Occlusion Culling and the hardware survey: question to the tech comunity

Hi, i don't know anything about graphic rendering, so i might be totally wrong about this, but i'm a bit confused about their hardware survey when a lot are pointing out the alleged lack of optimizations and rendering issues with odyssey:


So, i'm wondering what's the point of an hardware survey in those conditions, and my question is, since as i said i'm clueless about it, am i wrong and the lack of this occlusion culling might be not such a problem as it sounds it is?
 
So, i'm wondering what's the point of an hardware survey in those conditions

Some possibilities:

  • To deflect from the actual issue.
  • Hope that they'll find some pattern in the submissions that can be used as a scapegoat.
  • Complete ignorance of the managers and discrete development teams about what others are doing.

am i wrong and the lack of this occlusion culling might be not such a problem as it sounds it is?

I can't say it's the only problem (it's almost certainly not, and may not even be the largest), but even with my limited experience on the software side of things, I can tell there is something seriously wrong with how the game has been handling occluded objects simply by observing how it performs on various systems in various scenarios.
 
Last edited:
What is clear from those posts is that the current engine is almost certainly not culling anywhere near as much as it could. That said, many reports of performance issues are coming from folks with beafy enough hardware that it should just be able to power through the inefficient rendering pipeline, while there are people on more modest hardware that aren't suffering the same level of performance problems.

This disconnect is why FDev are looking to see if there's a pattern to the perfornance problems with the request for data. If they weren't aware of the culling issue before - which seems impossible, given culling optimisation was mentioned during one of the alpha phase livestreams - they certainly are now, but it's clear that they don't believe that alone will solve the performance issues, hence the request for data.

The lighting / colouration issues are a different beast: they're affecting everyone, and so there's no need to gather additional data. Whether this has been caused by a bad code merge, by the rush to get to release or something else we'll likely never know.

A general piece of life advice that has served me well is to always assume good intent unless presented with evidence to the contrary. I have confidence that FDev are working on this, and the more people with performance problems that give them the feedback they've asked for (in the way they've asked for it) the faster they'll be able to identify the cause and fix it.
 
They are most likely hoping of identifying a common nominator that is bottlenecking the game from performing as well as it should. If they can find it, they can, hopefully, try to fix it. Some people have, for example, said that Elite is not utilizing their CPU's and GPU's properly, which can cause the framerates to drop and as the hardware landscape is very heterogenous, it's almost impossible to get the needed data from anywhere else other than the players.
 
...I got interesting thought and send em too last minute.
I play from Linux and I have disabled spectre/melt protection explicit, so my CPU usage is about 35%.
If windows users have it enabled on w10 and not present on w7/8, then that could be a reason. Difference there is huge, especially for gaming as it invalidates all last 20 years inventions.
 
Some possibilities:

  • To deflect from the actual issue.
  • Hope that they'll find some pattern in the submissions that can be used as a scapegoat.
  • Complete ignorance of the managers and discrete development teams about what others are doing.



I can't say it's the only problem (it's almost certainly not, and may not even be the largest), but even with my limited experience on the software side of things, I can tell there is something seriously wrong with how the game has been handling occluded objects simply by observing how it performs on various systems in various scenarios.

Good grief. This is a quite strange post.

Here's something closer to the truth. PC`s use a myriad of different hardware in a myriad of different configurations. In order to target optimisations efficiently, they need to know how different configurations are performing.

This is important because there are only a finite number of configurations they can test in house.
 
What is clear from those posts is that the current engine is almost certainly not culling anywhere near as much as it could. That said, many reports of performance issues are coming from folks with beafy enough hardware that it should just be able to power through the inefficient rendering pipeline, while there are people on more modest hardware that aren't suffering the same level of performance problems.

This disconnect is why FDev are looking to see if there's a pattern to the perfornance problems with the request for data. If they weren't aware of the culling issue before - which seems impossible, given culling optimisation was mentioned during one of the alpha phase livestreams - they certainly are now, but it's clear that they don't believe that alone will solve the performance issues, hence the request for data.

The lighting / colouration issues are a different beast: they're affecting everyone, and so there's no need to gather additional data. Whether this has been caused by a bad code merge, by the rush to get to release or something else we'll likely never know.

A general piece of life advice that has served me well is to always assume good intent unless presented with evidence to the contrary. I have confidence that FDev are working on this, and the more people with performance problems that give them the feedback they've asked for (in the way they've asked for it) the faster they'll be able to identify the cause and fix it.
So they might be trying to get ahead of the lack of optimizations, because they realized that some systems are performing worse than they should even with it?
 
it's clear that they don't believe that alone will solve the performance issues

It won't. Odyssey still performs like crap, relative to Horizons, even where there is negligible potential for occlusion to improve things.

The lighting / colouration issues are a different beast: they're affecting everyone, and so there's no need to gather additional data. Whether this has been caused by a bad code merge, by the rush to get to release or something else we'll likely never know.

