Almost every single time I open social media I am faced with a familiar picture – a video of an old classic, perhaps Half-Life 2, maybe Batman: Arkham Knight or Crysis, and underneath it someone explaining how it was “more optimized” than modern games. How it runs better, sometimes how it even looks better.
To me this is obvious engagement farming, but some people do genuinely believe these things. Claims that Unreal Engine ruined games and performance, that Frame Generation or Reconstruction are evil, or that modern games are uniquely demanding are all topics we at Sapphire will address eventually. For now we start with the first claim.
Were old games more optimized? Did they run better?
Understanding optimization
The first thing readers must understand is that optimization is very different from being easy or hard to run.
If tomorrow I make a game that looks like classic Doom 2 on a modern engine and it runs at 300 fps on an RX 6500 XT, that is not optimized. It is easy to run, yes. On the other hand, imagine a game that looks like real life and runs at 1 frame per second on an RX 9070 XT. That is extremely demanding — yet it is optimized. Incredibly optimized, even.
Optimization is not about how easy a game is to run. It is about using available hardware and software intelligently to extract the highest possible visual fidelity, simulation complexity, scale, and game-design ambition. Modern ray-traced titles like Doom: The Dark Ages or Alan Wake 2 can be very demanding depending on settings, but they are highly optimized. Elden Ring is far easier to run, yet considering its fidelity, scale, and simulation it is comparatively unoptimized.
This does not mean a game is bad or good. We should separate the game as an art form or entertainment from the purely technical discussion.
The same applies to art direction. Art direction is more important than raw graphical fidelity — few people would contest this in good faith. But art direction does not determine optimization or how hard a game is to run. It is an artistic choice. At some point we must still be able to discuss fidelity on its own terms.
Hardware and software also change over time. Optimization is a moving target. Crysis was very demanding for its day, yet genuinely optimized for the hardware of 2007. It featured meaningful multi-core CPU usage, a 64-bit version, and heavy use of advanced GPU features for the time. From a 2026 perspective Crysis is easy to run, but its optimization belongs to a different hardware era. It was never designed around today’s many-core CPUs, modern graphics APIs, or current GPU feature sets, so it does not scale like a modern engine. Even a well-optimized modern title such as Doom: The Dark Ages will not look especially impressive when we run it on 2040 hardware — even if doing so will be trivial by then.
Old games crushed classic hardware — and they were often attacked for it
Complaints about how games run are as old as PC gaming itself. They simply appear more frequently now because PC gamers discuss performance more openly. Every new release draws claims that it is unoptimized, too demanding, or “ruined.”
The more things change, the more they stay the same :D
When those claims appear, the same people almost always hold up an older game as the opposite — something optimized and well-running. Very often they are misremembering or being selective with the truth.
Many of the most beautiful and influential games in history were extremely hard to run, or at least to max out, when they launched. Doom 3 was effectively impossible to max out on release hardware. F.E.A.R. hammered PCs of its era. Return to Castle Wolfenstein, Quake, and Unreal all generated forum meltdowns over performance. Company of Heroes, Oblivion, S.T.A.L.K.E.R.: Clear Sky, Far Cry, The Witcher 2 and 3 — all produced real controversies about how they ran on contemporary hardware. HairWorks, anyone? Batman: Arkham Knight is easy today, but the Radeon R9 Fury X of its time was not laughing (and the game was also broken at launch). The same was true of Assassin’s Creed Unity. Half-Life 2 could run acceptably on hardware of its era until you examined frame-time behavior. The game stuttered so badly at launch that if it released today we would call it Stutter-Life 2. Valve’s patches, engine updates, and the sheer leap in hardware since 2004 eventually tamed it.
Why the comparison is unfair
Old games benefit not only from two decades of hardware progress but also from being fully patched and supported by mature drivers. Many received substantial improvements over the years. The Half-Life 2, Quake II, or Doom 2 you launch on Steam today are not the original release versions. Developers continued working on them; the experience is more polished and often higher-fidelity than what players saw on day one.
https://www.youtube.com/watch?v=jlJxXMaRWBk
Half Life has been changing slowly over time!
Judging a brand-new modern release by that same standard is simply not fair.
