Suva 3D Engine. Pure C# ...with Speedtree!

[jfberge]

this looks interesting. I've been wanting Speedtree for .NET for a while now, and this engine has it integrated. It also has a terrain generator, physics (ODE), sound, networking, shader 2.0,  and the usual stuff.
No prices are mentioned, however, so it'll probably be out of my budget.

[JeffWeber]

Suva 3D is a managed .NET game engine for software titles on the PC and next generation XBOX platform.  Written entirely in C# and Managed DirectX

As far as I was aware, XBOX 360 isn't going to support Managed DirectX.  Has anyone heard that it will?  Officially!

[Yukito]

Well... i havent heard about it, but afaik we have currently no REAL managed DX for Windows either until WGF. Currently it is only a wrapper(?).

[billythekid]

nvm

[potato]

As far as I was aware, XBOX 360 isn't going to support Managed DirectX. Has anyone heard that it will? Officially!

That would be weird. Why wouldn't MS support the tech that they've spent so much time and effort pushing?

[JeffWeber]

That would be weird. Why wouldn't MS support the tech that they've spent so much time and effort pushing?

I think it's because the game dev community has not moved far enough in that direction yet.  I am guessing that the next version will support it... Who knows though.

[Borys Pomianek]

Maby becouse managed directX would be a step backwards in terms of speed ?

You really dont need that much speed when creating apps like Word, Exel etc.

But when you make games you really need the speed.

BP

[billythekid]

Managed DX is only slightly slower than unmanaged DX and that has to do more with the fact that managed DX is really just a wrapper for unmanaged DX than anything else.

Also the reason to go toward managed is because it reduces development time greatly. Hardware is so fast now and does so much for you that the software engineering focus is shifting for games and other types of programs. Software engineers are starting to care more about development time, reliability, security, etc than efficiency now. Efficiency is still important of course but for most things making things run as fast as possible isnt necessary. And that is starting to include games.

If I had to put money on it, Id say XNA or WGF or whatever (new DX) will have a managed component. And the whole point of XNA is so that if you develop on the PC, the same code will work on XBox. But sorry no official source for that.

[Borys Pomianek]

Well everybody says that everything is so fast but i have a "more than avarage" pc and i see a lot of slow games on it.

I never actually used any managed language or managed DX

I like to know what i do with my memory and other things.

BP

[billythekid]

If you have no knowledge of managed languages then why do you comment on them? That would be like me talking about the mating habits of cockroaches even though I know nothing about the mating habits of cockroaches.

[Borys Pomianek]

If you have no knowledge of managed languages then why do you comment on them? That would be like me talking about the mating habits of cockroaches even though I know nothing about the mating habits of cockroaches.

Becouse i dont have to know mating habits of cockroaches to know that they are animals and to know how animals use to behave.

I know how managed languages work i just dont use them.

Thats why i dont comment but ask, in case you dont saw the "?" at the end of the line.

BP

[gpx]

The problem with Managed DirectX is that it is easy to kill performance using standard .NET coding practices so normally your first attempt will suck (and unfortunately most people just give up there).  For example:

[1]
Your choice of game loop is critical (not so much in C++).

[2]
Efficient memory usage is still important with .NET (easily overlooked).

[3]
Automatic boxing/unboxing will sap performance and is also something you need to be aware of when writing managed games but is generally ignored when using .NET in general (same applies to event handlers).

So, managed games can be pretty efficient (90% +) PROVIDED you follow a few simple rules but unfortunately they aren't exactly intuitive and go against the general idea of .NET programming.  In a nutshell, .NET doesn't obviate the need for efficient programming and thinking it will causes all sorts of problems.

[Mutos]

Hi gpx, hi all,

[1]
Your choice of game loop is critical (not so much in C++).

[2]
Efficient memory usage is still important with .NET (easily overlooked).

[3]
Automatic boxing/unboxing will sap performance and is also something you need to be aware of when writing managed games but is generally ignored when using .NET in general (same applies to event handlers).

