Star Citizen Media Musings

[dsmart]

Because 40fps at 1080p is still better than 6fps at any resolution.

Yeah, that's not how that works.

Also, his claims were performance improvements at 1080p, while comparing them to 2K. That's a meaningless test as he would have had to test at 2K in order to make that sort of comparison worthy of note.

[dsmart]

I'm very much interested to see if they can deliver any performance. But for a valid test you cannot alter anything in your test bed and then claim a performance increase. My contention is that he new his baseline fps with 2k, and for some reason he changes it after the update and then claims fps improvements makes everything he said null and void.
The real question is what is he calling 2k maybe (1440p)?

Precisely. They are using it to - once again - make baseless and unfounded claims. It's what they do. Confidence is shaky and at an all time high.

Meanwhile, as if we were expecting any less, in 8 days, 3.1 would have missed the Q1/18 deadline. And most of the shit they promised, aren't done.

https://robertsspaceindustries.com/roadmap/board/1-Star-Citizen

[Kyrt]

I'm very much interested to see if they can deliver any performance. But for a valid test you cannot alter anything in your test bed and then claim a performance increase. My contention is that he new his baseline fps with 2k, and for some reason he changes it after the update and then claims fps improvements makes everything he said null and void.
The real question is what is he calling 2k maybe (1440p)?

I disagree.

The game was unplayable for many before and the resolution had minimal impact.

If the game is optimised to the extent that changing resolution makes it playable then that is an improvement.

Yes...you can't do a like for like comparison. But if whatever optimizations CIG has added are seen as making the game generally playable, then a like for like isn't necessary.

This isn't a scientific double blind test. An exact comparison isn't necessary, mandatory or even particularly useful.

All I need to know is....is the game "playable"? Yes....1080p is skewing the result in CIGs favour, but since the existing situation is that the game is often unplayable no matter what the resolution is, having the game be playable at 1080p is still an improvement.

Whether there is actually an improvement is unknown right now. But as I said, I believe we can expect some degree of improvement in performance. It could hardly get worse ;)

But the point I am trying to make is simple...

Would you consider 40FPS at 1080 an improvement over 6FPS at 2k?

[jwh1701]

Would you consider 40FPS at 1080 an improvement over 6FPS at 2k?

I would agree but my question why is it better, since variables changed
then we do not have sufficient data. I did not see anything in cig documentation
concerning 1080p.
Would 1080p yield better fps pre-patch and do we have the data?
Since we do not have his fps before the patch then I cannot determine if it was the patch
or resolution change.

[dsmart]

I disagree.

The game was unplayable for many before and the resolution had minimal impact.

If the game is optimised to the extent that changing resolution makes it playable then that is an improvement.

Yes...you can't do a like for like comparison. But if whatever optimizations CIG has added are seen as making the game generally playable, then a like for like isn't necessary.

This isn't a scientific double blind test. An exact comparison isn't necessary, mandatory or even particularly useful.

All I need to know is....is the game "playable"? Yes....1080p is skewing the result in CIGs favour, but since the existing situation is that the game is often unplayable no matter what the resolution is, having the game be playable at 1080p is still an improvement.

Whether there is actually an improvement is unknown right now. But as I said, I believe we can expect some degree of improvement in performance. It could hardly get worse ;)

But the point I am trying to make is simple...

Would you consider 40FPS at 1080 an improvement over 6FPS at 2k?

1) 2K      = 2048 x 1080, 2048x1152, 2014x1440

2) 1080p = 1920x 1080

#1 pushes a LOT more pixels than #1.

It is game in which most of its bottleneck is CPU - not GPU - bound. At least the last time I checked.

The performance in 3.0 remains abysmal. At any resolution. Of course if you play at LOWER resolution, you will get better performance. ergo. drop from 2K to 1080p. That's the way it is with ALL games.

3.1 comes out.

