Build Your Own Arcade Controls Forum
Main => Main Forum => Topic started by: MikeQ on January 13, 2006, 08:15:17 pm
-
I'm aware of VCMAME but would have a solution that can compile MAME32 and MAME separate or together.
-
Question: If available, would you use Microsoft solutions for MAME32 and MAME?
Not in a million years.
-
Not sure what is the purpose of this survey.
VCmame could compile mame or mame32 (haven't used it an a while)
Are you questioning how good the VS compiler is?
What about the Intel compiler? Fastmame?
-
The purpose is to determine if people would like to have the ability to compile my version of mame with Microsoft compilers.
-
Like I said it has been a while since I have played with VCMAME. I seem to remember that there where a couple of things that where different on VCMAME.
VCMAME did not used to support the DRC for MIPS. ( so games like KI, where not playable at full speed using VCMame.)
Also I don't think it supported the assembly CPU cores. (There were issues with these so may not be a big deal).
There are probably more things I can't think of.
Good luck
-
Are you talking about differences from a performance standpoint or as a developer?
Personally, I use the standard mame compiling tools (mingw) when making a build that I'd use in my cabinet.
However, for development work on mame, I use visual studio 2005 because it's much more developer friendly than command line tools and a text editor. The debugger in particular makes it totally worth it.
As a side note, I believe it's good for the mame project to compile mame with as many different compilers as possible. Each compiler catches different errors and warns about different things and fixing as many of the errors and warnings as possible and submitting them back into the main source code base results in better code for everyone.
-
The purpose is to determine if people would like to have the ability to compile my version of mame with Microsoft compilers. If no one wants this, then I would just use the MINGW mechanism.
MS Visual Studio tools and Intel's ICC both cost an arm and a leg. 99.9999% of users are NOT going to have this software on hand, and are NOT going to pay money for it just to compile something free like MAME (or a spinoff/utility/front-end/whatever).
Stick with MinGW/GCC. With the upcoming GCC4.0 builds there will be enough speed improvements to negate the need for other expensive compilers and their 5% speed boosts.
-
MS Visual Studio tools and Intel's ICC both cost an arm and a leg. 99.9999% of users are NOT going to have this software on hand, and are NOT going to pay money for it just to compile something free like MAME (or a spinoff/utility/front-end/whatever).
Last time I checked VC 2005 Express was free.
-
MS Visual Studio tools and Intel's ICC both cost an arm and a leg. 99.9999% of users are NOT going to have this software on hand, and are NOT going to pay money for it just to compile something free like MAME (or a spinoff/utility/front-end/whatever).
Last time I checked VC 2005 Express was free.
VC 2005 is free. Not an arm or a leg.
I've gotten free Intel compilers before too. They use to have a 30 eval period but after 30 days the product still functioned. You just didn't get support.
-
The purpose is to determine if people would like to have the ability to compile my version of mame with Microsoft compilers. If no one wants this, then I would just use the MINGW mechanism.
MS Visual Studio tools and Intel's ICC both cost an arm and a leg. 99.9999% of users are NOT going to have this software on hand, and are NOT going to pay money for it just to compile something free like MAME (or a spinoff/utility/front-end/whatever).
Stick with MinGW/GCC. With the upcoming GCC4.0 builds there will be enough speed improvements to negate the need for other expensive compilers and their 5% speed boosts.
I'm not suggesting I'd only provide a Microsoft project only. I'd still proved the MinGW/GCC make file. The Microsoft project would be additional.
-
I would love an Visual Studio solution. So please go for it. :)
Actually what would be infinitely better is if mame could be rewritten in C++. The design would make more sense visually when looking at the code and would be much cleaner and easier to read/maintain/modify.
-
I would stick with mingw because
1) it is what mame uses
2) it is much easier to setup. So even a noob can compile a custom mame if needed
3) takes up a smaller footprint on harddrive
You don't gain that much by using VS except for debugging, which most people who compile mame for a custom compile don't need.
But if it is going to be additional then sure, go ahead.
-
I voted no, but SirP said it:
I would stick with mingw ...
But if it is going to be additional then sure, go ahead.
-
It would be additional.
The WPO option in Microsoft can buy you a lot too. I've Vtuned Mame and looked at the module percentages and Microsoft is routinely a faster build. I haven't gone as far as measuring FPS differences though.
The main reason however to provide the Microsoft solution however is debugging like people have said but also so that you can use all the nice features that 3rd party development tools allow. I use Incredibuild for "compiler farm" compiling (compile mame in under a minute). Vtune and NuMega tools, etc.. All these tools tightly integrate with Visual Studio.
-
I'm not suggesting I'd only provide a Microsoft project only. I'd still proved the MinGW/GCC make file. The Microsoft project would be additional.
Well in that case I don't mind either way. I'm 100% GNU myself (big stinking Linux hippy) and if the MS stuff is merely an addition I can ignore, then it affects me naught.
It's all down to how much effort you can be bothered putting in for each release. Although for pure acedemic reasons, I'd be very interested to see them run side by side for benchmarking purposes.
-
I'm not suggesting I'd only provide a Microsoft project only. I'd still proved the MinGW/GCC make file. The Microsoft project would be additional.
Well in that case I don't mind either way. I'm 100% GNU myself (big stinking Linux hippy) and if the MS stuff is merely an addition I can ignore, then it affects me naught.
It's all down to how much effort you can be bothered putting in for each release. Although for pure acedemic reasons, I'd be very interested to see them run side by side for benchmarking purposes.
Well, I'll work in Visual Studio so putting the solutions out won't be a problem.
We had some Linux hippies at work that were always singing the praises of Linux. When our company had to start supporting device drivers for Linux, these guys were assigned to do the work on Linux since they loved it. Now they can't stop complaining about how much better Windows is. :)