The broad conclusion is credible, but the precise numbers need qualification. Noticias3D reported the results from TechPowerUp’s testing, while TechPowerUp’s original charts, annotations and complete test configuration were not directly verified. One particularly unusual finding—no detected frame-rate improvement from Intel frame generation on an Arc B580—therefore remains an unexplained launch observation rather than proof of a known hardware, driver or game defect.
Native maximum settings impose a steep GPU cost
Noticias3D’s secondary account places the GeForce RTX 4090 at approximately 51 FPS and the RTX 5080 at 43 FPS at native 4K with maximum settings. It also reports about 48 FPS for the Intel Arc B580 at native 1080p maximum settings.
Those numbers indicate a substantial rendering workload, but they should not be used to construct a definitive comparison between the individual cards. Without direct access to TechPowerUp’s original charts and full configuration, important details—including the test scene, system configuration and chart qualifications—remain unconfirmed.
The larger performance conclusion has independent support. PC Games Hardware found high load on both the processor and graphics card and described upsampling as nearly compulsory. That does not verify each Noticias3D figure, but it supports the broader finding that native maximum-quality rendering is exceptionally expensive.
These results also concern the highest settings rather than every practical configuration. A maximum preset can function more like a stress test than a sensible default, especially when the visual difference from the next preset is not established. Windows players should not assume that a recommended-specification PC is intended to run the game at native resolution with every setting maximized.
EA’s 1440p target includes upscaling
EA describes Zero Company as CPU-bound, but its official recommendation does not promise native 1440p at 60 FPS. The company says that the target is best achieved with TSR or vendor upscalers such as DLSS and FSR.
That detail is central to interpreting the system requirements. Upscaling renders the game internally at a lower resolution and reconstructs the output image, reducing part of the GPU workload. If a player meets the recommended specification but falls short of 60 FPS at native 1440p, that result does not necessarily contradict EA’s target.
The CPU-bound description should not be read as a claim that graphics performance is unimportant. A game can place considerable pressure on both components, with the limiting factor changing by scene, resolution and graphics setting. Lowering internal resolution can help when the GPU is the primary constraint, but it cannot eliminate processor-heavy simulation or rendering work.
For troubleshooting, resolution, preset and upscaling mode matter as much as the names of the installed CPU and GPU. A fast graphics card cannot guarantee a locked frame rate in a CPU-limited scene, while a strong processor cannot overcome the full cost of native 4K maximum settings.
The Arc B580 frame-generation anomaly remains unexplained
Noticias3D reports that the Arc B580 achieved approximately 48 FPS at native 1080p maximum settings but showed no detected FPS increase when Intel frame generation was enabled. That is the most unusual result in the secondary benchmark account.
Intel’s documentation says Arc B-series hardware supports XeSS 3 Multi Frame Generation, with as many as three AI-generated frames inserted between conventionally rendered frames. That capability is consistent with x2, x3 and x4-style options appearing in a game menu. It does not guarantee that selecting x4 will quadruple the measured frame rate in every game or configuration.
The reported lack of detected uplift does not establish that the B580 lacks frame-generation support. Nor does it prove that generated frames were never produced: measurement software can differ in whether and how it counts generated output. Conversely, Intel’s hardware-level support does not invalidate the reported observation. The two facts can coexist—a supported feature and a test in which no improvement was detected.
The cause remains unknown. The available evidence cannot distinguish among a game-integration problem, driver behavior, a setting interaction, a measurement limitation or another test-specific factor. The exact Intel driver, Windows build, game build and XeSS runtime used in the benchmark are not established.
There is also no verified later official patch or driver notice establishing whether the behavior was corrected or acknowledged on newer builds. The finding should therefore be treated as a dated launch observation, not evidence that every current Arc B580 system will reproduce it—or that a fix definitely exists.
Resizable BAR is a separate recommendation
EA separately advises Intel Arc owners to enable Resizable BAR to improve FPS performance. Resizable BAR allows the processor to access a larger portion of graphics memory at once, and it is an important general platform requirement for obtaining expected Arc performance.
