Star Swarm Benchmark Will Make Your PC Weep

[Andy Chalk]

Star Swarm Benchmark Will Make Your PC Weep

​

The Star Swarm benchmark uses Oxide Games' Nitrous engine to render a glorious multi-thousand-ship battle in space that will bring your PC to its knees.

The Nitrous engine put on display by Oxide Games and Stardock last month is noteworthy for its ability to render and control multiple thousands of on-screen units simultaneously, enabling the creation of battle scenes on a scale never before seen. Of course, that sort of thing requires a certain amount of processing power to get done, and thanks to the recently-released Star Swarm stress test/benchmark, you may now quickly and easily determine that you don't have it.

To be clear about it, this isn't a game, it's a demo that sets two huge, AI-controlled space fleets against one another in a real-time battle. It requires a DirectX 11-compatible video card and will also support AMD's Mantle when it becomes available, and offers multiple scenarios and visual settings from "low" to "extreme."

Because it's not a "deterministic simulation," which is to say that everything is being computed in real-time, the results of each benchmark will vary somewhat from run to run, but it's enough to give you a good idea of how close to (or far from) the bleeding edge you are. My own aging-but-still-viable rig, a Core 2 Quad backed by a GTX570, pulled down an average of 41.1 FPS at low settings but managed only 15.99 at extreme and several times bogged down to lowly, lonely 1 FPS. That's not a true mark, since I haven't rebooted in days and had a bunch of stuff running in the background, but the environment was equivalent and it was a fun show to watch, and that's good enough for me.

If you want to give it a rip yourself, you can grab the free Star Swarm stress test from Steam [http://store.steampowered.com/app/267130/]. Post your results in the forums and see where you stand!


Permalink

 

[TiberiusEsuriens]

I find it funny that the benchmark exists right now for the public, because as stated the Nitrous engine is specifically designed to take advantage of AMD's Mantle. This not only means that we need an AMD processor to ideally use this, but since Mantle isn't even publicly available yet the benchmark will ALWAYS return non-ideal results.

 

[Xan Krieger]

I don't even meet the minimum requirements since it won't run on windows XP. I'd upgrade but I don't know where to get a free copy of windows 7.

 

[SNCommand]

I find it funny that the benchmark exists right now for the public, because as stated the Nitrous engine is specifically designed to take advantage of AMD's Mantle. This not only means that we need an AMD processor to ideally use this, but since Mantle isn't even publicly available yet the benchmark will ALWAYS return non-ideal results.

AMD just released public beta drivers which enables mantle support for Battlefield 4 and Swarm War, users are already reporting an increase between 10% and 40% fps

 

[TiberiusEsuriens]

I find it funny that the benchmark exists right now for the public, because as stated the Nitrous engine is specifically designed to take advantage of AMD's Mantle. This not only means that we need an AMD processor to ideally use this, but since Mantle isn't even publicly available yet the benchmark will ALWAYS return non-ideal results.

AMD just released public beta drivers which enables mantle support for Battlefield 4 and Swarm War, users are already reporting an increase between 10% and 40% fps

That's pretty cool, didn't realize they had those out. It still doesn't change the fact that they're pretty much saying, "If you want to play these new games you need to buy new hardware," though. I know that is the entire point, but I'm a fan of making hardware as generic as possible in order to make software more universal.

 

[oldtaku]

It doesn't even look all that good compared to space combat sims, even on Extreme setting, but I guess that's the price you pay for 6000 ships onscreen at once.

 

[truckspond]

My i7 should be able to handle the CPU side but the whimpy GPU will most likely not even run it on low

 

[RicoADF]

Hmm I'll give my new laptop a run on this:
i7 4700HQ
16GB 1600mhz RAM
1TB HDD
Nvidia Geforce 750m

My desktop will have to wait until I install the new 780ti.... actually bugga it I'll install it now and find out.

 

[Kaymish]

well im on an AMD FX-8320 with a sapphire HD 7970oc and in some places i was getting as low as 9.8 FPS on extreme settings and only getting up to about 70 or so max FPS averaging out at around 38.4FPS though to be fair i loaded it onto my SSD i have all my games bar the ones with the most sluggish load times on my regular HDD but im not sure how much difference it makes i couldn't be bothered putting a copy on my HDD

 

[Jaylo69]

I don't even meet the minimum requirements since it won't run on windows XP. I'd upgrade but I don't know where to get a free copy of windows 7.

People still play games on W7? What the fuck?

 

[Strazdas]

you know who else simulates thousands of spaceships fighting in real time? Eve online universe [http://www.escapistmagazine.com/news/view/131799-EVE-Online-Mega-Battle-Breaks-300-000-In-Real-World-Losses]. Also if the video is accurate to judge from eve use more complex graphics too. does not use shields but does have GPU-breaking (true story, i actually burned my laptops GPU by being in a place that had 200 bubbles) bubbles of warp dampening.
So i dont see why would it break benchmarks unless its terribly optimized.

Btw, stalker Clear Sky benchmark at 1080p, max settings and A-tested AA on manages only 40 on a 770... a game from 2008...

thats probably because Clear Sky has as many bugs as the code has lines and it probably bogs down on performance. It is one of two games that i didnt manage to finish solely based on just how badly the crashes were (imagine saving the game only to find out the game crashes after 70 seconds from loading, regardless of what you do). and thats 4 times in a row somewhere above half of campaign finished ( i looked at a walkthrough to determine how far in i got). damn that game was fun but buggy as hell

People still play games on W7? What the fuck?

