snc
Veteran
He literaly mention api as one of possible things
Current Generation Games Analysis Technical Discussion [2022] [XBSX|S, PS5, PC]
- Thread starter BRiT
- Start date
Bias is fine. There's nothing wrong with it. I just think his videos could use the clarification here on his behalf to showcase why he thinks it's fillrate/alpha transparency.
It just doesn't explain why, on a zoom up of a boss, with no alpha, and no overdraw drops to 30fps. But once zoomed out, and all alpha effects and overdraw is now apparent, it suddenly jumps back up to 60fps. This is a reversal of the claim that alpha/fill rate are the reasons why PS5 is holding an advantage over XSX. When this particular scene should have this reversed for instance. Both should be hitting 60 at the close up, and 30 at the zoom out with PS5 holding the advantage.
i'm aware, I watched it.
but he says PS5 is outperforming XSX here 'owing to alpha, overdraw and fillrate advantages'.
But when referring to API, 'possibly be'
I don't have a problem with either statement, but he should showcase why the first statement he believes is the case showcasing examples of where he thinks this is true. Because in his video I can easily find areas where the opposite occurs: in particular no over draw, no alpha, no fillrate issues, and the frame rate is tanking on both.
If these are the bottlenecks for engine performance, we should see consistency around this particular area.
Sony really need to patch VRR on PS5! I think I am going to play the game using PS4 BC (particularly with that excellent 1800p CBR that reminds me the excellent CBR used in Dark souls remastered) then I'll do another playthrough with another class using the PS5 version with VRR.VRR display owners, you're fine! Non-VRR display owners, just tune in soon and see yourfateprize!
snc
Veteran
Nope, he didnt say possibility but that fromsoftware engine favour ps5 configuration and api, also said many times that diff is more academic and versions are close, buffling me that such statements can trigger somebody
If Sony doesn't provide VRR by the end of the year, one would assume that the PS5 hardware is incapable of VRR support, and that Sony has made a grave mistake in mentioning it.
not triggered. Just offering constructive criticism on how he could improve his technical analysis here.
As for API, yes it's their first jump to DX12. Not sure if that's the same as favouring PS5 as not understanding how to DX12.
And as per does it actually favour PS5 configuration?
I don't think it does, because if it did it would run like the PS4 Pro version on PS5.
There is something dramatically wrong with the PS5 edition if PS4 Pro version runs so well locked at 60 on PS5.
If I had to make a statement here on 'favouring a platform' I would say the engine favours last gen if anything: given that we now have reports that the X1X edition of ER runs 1800p near locked 60 on XSX and we know PS4 Pro edition runs locked 60 on PS5.
snc
Veteran
Ok but this negativity for imo not controversial content at all is bizzare
I don't think I'm being negative here. I'm seriously just offering constructive criticism.
He needs to find his audience here; if the goal is to provide surface level technical analysis for a buyers guide, he should stay at the surface level and just talk about performance. Once you start talking about the reasons why you think the performance is dipping a particular way, you should have to explain to the audience what it is you're referring to, how it works, what they observed, how they could reproduce the phenomena. There should be a consistency to trends. There should be consistency to seasonality, and cycles. When there is no consistency, it is just noise and random variation.
Being able to separate these items for viewers is critical for a deeper technical analysis for viewers who want to _learn_ more.
This is constructive criticism on how to improve his technical analysis because these are the things he is missing to back his claims. Otherwise his claims are... well just claims.
snc
Veteran
Points that I agree with but it started with some io talks that wasnt in this video in terms of performance
Exactly. We have loading times showing the PS4 loadings (with HDD) are actually quite faster than on XB1 / X1X (HDD) and not far from Series X levels (using SSD). And as you said the game isn't stable 60fps on PS5 either and loading times don't improve that much on PS5 compared to PS4. The difference is only because of SSD (3 times faster, about same diff from X1X to X series X actually, so HDD -> SSD), not custom PS5 I/O (otherwise we would see 20 times faster like in many others titles).
On the other hand the game is performing relatively (~30fps) really well on PS4 and Pro (and X1X too as it actually runs a bit better than Pro here, I still don't know if X1X runs at native vs Pro). It's obvious the game was optimized (framerate) to run well on old gen hardware with a focus on faster loading times on PS4. Maybe they use Oodle compression API on PS4 as it was available since some time there which would explain the anomaly of loading times on PS4 (inherited by Pro and PS5).
All next-gen versions seem to be a quick port from old gen version with relatively no specific and drastic I/O or performance improvements.
Last edited:
snc
Veteran
And that I dont understand what was the purpose to talk about under elden ring video. Its like df sometimes made mistakes and on every new video without watching it we should post about it. But imo enough about it, many people here dont like nxgamer videos so maybe indeed good idea to not post it.
its okay to make mistakes. Mistakes do happen, we have limited information.
But NXG could resolve his mistakes a lot more often if he just did some quick checks and showcase whether it holds true or not. Even the simple graph listed above provides some very interesting information about the state of the game on release.
The main difference between NXG and DF is the depth their willing to go to showcase their theories about the data that is coming back. At least I can follow DF's arguments as to their claims, they showcase it. They reproduce it. They double check it with PC. They setup tests to showcase this isn't some noise/random variation. I can't follow NXG. And that's really the crux of my criticism here. I don't mind that he's wrong, but there should be something tangible for me to follow to understand why he thought his claim was true for a technical analysis that wants to go a layer below reporting on surface level metrics.
He doesn't provide that however, and it's something I'd like for him to do. Or if he doesn't want to waste his time on it, then he shouldn't name drop reasons for why the numbers are that way. There's nothing for the viewer to follow. He tarnishes his own reputation by doing this imo.
NXG's opinions shouldn't be presented as anything other than a theory and a starting point for a conversation, same as any other. In the absence of convincing evidence, no particular weight should be afforded the presented interpretations of the engines. As he's not here to discuss his ideas, feedback on his content isn't particular useful to B3D.
Unlike Dictator. Feel free to rake his questionable interpretations over uncooled CRAYs.
If there's a theory, just posit "here's a theory from (source)" and then follow on with evidence for/against from whatever sources.
Unlike Dictator. Feel free to rake his questionable interpretations over uncooled CRAYs.
If there's a theory, just posit "here's a theory from (source)" and then follow on with evidence for/against from whatever sources.
I love ACG, but he's probably the least technical tech youtuber I could name off the top of my head. Great content, though. He has opinions. I don't always agree with them. But I believe that he believes them, and that mean he isn't pandering to his audience. And that's a respectable place to be, in my book.
Interesting. Right, per your article, you saw the stutters from the start. ACG’s review play-through was “stuttery, it had fps issues, but it was playable” (from his latest podcast). He has a 1080Ti, 2080Ti, and 3080. Skill-Up had a “near flawless” experience on his 2080Ti. I don’t know their OS or other system stats. I don’t believe they mentioned pervasive stuttering to such a degree in the beta test, and it’s hard to believe nothing changed in the game between the beta and review copies.
From gonna From, I guess.
Beyond the shader compilation, I wonder how much of the stutter can be attributed to EAC or sidestepped by running offline? Proton vs. Windows would also be interesting.
Similar threads
- Locked