Please post your specs and framerate

#1 A friendly word of caution here...

Nothing is strange on T2D at the moment. I keep saying it but T2D is in EA and doing performance testing trying to "figure things out" is pretty pointless.

So much is changing and so much has yet to change that your results won't really serve much purpose. As soon as the next update comes out, your results won't mean anything.

There is no magic limit; it all depends on how you setup your game. You need to understand how some of the scenegraph parameters work first. David Grace found that the old defaults did not suit his CloudBurst demo and changed the default bin-size which changed his performance drastically. In the shooter-demo, we constantly create/destroy the projectiles and this takes way too much time when things get hectic; they should be reused. The demo was written to show how quickly it is to just get to the "it's working" stage and to also serve as a tutorial.

The demos have had no performance tuning at all. Perhaps that's a mistake but we've constantly said how quickly they were put together so doing performance comparisons knowing that, doesn't make sense. It'll only show how much the demos need more work than just 24 hours. Packaging them up took as long as developing the code.

They are certainly not glossy high-performance game demos, we're just trying to show how easy it is to get stuff running, not how fantastic performance you can get. Maybe that's not come across very well but that's the way it is. We'll definately be doing that in later demos and spending much more time to show what T2D can do. We'll try to spend more than a day doing so. ;)

It is your time to spend though. ;)

- Melv.

#2 Maybe it was a mistake by us for not spending enough time getting those demos ready but the result is that they are not the top performers and people do tend to assume that they are the best that T2D can do.

We'll try to address this issue but eventually, T2D will show more than just those quicky demos. :)

- Melv.

#3 I am getting 115-149 FPS (FRAPS stat) on the Fish demo. Prescott 2.4, 1 gig of Hynix dual channel memory, 5900 Nvidia. The human eye can't see more than 60 fps anyway for practical purposes. We may be able to "sense" things up to 1/200th but that's not really an issue.

Edit add: @Robert

Quote:
My minimum spec is actually 400mhz-500mhz and a Geforce2 MX. This is the basic requirement games should run at as defined by the portals such as Real Arcade, Big Fish games, Reflexive Entertainment and so on... so I am anxious to get it working at about 30fps on those machines.

I think that could be realistic :) I don't do physics really, but I do a fair bit of scripting.

-Rob

In all due respect. That hardware is obsolete and the video card is not even fully DX 8.0 / 9.0 API compliant. Neither is it DX 9.0 certified. I understand that a 2d game should take less "beefy" pc. But people that own pc's like that don't spend big bucks on games anyway. Your main game client (in U.S.) is male 20-45 years old. And they are gonna have some hot machine with big 400 watt power supply in most cases. The Halo 2 mile long midnight bodysurfing crowd. Maybe things are different around here where I live but I would even be afraid to show my game on say a P4 1.6 with a 5200. My friends would laugh me out of the building.

#4 Oh and I was able to make my machine cry with torque t2. On Jedi Commander I put in about 200 droid enemy ships and a couple big enemy cruisers and T2d crashed and burned down to like 5 frames per second hahah. (Just a funny experiment).

#5 Not necessarily a demo, but I made a script that brought my 2.3 ghz machine with a Radeon 9700 to its knees. It took 7 seconds for the console to open/close after I hit '.

Granted it was rendering 2^15 + 1 sprites...

#6 Robert if you want to try slowing things down, open up the mainScreenGui file, replace the existing background bitmap with a new one, a really small one, then set it to tile.

I'd be curious to see if that slows things down at all. I found accidently last night that it practicly halves my FPS when I do that.

#7 I would love to see batching in t2d so there is just one texture state change and one draw call per different texture, this would speed up the engine no end.

#8 A special rendering layer for T2D which includes batching is planned but I didn't tell you that so keep it secret. ;)

- Melv.

EDIT: typo

#9 WOW! Now that is excellent news :) Remember, we didn't hear it from Melv *cough* :)

#10 We didn't hear what from Melv ? ;)

#11 Hehe ;)

#12 Now batching could be the answer - only a genius, Melv, could possibly cook up such a potential remedy and thanks to Timlon for bringing up the idea! If I'm not mistaken, that's exactly what some other high-performance, commercial quality 2d engines do to maintain smooth performance on low-end computers!:) Great thinking Melv!:) Trust me to keep a good secret!:)

#13 Heh advanced? Not really. Batching state changes is pretty fundamental actually. Not to say it isn't complicated and important -- it is, but its nothing groundbreaking.

#14 @Robert
Hum I don't see anywhere in my post that I took a shot at you like that

Quote:Are you even in the real world? you're talking as if T2D is going to sit next Doom3 on the shelves.

(no offense though :)

-Rob

I was addressing a subject not creating a personal attack. I didn't say "dude your pc is crap". etc.
And Yes I want T2D on the shelf next to Doom 3.