I may be a total moron but you either said too little or too much :lol: I understand you must be true but what do you mean exactly by, for instance "choice of game loop" or "Efficient memory usage". What I was expecting from the first sentence of your post was some precise do's and dont's, rather than generic ideas :twisted:

[billythekid]

gpx - Thats very true. Managed languages have a different set of performance pitfalls (and similar too like memory management) compared with unmanaged languages. However I dont really see that as a "problem". Just get yourself educated. I mean its quite easy to be a really crappy unmanaged programmer aswell if you dont watch out for the performance pitfalls.

[Borys Pomianek]

gpx - Thats very true. Managed languages have a different set of performance pitfalls (and similar too like memory management) compared with unmanaged languages. However I dont really see that as a "problem". Just get yourself educated. I mean its quite easy to be a really crappy unmanaged programmer aswell if you dont watch out for the performance pitfalls.

Yeah but if someones is efficient in a not .net language he or she may not want to start educating them selves again.

BP

[jfberge]

I think that as games become ever more complex and the .NET framework becomes more efficient, we'll eventually see a move to C#. There is a large inertia going for C++ in most application development, as it's been the predominant language for decades. C# has only been around for a few years now. As new coders emerge from college and cut their teeth on things like C#, I think that industry may begin to see it as a viable option for "serious" applications. This will be bolstered by the tighter integration of .NET with newer versions of Windows and the new Avalon subsystem. The migration from C++ to .NET looks to me much like the migration from assembly language to C. Sure, there's a tradeoff in terms of speed and efficiency, but there's a disproportionate gain in terms of development time and code managability. This will all take time, but I don't think it's silly to imagine that in a decade, managed languages (maybe Java, maybe C#) will become heirs to the throne for application development. You'll always have C++ for embedded systems and device drivers, but such low level control will become less relevant in a world of faster, multi-processor machines.
For those looking for a good reference on writing fast managed code, this article does a good job of diagramming the cost of various procedures in .NET.

[JeffWeber]

@jfberge

Well said!  I agree completly.

[darqSHADOW]

Managed languages have already started to make it into the mainstream.  I am a Managed programmer at my work, even.  I was solely responsible for building a huge document creation system from the ground up, the only thing was I had to use one C++ calculations library.  I wrote the entire web-system in ASP.Net (VB.Net), and built 3 different distributed daemons to handle the document prep and delivery.  With over 50 people on my system constantly, and thousands of completed doc packs (500+ pages each, with 5000+ filled info fields) I can generate the docs in seconds (average of 3 seconds).  This is all while running thousands of XML rules to choose the documents, then running thousands of complex calculations to create the data -- all live, on the fly, and all in managed languages.

This entire system was written from the ground up in less than a year, by a single person (me).  Had I been forced to use ASP or PHP, and C++ this project would have taken 3 years EASY to complete, and would have been harder to maintain, and harder to debug bottlenecks.

Now, someone want to try and explain why managed languages are not viable in the market?  

In the past few years, I've written many different things in VB.Net -- and I have no formal training in VB or VB.Net -- not a single class.  I had zero problems going from C and C++ into VB and VB.Net -- its just what programmers do, we evolve as the market does.  In the past 15 years that I've been programming, I've had to learn over 30 programming languages, some of which survived, some of which did not.  I, however, do not see the managed languages going away at all, nor do I see why some people do not see them as viable gaming platforms.

DS

[Borys Pomianek]

Managed languages have already started to make it into the mainstream.  I am a Managed programmer at my work, even.  I was solely responsible for building a huge document creation system from the ground up, the only thing was I had to use one C++ calculations library.  I wrote the entire web-system in ASP.Net (VB.Net), and built 3 different distributed daemons to handle the document prep and delivery.  With over 50 people on my system constantly, and thousands of completed doc packs (500+ pages each, with 5000+ filled info fields) I can generate the docs in seconds (average of 3 seconds).  This is all while running thousands of XML rules to choose the documents, then running thousands of complex calculations to create the data -- all live, on the fly, and all in managed languages.

This entire system was written from the ground up in less than a year, by a single person (me).  Had I been forced to use ASP or PHP, and C++ this project would have taken 3 years EASY to complete, and would have been harder to maintain, and harder to debug bottlenecks.

