Performance issues.

I don't mind my constant 250+ fps... smooth in any dock or dogfight.

System:Intel i7 4790k @ 4.4ghz, Kingston Genesis 16GB 2400Mhz, Asus GTX970 4GB OC SFF Edition, OCZ 256GB SSD, Twin Viewsonic Monitors @ 1680x1050, Windows 7 Pro 64 Bit etc etc.Settings:Borderless, Vert Off, Rate Limit Off, Everything else on High
-
I think it's important to have a well balanced system..

If you don't mind me asking, what models are the viewsonic monitors?
 
Sitting inside the hangar in a polygon-type station, 50FPS steady. 50-60 when launched, 110+ outside. This is running 2715x1527 (3k Dynamic Super Resolution) with everything turned up to max and 1x anti-aliasing. I don't notice the drops below 60fps due to G-sync.

I did check video memory usage at this resolution. Currently the game is using 2270mb of VRam. I suspect it would be less at 1080P. If you're having frame-rate issues at 1920x1080, turn anti-aliasing down or off. The 660 is limited on memory bandwidth and total VRam. When you can, upgrade to anything newer within your budget. Cards like the R9 285 or GTX960 are under $200 USD and will be a massive upgrade from a base-version GTX660. If you do upgrade, consider if you intend to buy a variable refresh monitor in the next year also - on the AMD side, you'll have to be sure to get something which supports the newer Displayport protocol which allows variable refresh (Freesync) and you'll want to consider a card like the 290/290x, or stick to Nvidia and plan to spend a bit more for G-Sync. Checkout PCPER.com for more info on the latest monitor tech.

As far as my system - 4790K @ 4.4ghz all cores, EVGA 780ti Superclocked.
 
Last edited:
Thanks for the reply, great readup on the current trends.

However, it dawns on me that people simply don't read what one writes, and just get on with their own thing.

Again, the problem is not with the card (it runs all other games decently), nor with the game in general (space, RES are 60fps solid), but with this particular scenario that is simply put unoptimized and shabby.

At this point my only hope is the 'low level optimizations' advertised for 1.3
 
The game does gain more lag with each update and its not my pc, GTA5 runs a constant 60 frames and it has to be said a lot more fun.

Just before you tell me to clear off and go play it just wait because by the time you read this I already will be. :D
 
The game does gain more lag with each update and its not my pc, GTA5 runs a constant 60 frames and it has to be said a lot more fun.

Just before you tell me to clear off and go play it just wait because by the time you read this I already will be. :D

It's worth noting that games like Battlefield 4 and Hardline are heavily optimized for multi-threaded CPUs, whereas many games are not. BF4 is a very pretty game from a texture standpoint, but strongly doubt that the poly count in BF4 is as high as that of a spacestation in ED. I also wonder about the sheer number of light-sources inside a station. If every one of those landing pad lights is real and not just a texture, it would certainly put a strain on a GPU.

Also, from a pure development standpoint, Origin is a part of EA, a publicly traded company with a market cap of 18.54 billion as of today, and development spending for it's main headlining game is likely 10x that of a game like ED (or more). Compare that to $1.578 million pounds from the kickstarter plus game sales we don't know totals for. From my perspective, Frontier has done a pretty good job so far and it's likely to get better as time goes on.

I realize that if you're not able to upgrade, no amount of discussion is going to change that fact. Fundamentally, there's nothing wrong with a 660... it's a good card. I used to run a pair of the Ti versions in SLI and the performance was really decent but a single 780ti is overall better, for example, and a 970 is basically as fast as my 780ti until memory bandwidth becomes a factor (resolutions above 1080p or with 4x AA).

GTA5 on a 660 at 1080p with "normal textures" benchmarks at 42FPS average. A 970 at that setting would be 88 average. I personally don't consider ~40fps at 1080p as playable, especially for a FPS or action type game. In a flying sim like ED, the poor framerate simply reduces the sense of realism when dogfighting, but I'd expect it's playable.

http://www.techspot.com/review/991-gta-5-pc-benchmarks/page2.html
 
Last edited:
The game does gain more lag with each update and its not my pc,

Your right it does, I've noticed that. I'm sure the devs have too.
It all depends on how much they value optimisation over new content. they're not being lazy and they're definately not incompetent, they just don't have the resources to do both.

