Advertisement Jump to content
Sign in to follow this  
proanim

system requirements

This topic is 1882 days old which is more than the 365 day threshold we allow for new replies. Please post a new topic.

If you intended to correct an error in the post then please contact us.

Recommended Posts

I am looking to create a game, and as a test machine I would use my own pc. It is fairly  average pc, that can run most games to date. How can I calculate how much ram I would need for my game, without sacrificing any of the game assets. For example if I have characters that have tons of detail in textures and detailed environment how can I be sure it doesn't exceed my memory limit. Are these things known in advance? or, are game assets corrected to fit in memory later?

Share this post


Link to post
Share on other sites
Advertisement

My suggestion for starters is: just don't worry about it. You probably won't even half a GiB.

More involved answer.

RAM estimation is based on your system. For a start, you need to load at least once all the textures and all the models.

Maybe some models will be CPU-vertex-skinned. Then you need an extra model copy for each interpolated version of each animated model.

Or maybe you have particle systems, each will have its own particle buffer, with each particle being (ex: float3, float4) and N particles for which you have a total buffer size of N*sizeof(float)*7. Maybe those are only in GPU memory.

And then you have physics. In my experience it's fairly compact representation, I'm pretty sure it takes less than 256k for me (only including level data representation).

 

In general, when the game is designed, the requirements are set in the design document. It is not best practice to cut them later, but it still happens. Sometimes this can be done even with no quality loss - for example, one might cut in half a texture which is only seen from a distance.

 

As a first step, you might elaborate on your definitions of "tons of detail in textures and detailed environment". I see this got an "IF", so it's only supposition. You want to care about the dataset you're going to use right now or ... at worse... a year from now. Not the AAA asset from the game you'll be producing 10 years from now (if you're extremely lucky).

Share this post


Link to post
Share on other sites

Yeah dont worry, as soon as you exceed the available memory you will notice from either an error from a function not working, an error message or a slowdown (when the system starts to get into an endless cycle of swaping pages in and out and the HDD cant keep up). Thats why its even a good thing to always test on a worse PC.

Just dont waste memory, only use what you need and you will have no problem unless you load entire movies into main memory.

Share this post


Link to post
Share on other sites

Generally system requirements are made for financial rather than technical reasons.

 

You need to be able to sell your game X number of times to break even. You know Y people have systems that are at least a certain level of power, you know Z people have systems that are at least a certain other level of power. You would base your target requirements based on the size of the audience you generally need to sell to in order to make money.

 

For indie games you generally aren't pushing any limits and you generally don't support varying rendering quality (no high/medium/low detail). In general you should probably pick a reasonable minimum requirement spec and use that as your test case.

 

For high performance games there is a great deal of work done up front to assess feasibility of your game. Each member of the team will work out the memory/performance/etc. requirements of their areas, usually using past games. If it's too difficult to assess, or there is no basis for comparison (i.e. a company's first game or a new engine), it is important to prototype all risky features to the level that you can accurately enough estimate your requirements. From these metrics you can scale your requirements to the estimated full requirements of the game.

 

Once the leads are convinced of technical feasibility only then will (or at least should) the team move forward, otherwise some sort of compromise needs to be made in order to fit in your chosen specs (either cut features, or change your target specs, or R&D how you can make it work).

Share this post


Link to post
Share on other sites
Sign in to follow this  

  • Advertisement
×

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!