Zen 5 vs Arrow Lake Thermals Under Extended Gaming Workloads

Two Architectures, One Thermal Question
When you push a modern desktop CPU through three or four hours of continuous gaming – not a benchmark sprint, but actual extended session workloads – the thermal story gets complicated fast. AMD’s Zen 5 chips and Intel’s Arrow Lake lineup both promise efficiency gains over their predecessors, but their behavior under sustained load tells very different stories about how each company thinks about heat management.

How Zen 5 Handles the Long Haul
AMD’s Ryzen 9000 series, built on the Zen 5 architecture, carries forward the chiplet design philosophy that has defined Ryzen for years. The compute die is fabbed on TSMC’s 4nm node, and the I/O die sits on a separate 6nm process. That split matters thermally because heat is generated across two physically distinct areas rather than one dense monolithic piece of silicon. Under gaming workloads specifically, the CCD handles the heavy lifting while the I/O die stays comparatively cooler, which means a good cooler doesn’t need to fight a single concentrated hot spot – it’s managing a more distributed thermal load.
That said, Zen 5 is not a quiet architecture. At stock settings, chips like the Ryzen 9 9900X push into the high 80s Celsius under extended gaming, particularly in titles that keep multiple cores active across long sessions. Games like Cyberpunk 2077, Microsoft Flight Simulator 2024, and CPU-intensive strategy titles like Total War: Warhammer III will consistently drive temperatures upward once you get past the first hour of play. The architecture does throttle gracefully – AMD’s precision boost algorithm backs off boost clocks before temperatures become a problem, so you rarely see thermal runaway – but you will see clock speeds drift down from their peak numbers.
One genuinely useful behavior in Zen 5 under gaming is how the architecture handles thermal headroom between workloads. During loading screens, menu navigation, or scene transitions, the chip drops power draw quickly and temperatures recover within seconds. This means sustained gaming sessions don’t necessarily mean sustained peak temperatures – there’s enough natural pacing in most games to give the cooler time to catch up. A 240mm AIO or a quality 120mm tower cooler can handle Ryzen 9000 series gaming loads without drama, provided your case airflow isn’t fighting you.
AMD’s Eco Mode is worth mentioning for anyone prioritizing thermals over maximum frame rates. Dropping a 9900X to its 65W TDP limit costs only a few frames per second in most gaming scenarios but brings operating temperatures down by 15 degrees or more in extended sessions. For a living room PC build, a quiet setup, or anyone using a compact mid-tower with modest airflow, that trade-off makes practical sense. The performance-per-watt story for Zen 5 at reduced TDP is strong enough that most gamers won’t notice the difference in actual play.

Arrow Lake’s Thermal Profile Is a Different Problem
Intel’s Core Ultra 200S series, based on Arrow Lake, took a different architectural path – one with real consequences for thermal behavior. Intel moved to a tiled design similar in concept to AMD’s chiplet approach, but with a hybrid core layout combining P-cores and E-cores across separate tiles. The P-core tile handles the performance-critical work during gaming, and in most game engines, that tile sees intense, concentrated utilization while the E-core tile sits at relatively low load. That imbalance creates an uneven thermal signature that some cooler designs handle less effectively than others.
Under extended gaming, the Core Ultra 9 285K runs cooler than its Raptor Lake predecessor in absolute terms, which is a genuine improvement Intel earned through the process node transition to TSMC’s 3nm for the compute tiles. You can expect P-core temperatures in the mid-to-high 70s Celsius under sustained gaming with a competent 280mm AIO, compared to the low-to-mid 90s that the Core i9-13900K regularly hit. Arrow Lake’s power draw during gaming is notably lower than Raptor Lake’s, and that directly translates to less heat generated per session. For a platform comparison, the Ryzen 9 9950X3D versus Core Ultra 9 285K breakdown goes deeper on how these differences show up in actual game benchmarks.
Where Arrow Lake runs into trouble is with voltage behavior out of the box. Intel’s default settings on Z890 motherboards allow aggressive voltage spikes during burst workloads, and while gaming isn’t a sustained all-core cinebench run, titles with frequent physics calculations, AI pathfinding, or large open-world streaming – Elden Ring, Star Citizen, or modded Skyrim at high NPC counts – will trigger short but intense power spikes that translate to temperature jumps. These spikes don’t last long enough to cause throttling in most cases, but they keep the average operating temperature higher than Intel’s marketing materials would suggest.
Intel’s Thread Director technology, which assigns workloads to the appropriate core type, has improved substantially since its initial Alder Lake debut. In extended gaming sessions, it generally does a reasonable job keeping game threads on P-cores and background processes on E-cores, which helps prevent thermal crossfire between both tile types running hot simultaneously. But some game engines, particularly older or poorly threaded titles, don’t interact cleanly with Thread Director, leading to P-core underutilization and occasional thermal anomalies where E-cores spike unexpectedly. It’s an edge case, but it shows up in testing.
Arrow Lake also behaves differently depending on the motherboard vendor’s power delivery defaults. MSI, ASUS, and Gigabyte all ship Z890 boards with different out-of-box power limit configurations, and the thermal difference between a conservative board and an aggressive one can be 8-10 degrees under identical gaming workloads. AMD’s platform is not entirely innocent of this problem – X870 boards have their own variance – but the gap is narrower because AMD’s boost algorithm takes more direct control of the process.
Which Architecture Rewards Patient Cooling Choices
For builders specifically thinking about long gaming sessions – marathon RPG runs, extended esports practice, overnight streaming setups – Zen 5 is the more predictable platform to cool. Its temperature behavior under gaming is consistent, its power management responds well to cooler quality, and the distributed chiplet design means you’re not fighting a single-point heat concentration. Arrow Lake runs cooler at stock than Intel’s previous generation, but its voltage behavior and board-level variability add complexity that Zen 5 largely avoids.

The practical question isn’t which chip gets hotter in a worst-case scenario. It’s which architecture lets you set up a system once, walk away, and trust that hour four of a gaming session looks the same thermally as hour one. On that specific measure, Zen 5’s boost management and the maturity of AMD’s platform-level thermal controls give it a clear edge – though if Intel tightens up Arrow Lake’s default voltage behavior through future microcode updates, that gap could close without any hardware change at all.