IMO :D
 
......The station is nowhere complex to modern games' complex......

Sorry but I think you may be greatly underestimating the complexity of producing a single object with massive view distances and still retaining high details (like paint chipped off grills inside the endcaps where very few people even go for example) at close distances involved in presenting the multi-lit and shadowed, moving station to you in a fully 3d environment.

Anything you cannot see in the station (or anywhere else in the game) will not being rendered but the object still has to have all of the complexity ready to be rendered at the right time. Suggesting

Most games obscure objects and the maps with other objects to reduce the amount of work involved (so that as little as possible actually has to be rendered). The station is effectively a map that you can see in its entirety - there are not many games that can do that and even fewer that do it without serious amounts lower resolution textures and details on objects at a distance. Maps are nearly always designed so that you are limited in the amount of detail you can see at any one time + they restrict your movement. ED does not have that luxury because you have free movement and a massive range difference.

Personally I think 30-45fps in a station on a gtx 660 is perfectly acceptable and if I were uncomfortable with how hot/hard my card and/or pc are running. I would look seriously into getting better airflow/cooling or turn down the graphics.
 
Last edited:
You miss my point again.

For all the things you've mentioned, there are optimizations known for years. Mipmapping, lods, frustum culling, etc. The details are made with textures, bumpmaps and so forth. It's not rocket science.
My problem with this particular issue is that some years ago (10) I wrote my own 3d engine and a paper on drawing optimization (quad/octrees, bsp, bounding volumes, etc), and that's why it's itching me, because I know that this could be done better.

The very core of my issue is this - In the closed bay i have 45fps, when I'm ejected to main deck of the station, 55fps, when I take off, it's 60.
If the station is so complex, why the framerate is lower in what is essentially a closed box? What is there to draw? Geometry wise, it's much harder to render a RES, or a minecraft world at a distance.

Of course I can (potentially) just buy a 970 and shut up. But then again, the company I paid money to, could launch a GPU profiler to see why the frametime renders are longer inside the goddamn bay. It's not a biggie for the engine guys, believe me.

What saddens me, is the level of denial and topic switching that I see here. Simple arguments such as "empty box - low fps"; "RES - high fps" don't seem to catch people's mind.
 
An empty box? My apologies for being so blunt but you obviously haven't got a clue what you are talking about if you think the only work going on is the drawing of the bay around you. I doubt you wrote your own 3d engine because I doubt anyone with the capability to do so would make a statement like, but I would like to see you try and build a station like the ones we have in ED - even using a 3rd party engine rather an your own - and then maybe you would be in a position to tell FD they don't what they are doing or why they get lower fps in a station.
 
Apprently you have not looked up any of the techniques I've quoted, nor followed (with understanding) the wiki link.

In short - frustum culling is a technique which allows the application to draw ONLY VISIBLE stuff. This means, that when you're sat in the bay, the rest of the station should not be pushed to GPU, as the view of the station is occulded by the bay's walls, or the 'empty box'.

Of course there could be a suboptimal particle emitter there throwing thousands of tris to render a cheap smoke, and that might be killing the GPU. Doesn't mean that it's good coding either.

Again, the inside of the station is not very complex, there's not a lot of geometry that would affect performance.

And yet again, why is my GPU load lower when I'm looking at the inside of the station, and higher when I'm sat under the deck?

For your reference, find a book titled "Real Time Rendering" in the local library, read through, and then we can get back to this discussion. All I read now is "hur hur, FD knows what they're doing and you are a random internet nobody, because I think so".
 
Apprently you have not looked up any of the techniques I've quoted, nor followed (with understanding) the wiki link.

In short - frustum culling is a technique which allows the application to draw ONLY VISIBLE stuff. This means, that when you're sat in the bay, the rest of the station should not be pushed to GPU, as the view of the station is occulded by the bay's walls, or the 'empty box'.