Ok your point makes sense try to keep on topic versus getting upset personally next time. Ok well you know you do have a great point and trying to make T2d games for lower end machines seems to make sense and for that I apologize. Just reading some of your other posts seems like you where complaining to GG and Melv about performance on what I would consider total crap hardware. But that's my professional opinion not personal. NOT TO YOU man I am talking about MX cards and stuff not people.

#15 There are a lot of GeForce2MXs out there and it would be a mistake to write off the people that have them. Ultimately what it comes down to is that you have to decide for yourself and for your project what your minimum spec will be and work towards that.

A puzzle game aimed at the casual market should run very well on that HW, but if your game is aimed at the hardcore gamer demographic then it's likely reasonable to aim higher.

#16 Thanks Rob.

... and all we need now is the time to get were we want to go.

Some of the issues here revolve around the fact that T2D started out to be something different from where it is now heading. As a personal project, I wasn't interested in market-sectors, it was just a fun bit of tech between meals. With the collaboration of GG as a partner and them having much more knowledge (and not nieve like me) about marketting, T2Ds development took a side-turn. We're still in the process of making a fundamental shift in how T2D gets things done.

The next few updates will highlight that with file-format, class-name and rendering changes as well as some more placeholder utilities. Things such as batching, software-rendering fallbacks, custom collision-responses, networking and other stuff wasn't exactly on my mind. The good thing was that the original T2D was well structured, commented and modular. This meant the drastic changes we've done so far to take it from where it originally was hasn't meant that it's ended-up in a hacked state.

In some ways, T2D can provide so much and can already enable the development of some pretty cool games. The other reality is that it's lacking in certain areas and that's where we're focusing but as with everything good and wholesome, it takes time.

There has been criticism of how/why T2D has/hasn't got said feature but the history of T2D explains that. I do respect that kind of criticism as it is essentially not wrong but the history of T2D cannot be ignored. From the time that GG took hold of the reins, we've made plans and come good on them without any iterative development or trial and error coding. Sure there has been delays but that's to be expected, particularly as I'm not fulltime (yet?).

The fact is that T2D is being worked on everyday but recently hasn't been getting that much closer to its goal. This is because the initial release has exposed lots of bugs and fixes and requirements for specific tutorials/tools and is to be expected. That's why we're so enthusiastic about having the community help out with tutorials and tools. We can focus on getting what everyone really wants out there.

We don't require people to follow our words like gods, we're just developers and our vision may not match everyones. We're very happy that you showed faith in purchasing T2D and for that we want to make what you've already purchased useful to you but the final thing we ask for is just a little time to get ourselves organised and get the rest of the features out.

If the fact that T2D seems complete right now is a problem then the coming months should prove to be even 'worse' because the feature-set is going to become even more complete. :)

Keep the faith everyone!

- Melv.

#17 @Melv: Very nicely put Melv!:) I'll help however I can to make the dream a reality!:) I can certainly wait for as long as it takes to get T2D where it is meant to be; a good lesson in patience to say the least. You are working hard enough as it is, perhaps its just about time for me to lend you a hand somehow, especially by allowing you and your team to focus on the important issues surrounding T2D and the expectations of the T2D community at large.

And as far as performance issues are concerned-I have just yesterday discovered that the makers of my GPU, eVGA, had issued a statement
regarding the latest Geforce4 MX 4000 drivers. The Opengl drivers are buggy and are currently being rewritten. There are workarounds to the problems however. The D3D drivers appear to work alot better in 16 bit mode while running the T2D demos. They are now running alot more smoothly, especially with the audio turned off. So, basically I guess what it actually boils down to, is a little tweaking to the driver settings and to the demos themselves. I am certain, that with the upcoming fixes you have in mind regarding T2D, especially batching, all of these issues will be resolved and will ultimatly lead to much more complex, smooth game play on obsolete hardware!:)

Do take care-you are a very valuable asset to the community and a pleasure to work with!:)

Gabor

#18 Well, Melv's already said it much better than me, but here's my two cents... I have been perfectly happy with T2D's performance, once I realized a few of the built-in parameters that can be adjusted for optimizations.

Needless to say, I still got a few complaints from some people about running my game under, say, Voodoo2 cards or other 3D hardware that is either ancient or not supported, or has a very buggy OpenGL implementation. The thing about T2D is that it's a 2D game engine -- based upon 3D hardware. There's going to be a minimum set of requirements for this, and people running very old systems might not have much luck.

As for the Geforce MX -- Cloudburst played just fine on my laptop, which is a 1.2GHz PIII (that actually runs at around 800 Mhz due to an overheating speed throttle problem) with a Geforce 440 Go video card. (Fairly close to the GF 4 MX.) I also gave the player several graphical details that they could enable/disable.

T2D is still EA, and as Melv has said, they're still trying to figure out what direction to take it. They are also working on the core bugs, and that's going to take a little while to work out. And that will rely a lot upon vigilance from the T2D SDK owners, as Melv and Josh can't test on every possible hardware/software platform.