A guy claiming to get bad performance in 3.0 when running at 2K, says he's getting better performance in 3.1 when running at 1080p. He never made ANY statements related to his 3.0 performance at 1080p.

Please explain to me - in simple terms - how exactly you came to the conclusion that his claims are related to "improvements" in 3.1 performance, as opposed to him just dropping down his resolution.

As a graphics programmer - even if I never played 3.0 or 3.1, none of that would make any sense.

The ONLY way to determine if in fact 3.1 has any performance improvements, is to run the game at the same resolution as 3.0, with the same config settings, in the same server environment (heavy vs empty), and in the same location (planet, space etc).

[dsmart]

Would you consider 40FPS at 1080 an improvement over 6FPS at 2k?

I would agree but my question why is it better, since variables changed
then we do not have sufficient data. I did not see anything in cig documentation
concerning 1080p.
Would 1080p yield better fps pre-patch and do we have the data?
Since we do not have his fps before the patch then I cannot determine if it was the patch
or resolution change.

It won't. Because it's nonsense.

[Kyrt]

Would you consider 40FPS at 1080 an improvement over 6FPS at 2k?

I would agree but my question why is it better, since variables changed
then we do not have sufficient data. I did not see anything in cig documentation
concerning 1080p.
Would 1080p yield better fps pre-patch and do we have the data?
Since we do not have his fps before the patch then I cannot determine if it was the patch
or resolution change.

Before the patch, people were stating that changing resolution up or down made little or no difference.

If changing resolution in 3.1 works, that indicates an improvement of sorts.

We don't know if this test was manipulated oi anyway (as it was in 3.0 with their hacks and running on empty servers), nor what actual performance will be like on live, nor what impact running with other players nearby will have or several other caveats.

But 40 fps, even at 1080p, is better than 6fps at whatever resolution.

The question I have is whether he manipulated the tests through means such as hacking, or running on empty servers, or editing. The question I have is are his results replicated throughout the player base. The question I have is how multiple players and ships affect fps rate.

Because given what players were complaining about wrt 3.0, at least the ones I saw, the ability to get semi decent FPS at least some of the time will be a massive improvement over 3.0, even if resolution has to be sacrificed.

Were I to be involved in narrowing down where and why and how such an improvement took place, then the change in resolution would indeed be problematic.

As a player, that the bottleneck has moved from where it was in 3.0 where 6or 7fps wasn't uncommon, and that 40fps is achievable will be all that I care about.

In short....resolution wasn't a major factor in the problems affecting 3.0. To a large degree, it was the CPU overhead caused by trying to process the huge amount of data coming in over the network, which hammered the system even before resolution became an issue. If we are now at the stage where we are pointing out resolution changes, that tells me that the previous bottleneck...the network...is no longer hammering the CPU and it is dealing with received information well enough that the bottleneck has moved to the point resolution is now an issue.

We don't really have enough information from this one test to determine if performance really has changed or improved. But we cannot discount  the "test" simply because of resolution changes for the simple reason, resolution wasn't an important factor in 3.0 and if it is now, in 3.1....that IS an improvement.

[Kyrt]

#2 pushes a LOT more pixels than #1.

Other way around, or am I missing something?

It is game in which most of its bottleneck is CPU - not GPU - bound. At least the last time I checked.

It is a game where much of the problem with performance was caused by the CPU trying to deal with a deluge on network trafgic, most of which was irrelevant.

If the performance is now at a point where the GPU affects performance in a meaningful manner,  that tells me the bottleneck has moved. It indicates that whatever kludge they put into place to filter the network traffic is having SOME impact to the point that the CPU can deal with the traffic it is receiving.

If that is the case, that would be an improvement over 3.0.

he performance in 3.0 remains abysmal. At any resolution. Of course if you play at LOWER resolution, you will get better performance. ergo. drop from 2K to 1080p. That's the way it is with ALL games.