Of course there could be a suboptimal particle emitter there throwing thousands of tris to render a cheap smoke, and that might be killing the GPU. Doesn't mean that it's good coding either.

Again, the inside of the station is not very complex, there's not a lot of geometry that would affect performance.

And yet again, why is my GPU load lower when I'm looking at the inside of the station, and higher when I'm sat under the deck?

For your reference, find a book titled "Real Time Rendering" in the local library, read through, and then we can get back to this discussion. All I read now is "hur hur, FD knows what they're doing and you are a random internet nobody, because I think so".

I used to the Digital Video/Audio Product Specialist for a major software company which is a household name for it's image editing software. I could talk the industry talk pretty well. I could even sit down and do a lot of general work with the tools despite having very limited real-world experience with them, simply because I was book-smart on the concepts and functionality. The reality was, I could talk about the tools knowledgeably, which was my job, but there's no way I could keep up with the level 2 or level 3 technical/development support staff. Your posts sounds like something I would have written back then... like you know enough about the topic to be dangerous. Are you a full-time professional coder now, or did you just take some classes back in college?
 
I used to the Digital Video/Audio Product Specialist for a major software company which is a household name for it's image editing software. I could talk the industry talk pretty well. I could even sit down and do a lot of general work with the tools despite having very limited real-world experience with them, simply because I was book-smart on the concepts and functionality. The reality was, I could talk about the tools knowledgeably, which was my job, but there's no way I could keep up with the level 2 or level 3 technical/development support staff. Your posts sounds like something I would have written back then... like you know enough about the topic to be dangerous. Are you a full-time professional coder now, or did you just take some classes back in college?

Like I said, I did a BSc. paper on spatial optimization for realtime graphics 10 years ago. I took a 2 year course dedicated stricly to graphics, so it's hardly "some classes". Now I'm managing software projects, but outside of the graphics spectrum.
Still, stuff should get better over time, not worse. I'd love to have some actual engineer school me on where I'm mislead.

At the end of the day, without looking at the code or talking with actual devs we can just speculate and that's it. My expectation is simply that if geometry is occulded, it should not be rendered.
 
Apprently you have not looked up any of the techniques I've quoted, nor followed (with understanding) the wiki link.

In short - frustum culling is a technique which allows the application to draw ONLY VISIBLE stuff. This means, that when you're sat in the bay, the rest of the station should not be pushed to GPU, as the view of the station is occulded by the bay's walls, or the 'empty box'.

Of course there could be a suboptimal particle emitter there throwing thousands of tris to render a cheap smoke, and that might be killing the GPU. Doesn't mean that it's good coding either.

Again, the inside of the station is not very complex, there's not a lot of geometry that would affect performance.

And yet again, why is my GPU load lower when I'm looking at the inside of the station, and higher when I'm sat under the deck?

For your reference, find a book titled "Real Time Rendering" in the local library, read through, and then we can get back to this discussion. All I read now is "hur hur, FD knows what they're doing and you are a random internet nobody, because I think so".

I still think you are underestimating what FD are doing with the graphics, what is going on in the background and you are still have a low end card which is a poor place to start making judgements from.

Quoting somone else's work is easy, doing is much harder - you think you can do better, then let's see it.....
 
Last edited:
For the record I'm experiencing the same issues with the same graphics card. I've had the same rig though Alpha, Beta, Gamma, and release and only recently has it started doing this. I used to get 45 - 50 fps inside stations no problem without the GPU doing a Boeing impression. Somethings most definitely up.
 
For the record I'm experiencing the same issues with the same graphics card. I've had the same rig though Alpha, Beta, Gamma, and release and only recently has it started doing this. I used to get 45 - 50 fps inside stations no problem without the GPU doing a Boeing impression. Somethings most definitely up.

Thanks for confirming that I'm not crazy. But watch out, soon you'll be told that because wings were added, now there's more to render in the docking bay.. ;-)
 
I've actually managed to fix my reduced framerate in stations by changing the refresh rate. It was set to 50hz, swapped it to 60 and it made a huge difference in FPS. The GPU still goes a bit mad fan wise in the stations though.
 