Whats the problem? W7 is the best option for gamers right now as XP is no longer supported due to diretx and windows 8 is uncomfortable as hell for anything else. Yes W8 is faster at its core but thats completely useless of thats all it is. i could understand your reaction if he said he played it on Vista, which is a truly "who the hell still uses that" deal, but win 7 is the most popular OS at the moment. and it still is the best selling one.

 

[WouldYouKindly]

I find it funny that the benchmark exists right now for the public, because as stated the Nitrous engine is specifically designed to take advantage of AMD's Mantle. This not only means that we need an AMD processor to ideally use this, but since Mantle isn't even publicly available yet the benchmark will ALWAYS return non-ideal results.

AMD just released public beta drivers which enables mantle support for Battlefield 4 and Swarm War, users are already reporting an increase between 10% and 40% fps

That's pretty cool, didn't realize they had those out. It still doesn't change the fact that they're pretty much saying, "If you want to play these new games you need to buy new hardware," though. I know that is the entire point, but I'm a fan of making hardware as generic as possible in order to make software more universal.

It depends on the level of detail you want. His machine is hardly a super computer and it did quite well at low settings.

 

[Caiphus]

i7 4700MQ
Nvidia GTX 780M
16GB 1600 mhz RAM
1TB 5400 rpm HDD
Windows 8.1

Just had a go at the low settings, some university stuff running in the background, nothing major. Averaged 72.08 FPS, using the Attract Scenario. Occasionally got pushed down to just under 20 FPS. Sooo yeah. Fun.

Edit:

Just got an average of 19.04 on extreme settings. The lowest I saw while peering up from my books was ~4.5.

 

[Kahani]

you know who else simulates thousands of spaceships fighting in real time? Eve online universe [http://www.escapistmagazine.com/news/view/131799-EVE-Online-Mega-Battle-Breaks-300-000-In-Real-World-Losses]. Also if the video is accurate to judge from eve use more complex graphics too. does not use shields but does have GPU-breaking (true story, i actually burned my laptops GPU by being in a place that had 200 bubbles) bubbles of warp dampening.
So i dont see why would it break benchmarks unless its terribly optimized.

The main difference is that most of Eve takes place on CCP's servers, and even then they need to slow the game down by a factor of 10 to prevent it from breaking down entirely during big fights. Having not just the rendering, but also all the calculations involved in every part of control and movement of ships all taking place on a single home PC is quite a big step up.

AI War: Fleet Command can have similar numbers of ships in action, but that is 2D with considerably worse graphics so probably not really comparable either.

 

[Strazdas]

you know who else simulates thousands of spaceships fighting in real time? Eve online universe [http://www.escapistmagazine.com/news/view/131799-EVE-Online-Mega-Battle-Breaks-300-000-In-Real-World-Losses]. Also if the video is accurate to judge from eve use more complex graphics too. does not use shields but does have GPU-breaking (true story, i actually burned my laptops GPU by being in a place that had 200 bubbles) bubbles of warp dampening.
So i dont see why would it break benchmarks unless its terribly optimized.

The main difference is that most of Eve takes place on CCP's servers, and even then they need to slow the game down by a factor of 10 to prevent it from breaking down entirely during big fights. Having not just the rendering, but also all the calculations involved in every part of control and movement of ships all taking place on a single home PC is quite a big step up.

AI War: Fleet Command can have similar numbers of ships in action, but that is 2D with considerably worse graphics so probably not really comparable either.

True, the simulation is done on the server for much of the calculation, but we are talking about graphical benchamarking of being able to draw thousands of ships on screen engaged in a fight in real time, and not the actual fight calcualtions, which is handled by your CPU anyway. so in this regard the GPU load in eve is no less than that of Star Swarm, in fact i woudl argue it is more as the models are more detailed (asuming you got draw distance set to high so they all load in eve battles, as you can actually make them invisible after certain distance to make these battles palyable on low end pcs).
The actual fight calcualtion isnt that huge of a deal actually. back in the days of sprite based games we had hundreds of soldiers fighting in games like cossacks on Pentium I. granted spaceships fighting will take more calculation to simulate but were not running 20 year old PCs either.
another game that did this was Supreme Commander and used actual 3D models, though granted once the fight numbers reached a thousand the lag was really rising to unbearable.

What really is a big step up here is not the Star Swarm but the fact that PC hardware have went so far as to be able to do this.

 

[Kahani]

True, the simulation is done on the server for much of the calculation, but we are talking about graphical benchamarking of being able to draw thousands of ships on screen engaged in a fight in real time, and not the actual fight calcualtions, which is handled by your CPU anyway.

No we're not. See the article:

The Nitrous engine put on display by Oxide Games and Stardock last month is noteworthy for its ability to render and control multiple thousands of on-screen units simultaneously...

Because it's not a "deterministic simulation," which is to say that everything is being computed in real-time

Yes, it's on the CPU. The CPU is there for a reason, and quite an important one.

The actual fight calcualtion isnt that huge of a deal actually. back in the days of sprite based games we had hundreds of soldiers fighting in games like cossacks on Pentium I.

Firstly, hundreds is not at all the same as thousands. Secondly, there's no sensible way to make a comparison without looking at what actually gets calculated in the different games. A 2D simulation of a relatively small number of actors with a very basic instruction set is not at all the same as a 3D simulation of a much larger number with more complex AI. 3D doesn't just mean models, it's a whole extra dimension for things to move in which greatly increases complexity. More actors can increase complexity exponentially depending on how they need to interact with each other, and of course if you make the AI complex enough even a single ship could be enough to make any computer grind to a halt. You can't simply dismiss all these factors by saying "it's on the CPU". Graphics is often the limiting factor these days, but if you manage to get around that it's trivially easy to max out CPU load, and figuring out how to get more out of it would be a big deal.