RE: Future of the series (Full Version)

All Forums >> [New Releases from Matrix Games] >> Distant Worlds 1 Series



Message

Raap -> RE: Future of the series (2/8/2011 1:00:34 PM)

A couple of things:

1. I didn't load that image into photoshop, I was simply watching it from my browser and scrolling it up and down. That took over 30% of my CPU power to do smoothly.

2. If I'm doing that 200 AoE weapons and 100ships test, then yes, *the game is completely smooth if I'm looking somewhere else in the same system while the combat is happening*. It doesn't start to struggle until I switch my view over to the actual graphical effects.

That's sort of been my point all the time, that's my 'data' if you will. The game doesn't start to struggle until the actual graphics appear.



Kayoz -> RE: Future of the series (2/8/2011 2:59:55 PM)

1. So you're comparing performance of DW to Internet Exploder? One has nothing to do with the other. Or are you seriously stating that you consider it a realistic performance comparison???

2. You're missing the point - it's still inconclusive. When you aren't watching, the combat engine probably approximates a lot of events - it doesn't bother rotating, scaling and placing weapon fire or shield glows. Does that mean it's related to hardware acceleration? No - the inefficiency could come from anywhere. Did you repeat your test with the same set of ships, but no combat occurring, to see what sort of performance degradation occurs in the absence of combat? Did you test with and without AoE weapons? Did you test with and without shields? Your testing is far from being indicative that hardware acceleration would make a big difference. Perhaps you'd like to post your full test matrix and results?



Raap -> RE: Future of the series (2/8/2011 5:04:20 PM)

1. No, I'm not comparing the performance. I'm trying to show you how CPUs are generally poor at handling images/textures/sprites.

2. Well, you asked.

quote:

DW doesn't have to do any graphics work if you aren't viewing the combat, right? So if the whole combat is taking place around the sun, and you put the camera close up on some asteroid on the edge of the system - then ostensibly the performance should shoot up, right?

And yes, it does. It becomes completely smooth.

Don't get me wrong, I definitely see your point; it *could* come from a lot of things. But my experiences with the game really make me believe it's coming from the graphics. Take another example as well; if you late-game zoom all the way out when playing in the largest galaxy/number of stars, you'll even feel how sluggish the whole game becomes simply from the ship and system icons( and the background image) moving around. Zoom in to a system and it's once again completely smooth. Why is this? Zooming out changes nothing in the gameplay, there's still the same amount of AI calculations, pathfinding and whatnot...I simply can't see any other explanation for such a huge difference in performance.

For yet another example:
[img]http://img404.imageshack.us/img404/4100/lightningbolty.jpg[/img]
As you can see, the game is paused. It's also extremely early in the game, few ships and few colonized planets( I have two). Panning over that very sprite, despite the game being paused, is basically crushing things to a halt. It's otherwise perfectly smooth. How could that be if the CPU wasn't having huge problems drawing the graphics?



Kayoz -> RE: Future of the series (2/9/2011 11:29:25 AM)

quote:

How could that be if the CPU wasn't having huge problems drawing the graphics?


It could come from a great number of areas. If you aren't watching something happening, most developers will take shortcuts and approximate. Does it matter exactly who does what in one of the dozens of simultaneous pirate battles? Does it really matter in the long run if they approximate results of the combat, or run it second-by-second?

You might be right - I really don't know. But you're stating your assumptions and guesses as absolutes. You state that hardware acceleration WILL make DW run smoothly - but you have no evidence to support that assumption. Inefficiency could be in many places - and you're looking at causality where there's no definitive link.

Here's an example:

In 1949, Dr. Sandler claimed that polio was linked to hot weather and ice cream consumption. If you looked at polio infection rates, they followed the ice cream sales quite closely. So since most victims of polio were kids - then they logically assumed ice cream consumption and polio were linked.

Sound silly? Perhaps - but what they did - looking at the data and seeing a causal link - is exactly what you're doing.



tofudog -> RE: Future of the series (2/9/2011 12:24:32 PM)


@Kayoz
You confirm my desire to remain a code-noob[:D]. What a fascinating world of knowledge, but so ALIEN to my thinking.
Ask me anytime, when you want to upgrade your home to modern insulation standards, but my highest piece of computer knowledge might forever be:

THOU SHALT NOT SPLASH COFFEE ON THE COOLER-GRILL[:'(], NOT EVEN IF SAID COFFEE IS COLD.




Kayoz -> RE: Future of the series (2/9/2011 12:51:02 PM)

Fair enough, tofudog - my last experience with home insulation was when I was 5 years old, and my parents were replacing the pink fiberglass in the attic. Big pile of soft pink stuff - what 5-year old wouldn't have a great time jumping into it?

I haven't had any desire to go near the stuff since then...



tofudog -> RE: Future of the series (2/9/2011 1:23:34 PM)

[;)]Those were the days. When I was in training, one summer day I fell asleep in that stuff and was my boss angry.
Would not use it anymore than youŽd use a floppy disc though.



Raap -> RE: Future of the series (2/9/2011 2:28:36 PM)

quote:

ORIGINAL: Kayoz

quote:

How could that be if the CPU wasn't having huge problems drawing the graphics?


It could come from a great number of areas. If you aren't watching something happening, most developers will take shortcuts and approximate. Does it matter exactly who does what in one of the dozens of simultaneous pirate battles? Does it really matter in the long run if they approximate results of the combat, or run it second-by-second?

