about game mechanics are very superficial.
Im starting to understand John Kaster. Just now I have understood as
difficultly to communicate with people when they see only that that want to
see. They try to discuss about what know only a small part. Oh, John, please
forgive me. I too wrote silly letters.
Example. Is RA units able to drawback having theirs health low? Is medic
itself able to go to wounded soldier and heal him?
Yes. Noticed that programs on scripted languages slower than programs being
compiled into binary modules?
AR>And btw how 3d Unreal relates to 2d IWD?
It relates to your PC (specifically CPU). Just to show how it weak. Even if
it will be equipped with high-end 3D accelerator and huge amount of RAM.
AR>may agree or not.
And also i have all rights to have (and declare) opinion on any subject.
Including your person, and your opinions. You may agree or not, may like or
not. :-)
AR>complicated maps with a LOT of units - again pathfinding
AR>calculations.
So Westwood programmers also not good if their app has difference in
performance on nearly same machines.>:->
AR>and was a little more complicated than it.
Please, prove that thought.
AR>even those times AFAIR.
Here again your representations are very approximate and superficial. And i
said about speed, not about principial possibility.
BTW, pattern recognition was possible even without PC. OCRs existed from
times when alphabet was created. :-)
 

AR>Huh?
AR>You need to draw every unit (in visible part of map), you need to
AR>pathfind for every moving unit, if it is a computer's army it should
AR>make decision for each unit,
Pretty simple decision. Move or fire.
Example for fire.
for I:=0 to FriendlyUnits.Count-1 do
begin
FriendlyUnit:=FriendlyUnits[I];
EnemyUnit:=EnemyUnits.GetNearestUnitInWeaponRangeTo(FriendlyUnit);
if EnemyUnit<>NIL then FriendlyUnit.FireAt(EnemyUnit);
end;
I see it is extremely resource demanding algorithm. :-)
AR>also calculate trajectories for bullets,
AR>missles, etc. More units -
AR>more calculations.
You really think that they calcultate trajectories for bullets? :-)))))
AR>Also map itself is vivid in sense you build/destroy
AR>buildings, eliminate trees and so on (against static maps in IWD) .
Replace one standard tile with another. Just change bitmap pointer for tile
from one bitmap in bitmap cache to another. Next drawing cycle will render
different bitmap with destroyed building. For animations there are few
bitmaps for sequential drawing cycles for states 0,1,2...N. I don't see
anything complicated. Changing few pointers at a time and that is your map
vividness, not big deal at least from first glance.
 

On Wed, 3 Sep 2003 16:54:16 +0100, "Per larsen"
<perlATturbopowerDOTcom>writes:
That is because that definition is nonsense.
Oz
 

Intel had it, while AMD did not. Does anyone know if the new AMD CPUs
(32 and 64 Bit) provide this, too ?
thanks, Michael