The Forever Winter

This isn't my own opinion, but it deserves to be etched in stone: Alright, sick of people who knows nothing about what they're talking about defending UE5. You're getting a wall of text. Cause at this point, “just blame the developers” is the lazy explanation. When the SAME TYPES OF PERFORMANCE PROBLEMS keep appearing across completely unrelated studios, completely unrelated genres, completely unrelated budgets, and completely unrelated teams, the common denominator starts mattering. And the common denominator is Unreal Engine 5. Shader compilation stutter. PSO compilation hitches. Traversal stutter. Asset streaming stalls. Heavy CPU overhead. Expensive Lumen. Expensive Virtual Shadow Maps. Nanite overhead. Reliance on aggressive temporal reconstruction to claw performance back. These aren't some bizarre collection of identical mistakes that dozens of unrelated developers all independently decided to make. They are recurring pain points associated with the technology stack they are all using. That is the entire reason blaming UE5 exists in the first place. Your argument essentially requires us to believe that studio after studio after studio keeps mysteriously producing the same categories of performance pathology, but the shared engine underneath all of them is somehow above suspicion. That's ridiculous. And Epic themselves clearly don't believe that, because Epic has spent enormous engineering effort changing UE5 specifically to address shader compilation, PSO precaching, traversal hitching, streaming, CPU performance, Nanite performance, Lumen performance, and other bottlenecks. Why would Epic need engine level solutions if these weren't engine level problems? Apparently Epic should just close the engineering tickets and write: “Skill issue. Developers should optimize better.” Problem solved. The existence of games that run well does not rescue your argument either. Of course UE5 CAN run well. Nobody said the engine physically prevents a game from ever achieving good performance. The argument is that UE5 makes certain performance problems far more common, and developers repeatedly have to engineer around those problems. Those are completely different statements. A competent team can spend enormous amounts of time profiling UE5, precompiling shaders, managing PSOs, controlling asset streaming, limiting Lumen, limiting Nanite, tuning Virtual Shadow Maps, building alternate scalability paths, optimizing CPU workloads, and designing their game around the engine's limitations. And then you point at the finished product and say: “See? The engine wasn't the problem!” No. You just pointed at developers successfully WORKING AROUND THE PROBLEM. That's like watching someone reinforce a structurally weak bridge and concluding that because it didn't collapse, the original engineering was flawless. Your list actually demonstrates this beautifully. VALORANT is a competitive shooter designed around extreme performance, with a comparatively constrained rendering workload. Tekken 8 is an arena fighter. Hellblade II is heavily controlled and linear. Fortnite has numerous scalability modes and can disable expensive rendering features specifically to accommodate weaker hardware. Different workloads place radically different pressure on an engine. Then you throw Black Myth: Wukong into the list as though people didn't extensively document shader and traversal stuttering in that game. And “STALKER 2 Patch 2.0” might be the funniest one of all. You are citing a game that required enormous post launch optimization work and an engine upgrade as evidence that the engine wasn't contributing to the performance problem. Read that sentence again. If STALKER 2 needed months of patches, engine changes, CPU optimization, streaming improvements, and rendering work before you were comfortable putting it on your “UE5 performs fine” list, you aren't making the argument you think you're making. You're demonstrating how much work can be required to beat UE5 into shape. And this is the important distinction you keep dodging: Developers control how well they optimize their game. The engine controls what they HAVE TO OPTIMIZE. UE5 determines the renderer. UE5 determines the shader pipeline. UE5 determines how PSOs are handled. UE5 provides the asset streaming architecture. UE5 provides Nanite. UE5 provides Lumen. UE5 provides Virtual Shadow Maps. UE5 provides the temporal reconstruction stack. UE5 provides the systems developers are fighting with when these problems appear. So yes, developers can absolutely make bad optimization decisions. Nobody is giving developers blanket immunity. But when developers from studio A, studio B, studio C, studio D, studio E, and studio F all encounter remarkably similar technical problems while using the same engine, at some point screaming “THE DEVELOPERS!” stops being analysis and starts being denial. Software engineers understand this concept perfectly well. If twenty different applications using the same library start throwing the same class of failure, you investigate the shared dependency. You don't conclude that twenty independent engineering teams coincidentally became incompetent in exactly the same way. And UE5 has another problem that makes this even more obvious: Its marquee features are EXPENSIVE. That's not an insult. That's literally what they are. Lumen is expensive. Nanite isn't free. Virtual Shadow Maps aren't free. High density geometry isn't free. Large seamless worlds with aggressive streaming aren't free. Temporal reconstruction exists partly because rendering these workloads natively at high resolutions can be prohibitively expensive. UE5 encourages developers toward extremely sophisticated rendering technology, and then developers have to claw performance back through scalability settings, reconstruction, dynamic resolution, shader management, streaming optimization, and enormous amounts of profiling. That is an ENGINE CHARACTERISTIC. So stop pretending the only two possibilities are: “The developers are responsible” or “Epic is responsible.” Developers can fail to mitigate an engine problem. That does not magically make the underlying engine problem disappear. And when the same complaints follow an engine from game to game, year after year, studio after studio, eventually the intellectually serious position is to examine the thing they all have in common. Unreal Engine 5. Your giant list of exceptions does not disprove a pattern. It proves there are developers who have successfully mitigated it. Those are not the same thing. https://steamcommunity.com/id/CapnJordy