Try to use the standard FOV. Not maxing it out. I think this also helps.
But yeah, stations are always the most FPS nagging b-a-s-t-a-r-d-s.
I think it is mostly because of the "steam" in there.
I play at Ultra and 1,5x Supersampling with my R9 280 and so get under 40 FPS inside stations, but don't mind it.
You get what you payed for. And i paid just 180€ on my Graphics Card 6 month ago. And comparing games did not help there.
Star Citizen i.e. runs a lot worse then Elite. But i don't complain there either. You always have different games being more or less demanding.
Maybe the XBox One port will bring us even better FPS on lower end systems. But you simply just can't max out settings on such systems and complain about low FPS on the same time.
 
Last edited:
You see, that's the whole problem I have with today's computing, enthusiasts and industry in general.
Back in the day, devs would go down to the lowest possible level and do optimizations to get stuff working. Nowadays, we just slap tons and tons of hardware on top of what seems to be poorly written code, or just a performance edge case.

Because it's the glitter that sells (and the most costly in game development -- that lead animator makes a bigger pay check than the lead programmer).

Unfortunately, even my tank of video card that is the only AMD video card I've had that ran cool ever, it even runs hot in ED. Guess it's more of a tuning issue that time will make better, but it does run hot to the point come summer it'll be unbearable. Pure heat comes from the fans, which never has been a case from ripping HD videos to running 3 game clients at once on ultra.

It. Runs. Hot.
 
Apprently you have not looked up any of the techniques I've quoted, nor followed (with understanding) the wiki link.

In short - frustum culling is a technique which allows the application to draw ONLY VISIBLE stuff. This means, that when you're sat in the bay, the rest of the station should not be pushed to GPU, as the view of the station is occulded by the bay's walls, or the 'empty box'.

Of course there could be a suboptimal particle emitter there throwing thousands of tris to render a cheap smoke, and that might be killing the GPU. Doesn't mean that it's good coding either.

Again, the inside of the station is not very complex, there's not a lot of geometry that would affect performance.

And yet again, why is my GPU load lower when I'm looking at the inside of the station, and higher when I'm sat under the deck?

For your reference, find a book titled "Real Time Rendering" in the local library, read through, and then we can get back to this discussion. All I read now is "hur hur, FD knows what they're doing and you are a random internet nobody, because I think so".

The "under deck" issue is probably a simple matter of it not being worth optimizing for. It doesn't matter if the frame rate is lower when you store your ship since you can't do anything in the dock. Why put in code that doesn't do anything
but make the FPS number larger in a situation where the player can't fly or shoot. The issue inside of the station in different. That is a place where improvements to FPS can make a big difference in the experience.

Frustum culling has advantages, although generally at a cost on the CPU side and it's not really compatable with modern GPU card design. It's also not friendly with head mounted displays since people can move their head very quickly. Better to have all the geometry already resident on the GPU ready to be drawn when the viewing direction is sent to the GPU.

GPU performance has two main bottlenecks, polyon performance or pixel performance. Culling out polygons helps with the first, but not with the second. I suspect it's not the polygon count inside the station that causes the major performance hit, but is the pixel shaders being used. Older cards were not optimised to run pixel shaders like the more recent cards. Perhaps FD could rewrite some of the shaders to be less processing heavy on older cards.
 
Last edited:
GPU performance has two main bottlenecks, polyon performance or pixel performance. Culling out polygons helps with the first, but not with the second. I suspect it's not the polygon count inside the station that causes the major performance hit, but is the pixel shaders being used. Older cards were not optimised to run pixel shaders like the more recent cards. Perhaps FD could rewrite some of the shaders to be less processing heavy on older cards.

It's something to do with the geometry, because when in a station and opening up the star chart to check different systems, that is when the video card pumps out the heat.

I can run 3 WoW clients on ultra and not produce this much heat for one game just processing star geometry data.

So it's more than shaders causing it. It feels more like the heat that comes from SETI or Folding WUs. It's really running the math hard.
 
Back
Top Bottom