That still makes me laugh, a 1060 runs about as well as a PS4 in this game. I don’t know how that’s possible.
A deep analysis of what’s going on with the port and performance would be interesting but I guess the answer is a lot of small issues and porting problems adding up.
Coding practices and other errors and mistakes or just a lack of time or not utilizing D3D12/Vulkan and the advantages of low-level API’s to their fullest.
I just hope a lot of it can be patched but I know some of it will be difficult if not too problematic like the 30 FPS animations.
I guess I can take Valve off the curse word filter now … I was able to convince them to refund my application fee.
That really pissed me off, considering no fewer than 3 times the entire premise of the software was explained and approved – once for the proposal to even be allowed to apply to list non-game software on the store, a second for the proposed proposal (lol), and then a third time when the store page was reviewed. It should have been brought to my attention that there were aspects of the design they did not like before I spent ~1 year in development (6 months too much to normally qualify for a refund).
I’ll have to see if I can get any specifics on what usage of SteamAPI was deemed so egregious that the build review process went straight from reject / revise / re-submit to “do not pass go, do not collect $200.” Might be pushing my luck, but this last bit of communication was a bit more … not completely 1-sided.
Don’t they literally mean just a complete removal of SteamAPI features? I personally doubt they looked into it properly, or it would’ve been approved.
And then trailers and screenshots that show no games, and the software is basically approved unless they have something else up their sleeves.
Edit: Oh cool i just realized you can 
posts.
2 Likes
I assume it’s some deeper Steam and Steam API functionality and some vague agreement but maybe poorly documented mutual understanding on never overriding the Steam API functionality from software.
…Though breaking it is apparently fine even when Valve then has to hot-patch the client into not going critical but some third party utility having key fixes for … a lot … of games hitting API usage issues no that’s a problem.
Open source too so the code review and team more involved with this type of thing could have done a lot within the first month or so I would think not this last part just suddenly halting and stopping things entirely although it’s good to hear you managed to work out a refund at least despite some initial issues with that. 
EDIT: It’s all on a case by case basis and Valve could and have altered client behavior a few times though they can’t outright go out after custom projects and tools hosted elsewhere. Makes it difficult to get some clear answers if one reviewer has the projected accepted and then another one blocks it and now discussions back and forth on why and how to resolve it.
(If you even get a actual detailed answer at all or if it’s some PR or corporate statement.)
How would that work? 
The entire point of me putting Special K onto Steam was to make use of Workshop and various other services that Steamworks offers. As an actual first-class citizen in the Steam ecosystem, it stops being weird that I am using SteamAPI.
The only place it remains weird, I think, is the BLoggedOn (…) override. I jokingly referred to that as “offline DRM bypass,” since Sonic Mania’s day one fiasco was caused by that API. That override has stayed in the codebase, however, because I know of multiple games that crash in the middle of gameplay if the logged on state changes, as well as games like Mega Man Collection that will crash during startup if not overridden.
I am a wealth of knowledge on all things that could possibly go wrong If anything, Valve should consult me rather than shove me out of the way. Anything SteamAPI-related weird in Special K’s codebase was put there due to a malfunctioning game on their store.
Thought it would entail removing custom achievements, global stats, perhaps screenshots etc - though i would think you could still make HDR screenshots work regardless of SteamAPI.
What you’ve stated does sound like too much to be expected to remove, but if Valve isn’t providing any clarity then they are basically saying “hey it’s too much effort to look into what you do with SteamAPI, and it makes our job easier if you scrap anything to do with it.”
The approval system is likely just a checklist of things to verify and they rush through it. You can ask them about it, but i wouldn’t bet on a clear answer.
Seems the links have been reinstated on the review.
A site using HTTPS does not mean it is “secure”. It only means the communication between your computer and the website is encrypted over the public internet.
Assuming it means the site itself is actually “secure” to use is a major unsubstantiated assumption that should not be made.
Since a few years ago fully automated and free certificate authorities such as Let’s Encrypt exists that basically lets any random website easily and rapidly support encrypted communication over HTTPS. It is, for example, what this Discourse and the SK GitLab subdomains make use of – all without ever having to go through the hassle of actually purchasing a custom SSL certificate.
In this case I imagine the shared links might’ve run afoul some automated system of Valve’s that temporarily saw links removed until it was manually approved or such. Right now the links seems to have been reinstated, with the usual link disclaimer warning.
1 Like
By the way, @Kaldaien, can you install the Discourse Cookie Consent Banner plugin so we get one of those random cookie banners for users?
Yeah, HTTPS means nothing other than your traffic’s not going out in plain text. Which is what all those stupid VPN services that advertise on YouTube want/need you to believe.
It’s just not the case, I got SSL certificates without spending a single cent or doing any of the normal things you would expect (e.g. getting a notary and proofs of identity). The “site is secure” indicates nothing about the party on the other end.
It’s a little different for software code signing, which I haven’t gotten around to yet. There are hoops you have to jump through and anual fees associated with that.