Scalability — a fair point
One advantage older games often held deserves acknowledgment. They were frequently demanding on Medium, High, and especially Ultra settings, yet their Low settings were usually quite accessible. Running Crysis was not hard; many systems below the official requirements could still play it. The same was true for Unreal Tournament, the early Wolfenstein titles, The Witcher 3, and many others. Reaching the highest settings was difficult, but simply playing the game was not.
https://www.youtube.com/watch?v=y4sG21EKGxw
Potato settings are so funny!
Modern games could learn from this and offer more scalable fidelity options. Some already do, but the principle is worth pushing for. At the same time, Medium or Low settings in a 2005 title are nothing like Medium settings in a 2026 title. Modern games remain visually impressive even on lower presets.
Ultra settings are usually for future hardware
A common pattern today is players with mid-range or older hardware insisting on Ultra settings for a brand-new game and declaring it unoptimized when it does not run well. This happened in the past too, though less often. In both cases it is the wrong approach.
Developers do not primarily target the absolute highest settings. They target consoles and a reasonable mainstream PC configuration. Ultra is largely for screenshots and future hardware. That is a useful rule of thumb. If you cannot max out a game, try Medium or High, stop overthinking it, and enjoy the experience. It is usually almost as good-looking while running much faster.
Some examples of people complaining about performance
Here are concrete examples of users complaining about performance in older titles, along with those games running on hardware of their era (adjust for inflation).Doom is widely regarded as a crowning achievement of optimization in 1993 — and it was. Yet on a nearly top-end $2,000 PC of the time (more than $4,000 in 2026 money) it often ran in the teens and dipped lower still.
This PC is 4000 USD when adjusting for inflation…
Half-Life 2 is one of PC gaming’s legendary titles. Digital Foundry footage from the period shows just how badly it stuttered at launch.
https://www.youtube.com/watch?v=xplR_JTXhbI&t=766s
Digital Foundry talking about the stutter in Half-Life 2
F.E.A.R. remains a masterpiece. Contemporary forum posts are full of players complaining about how hard it was to run. Nothing has changed.
Ah, a classic comment!
The conspiracy!
Some claim the original S.T.A.L.K.E.R. games ran on potatoes. They did scale to modest hardware, but high or maximum settings were another story. The ATI Radeon HD 4890 — a genuine high-end card of its day — was pushed hard on High settings and still could not fully max the game out. It took a R9 290X to do that, 5 years later.
Do note - this is NOT maximum settings!
Unreal Tournament generated entire threads of players complaining about performance.
Batman: Arkham Knight has outstanding art direction, but a single room in a modern game often contains more micro-detail than entire regions of its world. Performance-wise, contemporary video shows it punishing a GPU that would cost roughly $900 in today’s money.
Finally, a personal note: I am playing Far Cry 4 right now. It is an excellent game. Even on a Ryzen 7 9800X3D and a Sapphire Nitro+ RX 9070 XT I still see occasional drops to around 70 fps at 4K maximum settings. That is not disastrous performance, but a 2014 title behaving this way on current flagship hardware shows it is not optimized by modern standards. Great game, though.
We do not have it that bad
Every era of PC gaming has had its struggles. In the 1990s we lacked fully unified graphics APIs and high-end gaming PCs were rare and expensive. The early 2000s were a brutal slog: every new GPU was a leap forward, yet every new ambitious title was extremely demanding to max out. Later we fought shader compilation stutter, inconsistent frame times, and poor console ports.
The present day is transitional as well. We are still in the early years of ray tracing and path tracing. 4K has become a realistic standard and HDR is maturing. We also face higher-than-expected hardware prices and slower growth in consumer VRAM, partly driven by AI and datacenter demand.
DOOM 2 launched 1 year after DOOM 1 and crashed performance on the same hardware!
Yet there are clear advantages. Virtually every modern game remains playable on an RX 6600 XT — a mid-range card from 2020. A mid-range GPU from 2000 would not have handled Prey or Company of Heroes nearly as well. The 6600 XT can run current titles at respectable settings and frame rates, and modern games usually look strong even on Medium or Low.
Conclusion
Do not believe everything you see online. Many people who insist old games were perfectly optimized were not there at the time, or they simply do not check contemporary benchmarks and forum posts. The legendary PC titles of the past were no joke for the hardware of their day.
Optimization does not mean “easy to run.”
I do not want to make excuses for cases where we have a genuine lack of optimization in either old or new games — that remains a real issue. But we need to be fair and sober both about the present - and the past too!