I'm not sure how this differs from the performance issues, except that some people aren't noticing the performance drop they are surely experiencing.

PC`s use a myriad of different hardware in a myriad of different configurations.

Not in dispute.

However, Odyssey performs near universally worse and certain optimizations will improve performance no matter what hardware it's run on. These aren't outliers that are having issues and targeting specific configurations is not going to address whatever the primary underlying issues are.
 
I'm pretty sure it is not just the culling, but it has to be a big part. I went back, wiped my graphic settings again, and went with High, with the only change being ground textures to ultra and switching OFF blur. So it shouldn't impact space, right? Ryzen 7 2700X, RTX 2060, 16GB RAM Took off from a platform. So far so good, actual Horizons style 100+ frame rates. Fly over to a Coriolis. Arrive at Coriolis from the back 70fps. Go to door and face inside, where the concourse is in the back.. 50fps. :cautious:
 
I'm pretty sure it is not just the culling, but it has to be a big part. I went back, wiped my graphic settings again, and went with High, with the only change being ground textures to ultra and switching OFF blur. So it shouldn't impact space, right? Ryzen 7 2700X, RTX 2060, 16GB RAM Took off from a platform. So far so good, actual Horizons style 100+ frame rates. Fly over to a Coriolis. Arrive at Coriolis from the back 70fps. Go to door and face inside, where the concourse is in the back.. 50fps. :cautious:
So is it different if you face the back of the station or the front? Maybe there is some culling afterall
 
So is it different if you face the back of the station or the front?
Yeah, as far as I'm aware, when you approach the station and not facing the mail slot, it doesn't render the inside. But once you look at the mail slot, it does. And in there frame rate varies depending on you facing the part where the on-foot concourse is, or not. It is super jarring.

Same thing on planets really, if you as much as glance towards a settlement, even if it is kilometers away, and behind a mountain, your fps drops like a stone.
 
Yeah, as far as I'm aware, when you approach the station and not facing the mail slot, it doesn't render the inside. But once you look at the mail slot, it does. And in there frame rate varies depending on you facing the part where the on-foot concourse is, or not. It is super jarring.

Same thing on planets really, if you as much as glance towards a settlement, even if it is kilometers away, and behind a mountain, your fps drops like a stone.
However when you face the back of the station from the outside, it doesn't look like it renders the concourse right?
 
However when you face the back of the station from the outside, it doesn't look like it renders the concourse right?
That I would have to test, I think it is distance dependant, but outside everything is fine, until you see that little blink, when the insides of the station appear.
 
However, Odyssey performs near universally worse and certain optimizations will improve performance no matter what hardware it's run on. These aren't outliers that are having issues and targeting specific configurations is not going to address whatever the primary underlying issues are.
Of course, certain optimisations will bring the performance up universally, that's a given.

But some optimisations rely on profiling certain hardware configurations, that's why the information has been requested.
 
But some optimisations rely on profiling certain hardware configurations, that's why the information has been requested.

I certainly acknowledge the former, but I'm not convinced of the latter. I did send the the info, just in case they can do something useful with it.
 
I'd say the hardware survey is partly due to the incredibly variable results that people are getting. High end hardware is tanking.. Low end sucks.. Mid-range seems be kinda OK, or is for some. Occlusion does appear to be an issue, but I'm not sure it's all of the problem.

The results are not that varied - they seem to be quite consistent
mine for example: laptop, i7-9750h, 16gb ram (2x 8GB, HyperX Impact, DDR4, 2666MHz, CL15), GTX1660TI - i get as low as 30-32 FPS in Concourse (higher in orbital outpost)

The variable is the personal bias response to such low framerates

Personally i'm not that fussed about it, for me it is playable, but others are more sensitive to the fact that they are getting only 30fps on rigs better than recommended specs
 
The results are not that varied - they seem to be quite consistent
mine for example: laptop, i7-9750h, 16gb ram (2x 8GB, HyperX Impact, DDR4, 2666MHz, CL15), GTX1660TI - i get as low as 30-32 FPS in Concourse (higher in orbital outpost)

The variable is the personal bias response to such low framerates

Personally i'm not that fussed about it, for me it is playable, but others are more sensitive to the fact that they are getting only 30fps on rigs better than recommended specs
And the Concourse is not even a good datapoint anymore, as you don't have to do FPS there. CZs in big settlements with lots of actors (PC or NPC), large tourism hubs, that's where the chugging really begins.
 
Hi, i don't know anything about graphic rendering, so i might be totally wrong about this, but i'm a bit confused about their hardware survey when a lot are pointing out the alleged lack of optimizations and rendering issues with odyssey:
Becuase there's a lot more than one problem to solve here. Some of the performance problems are related to what you mention in your OP. But there are other, different performance issues which are more to do with specific hardware configurations.

The results are not that varied - they seem to be quite consistent
The results must be different enough that they want to target certain optimisations against specific hardware configurations.
 
  • Like (+1)
Reactions: LKx
Back
Top Bottom