I recompiled the entire server before realizing that’s not a plug-in 
It’s a theme component, and only takes 5 seconds versus 5 minutes.
Confirmed UE3 games that work so far: AC Chronicles: China, Arkham City and Origins, Bioshock Infinite, Bulletstorm Full Clip Edition (standard edition is DX9 and hasn’t been tested), Borderlands Game of the Year Enhanced, Shadow Complex Remastered and Thief 2014. The only UE3 games that gave me issues with HDR during the tests were Arkham Knight and Murdered: Soul Suspect. There aren’t a lot of DX11 UE3 games, sadly. The vast majority of them were DX9 only.
Do you have any examples you can show of this? For literally 15 years I’ve just straight up used 16x AF forced in Nvidia Control panel globally, and haven’t noticed anything out of the ordinary.
Rage is the only one I can think of, forcing anisotropic filtering has otherwise just worked although AMD and NVIDIA quickly add profiles in to cap or limit it where required so after a few driver updates Rage 1 only did 4x and no higher which worked with the Mega Texture and avoiding square pattern textures.
NVIDIA has NV Inspector so if the actual flag for it could be identified and you could get a list of profiles where that applies that would probably be a more extensive example just that current drivers are quickly updated to cover for this so even with global 8x or full 16x set the profile either hinders overrides or caps it to a compatible amount.
Was surprised to learn that for AMD neither AA or AF actually applies when forced to Vulkan or D3D12 whereas for NVIDIA Vulkan doesn’t seem to apply but for DirectX 12 anisotropic filtering does work although apparently here’s this first game with visible glitches when this is used and it has to be disabled.
(And the game updated to fix the AF option actually doing anything.)
Performance for Horizon ZD is much better here than what DF managed, maybe it helps to use a high end CPU and ram? This video’s results are more in-line with what i would expect from these GPUs. Weird thing going on with Vega though. Radeon 7 seems fine, but Vega 56 and 64 suffer horribly?
vs (also testing the benchmark run)
Edit: Although DF isn’t using original settings here, maybe something they enabled crushes lower-end GPUs?
Vega suffers compared to Navi cards as the game is RDNA optimized using SPD or single pass downsampling optimizations on some of the shaders as part of AMD Fidelity FX.
Potentially those GPU’s also suffers from a driver issue in Adrenaline 2020 with regards to async compute where utilization of this is causing lower performance on Polaris and Vega GPU’s and while it doesn’t apply here there’s a second ongoing issue in regards to lower Vulkan performance on at least Polaris GPU’s too.
Think that covers for that, restarting after shader compilation can also help or the game might end up hitting well into 20 GB RAM usage which for many users at 8 - 16 GB system RAM and hopefully a page file will start utilizing said page file which either slows things down or the game just crashes. (Lower sized fixed page file for example so it just runs out.)
EDIT:
EDIT: It’s a bit like Far Cry 5 and Vega benefiting from FP16 optimized shaders and being able to handle 2x of these at a time although far as I am aware or at least heard of only the water effect utilized it in this game and NVIDIA Turing cards also do FP16 even if the “Rapid Packed Math” stuff about doing 2x calculations together I think is AMD GCN specific and done on Vega and newer.
Or rather it’s one of those fancy optimization things and tricks on newer GPU hardware, RDNA is a bit different from GCN so I assume the hardware divide in feature and support here is a actual thing and not just a software and driver divide.
1 Like
Wonder if this goes into what Durante and some other tech sites discovered and the game is hindered by PCI bandwidth speed which is very uncommon.
Some deeper analysis would be needed but potentially the memory bottleneck is more pronounced thus why Vega falters behind the more balanced architecture on Polaris while Radeon VII just brute forces it by having all the bandwidth and memory availability.
Could also be a driver optimization issue, AMD noticed that after reports on the AMD Reddit and are investigating.
Curious as to why that got discovered there and not in the various tech benchmarks earlier as if AMD’s behind the games PC port or the tech it utilizes they’d be interested in seeing it perform as expected and would follow tech articles and videos and would be aware of this already.
EDIT: How was this called again, Vega GPU’s are front-loaded in terms of hardware and excel at computational tasks but the memory and raster performance hampers the overall experience though it still held up but hit a lot of issues though it overcame many earlier problems like what the Fury ran into and eventually the Polaris 580 started surpassing it in benchmarks.
Radeon VII with extra memory stacks and with HBM having even more bandwidth and total memory speed as a way to overcome at least this limitation.
Unsure if there’s been any testing for Navi and PCI Express and if the game also starts showing performance issues at PCI-E 4.0 8x versus 16x or when compatibility drops to 3.0 including on non-X570 hardware.
Certainly seems that a NVME device on a few of the less capable boards or chipsets where the lanes aren’t quite enough so it goes from 16x/8x to 8x/8x causes a bit of a performance dip at least.
I think some AMD employees are moderators on the subreddit, or at least they post on the sub from time to time.
I’m interested in seeing how the Navi cards fare in this game using PCIe 4.0, maybe there’s a boost in performance there.