View Full Version : Delphi8 for Win32 ?
Goliatus
28-01-2004, 05:24 PM
Delphi8 brings some great new syntax features, but managed code is very very slow(slower than java?) and unuseful for games i think.
I remember that in open letter borland said that they will work on win32 version in near future. So can anyone tell that when new Delphi for Win32 will be available?
Useless Hacker
28-01-2004, 10:11 PM
Managed code is faster than Java, I think about 90% native code speed. (And yes, that statistic was made up on-the-spot. Sue me. ;)) You can make games in it, and I doubt you will notice much difference (except perhaps an increase in the speed of development.)
Managed code is almost as fast as native code (Microsoft says that c# is ~10% slower than C++).
Also Delphi8 for Win32 will be coming out after the next version of delphi8 for .NET.
This is because they are awaiting to standardize the generics syntax that will be introducted with the next version of the .NET framework.
Goliatus
29-01-2004, 12:53 PM
Managed code is almost as fast as native code (Microsoft says that c# is ~10% slower than C++).
In some math, string, list operations test managed code was 2x slower than code compiled in D7 :( If .NET programs run on VM then it is not any suprise for me.
You can make games in it, and I doubt you will notice much difference (except perhaps an increase in the speed of development.)
Heheh, I don't agree with You. Nobody has seriously tested .NET in such things like physics or graphics where performance is really needed.
I think that .NET could find same use as Java in Chrome.
Useless Hacker
29-01-2004, 05:08 PM
After Windows Longhorn is released the Win32 format will be pretty much obsolete anyway.
Alimonster
30-01-2004, 01:42 PM
Any news on whether the generics will spill over from D8 for .Net into D8 for Win32? (Oh please oh please oh please...)
Clootie
30-01-2004, 05:21 PM
Managed code is faster than Java, I think about 90% native code speed. (And yes, that statistic was made up on-the-spot. Sue me. ) You can make games in it, and I doubt you will notice much difference (except perhaps an increase in the speed of development.)
a) WRONG - "Managed code is faster than Java" - Sometimes Java do better sometime .Net do better - their are more or less equal from technology side - so speed is comparable too.
b) WRONG - about 90% native code speed - it's the AIM and marketting FUD. May be then 2.0 arrives with with enchanced profiler when... But now it's certainly not 90% of native speed.
c) WRONG - You can make games in it, and I doubt you will notice much difference - have you tried this, have you seen any single real GAME made with .NET? Probably in 3-5 years will see some (definetly not all) games will use .NET technology in SOME codebase as this is done in Delphi8.NET IDE for example.
After Windows Longhorn is released the Win32 format will be pretty much obsolete anyway.
WRONG - Nobody will just jump in .NET bandwagon. On mine work we still use Win16 applications (althow 32bit version got available some years ago) - it can be bad example, but still shows what transition to .NET will take a long time. Dont' forget what system low level code/drivers will be native anyway.
In addition:
a) Performance comparision of different compilers can be found here: http://osnews.com/story.php?news_id=5602
b) Floating point performance in .NET is AWFULL because of unneeded checks and SSE is not used anyway. So in 3D games with a LOT of data .NET performance will be times slower when native.
c) "Preview" of VER160 Win32 compiler (aka Delphi8.Win32) used to build Delphi8.NET itself can be found here:
http://codecentral.borland.com/codecentral/ccweb.exe/listing?id=21333 :P :P :roll:
Managed code is almost as fast as native code (Microsoft says that c# is ~10% slower than C++).
In some math, string, list operations test managed code was 2x slower than code compiled in D7 :( If .NET programs run on VM then it is not any suprise for me.
The string & list stuff doesnt suprise me too much.
If you just port a bunch of string manipulation code to .NET it will be slow, because string are immutable.
This means a single change causes the entire string to be reallocated.
As for the list stuff, dynamica array have to be recreated then the old contents copied to the new. And boxing isnt hidiously efficent either
Heheh, I don't agree with You. Nobody has seriously tested .NET in such things like physics or graphics where performance is really needed.
Only real way is to test & find out. But the performance comparision of different compilers Clootie linked to showed that the various .NET langauges werent that bad.
Clootie
31-01-2004, 12:21 AM
If you just port a bunch of string manipulation code to .NET it will be slow, because string are immutable.
As for the list stuff, dynamica array have to be recreated then the old contents copied to the new. And boxing isnt hidiously efficent either
For String manipulation there is StringBuilder class.
But array bound are checked each time an array element is retrieved (even then it's not dynamic from Delphi point of view) - so perfomance hit is guarantied.
Goliatus
31-01-2004, 10:16 AM
.NET will find use in games but only as some kind of a scripting language. Java was used in that way in Chrome Game(polish production :D).
noeska
31-01-2004, 10:25 AM
.net programs are a great thing for webdevelopement. Allowing more powerfull web-driven applications without sacrificing security.
On the desktop the use of .net will first be for office use applications where raw speed is not a direct issue, but security and data sharing is.
Powered by vBulletin® Version 4.2.5 Copyright © 2024 vBulletin Solutions Inc. All rights reserved.