Most games. One of the complaints 3.0 had was that changing resolution had little or no impact and the rationale given was that the main bottleneck was in the network. That each client was getting hammered with too much information. That the CPU became the bottleneck in performance because the GPU rarely got a look in

If 3.1 really has changed things to the point changing resolution affects performance, that's an improvement over 3.0 where all too often it did not.

A guy claiming to get bad performance in 3.0 when running at 2K, says he's getting better performance in 3.1 when running at 1080p. He never made ANY statements related to his 3.0 performance at 1080p.

And given how Citizens manipulated the results for 3.0, if a simple resolution change change affected performance we would have seen them push  that and get CIG to optimise the graphic routines. They wouldn't have needed to hack their clients into offline mode or seek empty servers.

Resolution changes apparently had little impact on performance in 3.0. If it does now, there is a new bottleneck.

As a graphics programmer - even if I never played 3.0 or 3.1, none of that would make any sense.

That can be true only if you assume only one bottleneck in the game. Was the resolution a major determining factor in performance during 3.0? No. It wasn't. If it is now....and its too early to tell...then the bottleneck in 3.0 has gone and there is a different  bottleneck now.

Yes...all else being equal, resolution would be important. 

But it isn't equal and we don't need information about resolution to determine if the game can simply deliver acceptable performance at some resolution.

The ONLY way to determine if in fact 3.1 has any performance improvements, is to run the game at the same resolution as 3.0, with the same config settings, in the same server environment (heavy vs empty), and in the same location (planet, space etc).

Or we could just wait and see if the same complaints and symptoms crop up in 3.1.

To keep it short...

In your opinion, did changing resolution have a major impact on the performance issues noted in 3.0? Was lowering the resolution seen as a possible and viable solution?

 

[SpaceTroll]

Fun starts at 3:59:00

 

[Noztra]

I like how Boredgamer think that the video with Andy Serkins that shows off ray tracing are somehow gonna be how it will look in SC. He even mention that its not even CY but UE, but that doesn't matter. :P

And impressive that he thinks SC will be able to implement ray tracing into SC.

[dsmart]

We don't really have enough information from this one test to determine if performance really has changed or improved.

That's why it's particularly hilarious to me that a guy is claiming performance at 1080p is better in 3.1 than in 3.0. All he did was lower his resolution. Doing that in ANY game will yield better performance results.

But we cannot discount  the "test" simply because of resolution changes for the simple reason, resolution wasn't an important factor in 3.0 and if it is now, in 3.1....that IS an improvement.

Yes we can discount it because it's not only is it not based on empirical evidence, but also it's just nonsense.

Resolution is ALWAYS a performance issue in games; especially in CryEngine games which are notorious for pushing systems to their limits. I have no idea where your statement about 3.0 not being an important factor comes from. The performance in 3.0 is abysmal at ANY resolution. If that were not the case, there would be no complaints about it because people would just play at the standard 1080p and not bitch about it. In fact, this person dropping down from 2K to 1080p, pretty much confirms it, regardless of the veracity of his claims. And in his case, it's basically a false positive.

As I said, I've played it. I always play games at 1080p and I haven't noticed ANY performance improvements. Plus the crashes and hangs are worse in 3.1.

I remember when I was writing various articles for MONTHS on end about how performance in 3.0 was completely catastrophic. Those guys on Reddit were saying all kinds of things, attacking me etc. Then 3.0 dropped in Dec 2017. Silence. Then rage. Then acceptance. It's all Deja Vu.

[dsmart]

Other way around, or am I missing something?

It is a game where much of the problem with performance was caused by the CPU trying to deal with a deluge on network trafgic, most of which was irrelevant.

That's not true. That's just been an excuse harped by CIG and carried by Shitizens for the past TWO YEARS.