You might be right - I really don't know. But you're stating your assumptions and guesses as absolutes. You state that hardware acceleration WILL make DW run smoothly - but you have no evidence to support that assumption. Inefficiency could be in many places - and you're looking at causality where there's no definitive link.

Here's an example:

In 1949, Dr. Sandler claimed that polio was linked to hot weather and ice cream consumption. If you looked at polio infection rates, they followed the ice cream sales quite closely. So since most victims of polio were kids - then they logically assumed ice cream consumption and polio were linked.

Sound silly? Perhaps - but what they did - looking at the data and seeing a causal link - is exactly what you're doing.

I see what you're saying, and I'd probably be annoyed discussing with me seeing how stubborn I'm being on this subject.

I've seen no evidence of approximation in combat though; you can usually always notice those things, since the combat won't have proceeded as you'd expect. In most games you'll usually lose a lot more units than you would controlling it manually, and sometimes you'll lose less. That's not the case here, as when I send my ships on attack missions without watching them everything still goes as planned.

I also think it would be rather difficult to manage combat simulation in a real-time game where combats aren't a separate event. Not sure I can think of any games which do that, actually. All the games I recall with that feature are turn-based. Well, except for the X-series; those games do simulate the Out of System combats, but they also use separate sectors as 'levels'. And I expect it's as much from having to worry about pathfinding and collision detection as anything else. It's also *very* noticeable, as in combats don't go at all as you'd expect when in the system. Distant Worlds has an easier job there with the simple pathfinding and no collision detection, though.

Anyway, I suppose that would be something to ask the devs about, but it's also not really relevant to several of the other examples I brought up, like having poor performance when scolling over the larger sprites while having the game paused. So yeah, I'm not going to claim that global warming is caused by the dropping number of pirates in the world, even though they correlate very well, but I do believe GPU acceleration would have a huge impact on this game's performance.



Kayoz -> RE: Future of the series (2/9/2011 6:44:15 PM)

Unfortunately, DW has a lot of areas it could improve the efficiency of - so many that I'm hesitant to say that any one would make a significant difference. Can they improve it? Probably - they've probably learned enough lessons in writing it that they know where they've made mistakes - but correcting them might be more trouble than it's worth.

You're probably right that hardware acceleration could help - but my main issue is "how much?". Some of the game behaviour makes me cringe, as I have a pretty good idea what they've done wrong (or rather, they did it as they were supposed to according to documentation). But hey - new company, new product. We're human - we learn through making mistakes. Nah, you aren't stubborn - it's just that ASCII text isn't the greatest communication medium, unless you're willing to put in pages to explain something. I understand your position - I'm more concerned that it's a drop in the bucket.

I hope that they take the lessons on-board, and have a slicker next product. As an indie game, they've done quite well merely in getting something mostly stable to market. That puts them ahead of 90% of the crowd.



Raap -> RE: Future of the series (2/9/2011 8:01:20 PM)

Yeah, we'll find out how much they've learned with the next product. They've also mentioned better multi-threading support to come( with patches, I believe), which will hopefully end up having a noticeable impact.

Do you know approximately how much work one would be talking about to implement something like hardware acceleration though? Would it just be about switching out the drawing functions with their GPU-accelerated equivalents, or are we talking like weeks/months of rewriting? Guess I'm just sort of wondering why it hasn't already be done; if it's because it would require too much work and/or because they've already considered it and found that the benefit, or lack thereof, isn't worth the effort.



Kayoz -> RE: Future of the series (2/17/2011 5:34:42 AM)


quote:

ORIGINAL: Raap

They've also mentioned better multi-threading support to come( with patches, I believe)


Unfortunately this will probably require some serious re-engineering.

Think of single-threaded programs as one huge assembly line - they can get bigger and faster, but it's still only as fast as the slowest part.

Multi-threaded is like breaking that big assembly line into a bunch of little assembly lines. Instead of, for example, making the seats then the wheels and then bolting them to the car - multithreaded means making the wheels and seats at the same time, and bolting them to the car as they're done. With today's CPU architecture this means you can make more use of the multiple cores. But the architecture is quite a bit different. Sometimes you can have distinct worker threads tied together with events and semaphores - but it's not something that's going to happen with a patch or update.

quote:

ORIGINAL: Raap

Do you know approximately how much work one would be talking about to implement something like hardware acceleration though? Would it just be about switching out the drawing functions with their GPU-accelerated equivalents, or are we talking like weeks/months of rewriting? Guess I'm just sort of wondering why it hasn't already be done; if it's because it would require too much work and/or because they've already considered it and found that the benefit, or lack thereof, isn't worth the effort.


Unfortunately, the answer to that is the same as "how long is a piece of string?" - depending on their code, it may be a few man-days to a few man-years. It all depends on whether or not their current graphics engine supports hw accel, and if not, how similar the data they have is to what's required by one that does.

But a rough guess - I'd say 1-2 man-months of dev time to switch to a library that supports hardware acceleration. Then about the same to QA it. You have to remember that this would impact on every part of the program, bar none.

So long as they don't recycle too much of their old code, it shouldn't be a problem with the next version - but current DW - I wouldn't expect them to add it in. Too little bang for the buck, and too risky.



Webbco -> RE: Future of the series (3/10/2011 11:56:24 PM)

Just thought I'd bump this thread to maybe find out from Elliot or Erik what they have planned for the future? Could you guys give us a bit of a teaser for things to come?



Page: <<   < prev  1 2 3 4 [5]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
0.5957031