Now, someone want to try and explain why managed languages are not viable in the market?  

In the past few years, I've written many different things in VB.Net -- and I have no formal training in VB or VB.Net -- not a single class.  I had zero problems going from C and C++ into VB and VB.Net -- its just what programmers do, we evolve as the market does.  In the past 15 years that I've been programming, I've had to learn over 30 programming languages, some of which survived, some of which did not.  I, however, do not see the managed languages going away at all, nor do I see why some people do not see them as viable gaming platforms.

DS

quite a good answer.

Still you for some reason make TV 6.5 using c++ not c# or vb.net  

For instance i would like to use the .net platform but i dont want to pay for the whole .net thing and i dont want to scrap things iam lerning today and buy completly new books to start using .net.

I see a biig step forward when you think about typical programs, word like programs or the thing you did but some stuff changes a lot slower.

We still use ASM in many areas even when the hardware gets faster and faster.

Olso ofcourse in 10 yeers we may use c++ as we use asm today and .net as we use c++ kind of level languages but imho not actually today.

using managed languages like java for instance requires a lot diffrent way of thinking and time will have to pass before people start using it as it should be used.

for an example, i started with java, i was writing some stupid example rewries and doing other kind of stuff and then i started lerning c++. Now i am very happy that i know c++ ( i dont program actively right now becouse i dont have time ) becouse it made me think a lot diffrent.
I would love to be efficient in ASM and better in c++ becouse that way i could create whatever i wanted.
imho .net is aimed more towards the big market and it does not give anything special to a hobbyst programmer.

I would like to create my own webbrowser for instance in c++. In Java you can just use a library for the whole html thing, create a simple gui and voila you got it done.
Still it works like hell, and iam quite sure that if i would wrote that in c++ it would work as i want and not as the lib autorhs wanted it to work.

Ofcourse if i would be a java specialist i would not have that kind of a problem but iam not a java specialist and i think that .net is just not for me just like java is not good for my needs.

But i will learn those better when i have time.

BP

[jlucard]

Borys, I used to be a pretty good asm coder. I have written asm code for x86 and 68000 processors, as well as several microcontrollers. I am also a C++ programmer having 5 years of work experience with this language. Why am I mentioning this?

Because I use mainly two languages now. Delphi and C#. You see the problem is I haven't got an infinite amount of time at my disposal. Using .net I can make an app in almost no time at all. A project that could take a few months to complete using such a language would require years of work using pure asm (which is preciselly DS point). Using asm would be faster I hear you say. So what. If by the time the app is out it is obsolete that would be pointless. And what about portability. Asm code is hardware specific. Porting an asm program into another platform can take ages. Furthermore using asm you would need to program a bigger part of the code yourself (instead of using libraries provided by a vendor). That means you have much more code to debug and test, and that is code that is difficult to read, thus an increased likelyhood there will be bugs in your app. I am not sure where you got the idea that asm is the magic wand that could turn lead to gold or something, but this is not the case.

No you are wrong about this subject. Levels of abstraction are good. Using high level  languages is both appropriate for bussiness and for small indipendent developers alike. As an independent developer my first choise of tools would include one of them visual languages due to the fact I am not immortal. I want to finsih apps in a reasonable amount of time, and so do my potential customers. C++ is good as well, but I wouldn't use it with a development method such as prototyping, not in the windows platform anyway - I use a c++/qt combination when developing for linux. C++ is more appropriate for a bottom to top approach. As for asm... I might consider it only for optimising specific app parts (i.e. converting a loop from delphi to asm might make it somewhat more efficient). But generally speaking it hardly worths doing that sort of thing these days.

If you want my advice learn the new tools. Dot net would make you life easier, especially since in real life when you have real customers it is better to use a development method such as prototyping to ensure you have an appropriate requirments specification document. Showing a non working app interface to you target user could uncover potential misunderstandings and does create a feeling of them being involved in the process, thus reducing the possibility of resistance (a natural reaction of users when computerised systems are introduced to work environments previously operating using traditional pen and paper methods).

Sorry for the long message, DS had already presented some arguments supporting the .net case, however I thought it would be beneficial to elaborate on that matter a little bit more.