The game's performance is affected in many areas, and networking is only a part of that. And unless we see the game profiled in one of their dev shows, there is NO empirical evidence to support the theory that networking is primarily the main bottleneck. Things like bind culling and serialized variables are only going to go so far. And even if/when they implement either, the results will show precisely that. And whatever improvements they yield, would hardly be much of an impact. They know this. And that's why they've been putting it off for years now. Aside from the fact that, as I've written in my articles, it has the potential to break everything - completely.

If the performance is now at a point where the GPU affects performance in a meaningful manner,  that tells me the bottleneck has moved. It indicates that whatever kludge they put into place to filter the network traffic is having SOME impact to the point that the CPU can deal with the traffic it is receiving.

If that is the case, that would be an improvement over 3.0.

None of that has anything to do with the on-going discussion about performance in 3.0 v 3.1.

Whatever networking improvements they made - if any - would affect ALL resolutions. So the guy won't have had to drop down to 2K to play it because network improvements aren't specific to resolutions.

[dsmart]

I like how Boredgamer think that the video with Andy Serkins that shows off ray tracing are somehow gonna be how it will look in SC. He even mention that its not even CY but UE, but that doesn't matter. :P

And impressive that he thinks SC will be able to implement ray tracing into SC.

He's a moron.

Even the devs who created the tech know that ray tracing is YEARS away. I have a thread on Facebook right now discussing this. And besides Microsoft's own statements, there are leading graphics programmers who are friends (e.g. Andre LaMothe who has written so many gamedev books, papers etc I lost count) of mine, talking about precisely that.

[Noztra]

And It only took 4 NVLinked Tesla V100's to run that EPIC'S Star Wars demo at 1080p24. :)

[Kyrt]

That's not true. That's just been an excuse harped by CIG and carried by Shitizens for the past TWO YEARS.

Maybe it isn't true.

The game's performance is affected in many areas, and networking is only a part of that.

Ah....so it IS true.

I'm quite prepared to accept that there is more wronf with the engine than just the netcode. I'm even prepared to accept that the serialized variable culling they mentioned is little mire than a band aid fix, and one that quite possibly will have side effects, and very likely will end up being semi permanent despite being portrayed as a temporary  measure.

But ultimately, the only question of importance that I....as a gamer...am going to be interested in is does the game have acceptable levels of performance?

It is, to a large degree,  irrelevant what performance was like in 3.0 or what CIG fixed or didn't fix.

If they improved the netcode....that is an improvement even if the game performance still suffers. If the game performance still suffers, if it is just as bad for just as many players as 3.0, then they failed. But if there is a noticeable improvement in performance then how such an improvement is achieved may be of dome interest to technophiles like you or I....but the typical gamer isn't going to care.

And ultimately, unless you are interested in forensic analysis of the games performance, the actual resolution is unimportant and a bit nitpicky.

Has the games performance improved?

And unless we see the game profiled in one of their dev shows, there is NO empirical evidence to support the theory that networking is primarily the main bottleneck.

Quite bluntly...it doesn't have to be. It does appear to be a bottleneck, it does appear to have a negative impact on performance and whether you blame that on overutilisation of the network, lack of suitable filtering on the server side, the client side CPU getting hammered by network traffic or wherever, fixing that issue...even via a temporary bandaid...is still an improvement.

How much of an improvement is debatable but arguing that we can ignore the issue because it may not be the primary cause of performance issues isn't exactly helpful either.

Whatever networking improvements they made - if any - would affect ALL resolutions. So the guy won't have had to drop down to 2K to play it because network improvements aren't specific to resolutions.

Again, you appear to be arguing that the game has just one bottleneck affecting performance.

I don't believe that, nor do I think you do. Fixing the networking bottleneck could easily result in an improvement in performance by freeing up clientside CPU cycles, an improvement which can then be undone by maintaining an unrealistic resolution impacting the GPU which in turn can be fixed by reducing said resolution. Said improved performance will then continue until yet another bottleneck is hit...such as having several ships or players  onscreen at one time.

None of this means the netcode hasn't been improved or that overall performance hasn't improved. Nor does it mean it will.