That guidance must not be presented as an explanation for the frame-generation anomaly. The Resizable BAR state of the reported test is unknown, and there is no test evidence showing that disabling it caused the lack of detected uplift. EA’s general performance recommendation and Noticias3D’s report of the Arc B580 result are separate facts.
Arc owners should still verify Resizable BAR because EA recommends it for overall performance. They should not assume that doing so will resolve the specific frame-generation behavior. The only reliable way to determine that would be controlled testing with the same game scene and settings before and after a confirmed platform change.
Frame generation is not free performance
Frame generation can increase displayed frame rate by inserting generated images between conventionally rendered frames, but it carries its own processing and memory costs. It also does not raise the underlying simulation rate in the same way as rendering more original frames.
AMD’s general FSR documentation estimates approximately 313 MB of additional GPU-memory use at 2560×1440 and 410 MB at 3840×2160, before frame-generation swapchain overhead. These figures provide useful technical context for why frame generation is not a zero-cost toggle.
They do not explain any specific Zero Company benchmark result. No directly verified game-specific memory telemetry, exact Radeon frame-generation configuration or controlled reproduction establishes memory pressure as the cause of poor scaling on a particular GPU. Drawing that conclusion would require evidence about resolution, settings, existing memory use and what happened when capacity was exceeded.
Frame generation also works best when the conventionally rendered base frame rate is already reasonably stable. A higher displayed figure cannot remove all latency, frame-pacing or CPU limitations present in the underlying workload. Players should establish native or upscaled performance first and then evaluate frame generation as a separate step.
Launch drivers provide a troubleshooting baseline
AMD Adrenalin Edition 26.8.1 added Zero Company support, while NVIDIA’s GeForce 616.56 launch driver advertised day-one support for DLSS 4.5 Super Resolution and Multi Frame Generation. These versions identify the launch baseline for Radeon and GeForce systems, not necessarily the most appropriate version for every later game build.
Using an older, pre-support driver adds an avoidable variable and may omit game-specific profiles or fixes. At the same time, installing a launch driver cannot guarantee that every game-side integration issue has been addressed. Upscaling and frame generation depend on interaction among the game, driver, runtime, Windows graphics stack and measurement tools.
Players using later drivers should examine the release notes for the version they install and retest rather than assume that an undocumented change fixed—or introduced—a particular behavior. Comparisons are meaningful only when resolution, preset, scene, upscaling mode and frame-generation setting remain consistent.
Performance and stability are different problems
Demanding benchmark numbers do not establish that every PC will experience widespread instability. PC Gamer’s desktop review on a Radeon RX 9070 XT encountered two hard crashes but characterized most of its other technical problems as relatively minor.
Two crashes remain significant, particularly in a long game, but that experience illustrates why three categories should be tracked separately:
- Rendering performance: The conventionally rendered frame rate at a defined resolution and preset.
- Frame-generation behavior: Whether generated frames appear and whether the selected counter records them.
- Software stability: Crashes, hangs and other faults that can occur regardless of the displayed frame rate.
A PC can deliver low native performance without crashing, or run at a high displayed frame rate while retaining an unrelated stability problem. Raising the frame rate through reconstruction does not itself repair a crash, and a stable session does not prove that graphics settings are performing as intended.
A practical setup order for Windows players
A controlled configuration process is more useful than immediately selecting maximum settings and the largest frame-generation multiplier:
- Install an appropriate game-support driver. The identified launch baselines are AMD 26.8.1 and NVIDIA 616.56; later users should choose an applicable successor with relevant game support.
- Check Arc platform configuration. Intel Arc owners should confirm that Resizable BAR is enabled because EA recommends it for overall FPS, without assuming it explains the B580 frame-generation report.
- Establish a conventional baseline. Test a repeatable section with frame generation disabled and record the resolution, preset and upscaling setting.
- Enable upscaling before chasing maximum settings. EA’s own 1440p/60 target assumes TSR, DLSS or FSR, making reconstruction part of the intended performance path.
- Test frame generation independently. Change only that setting and use a counter capable of recognizing generated frames where possible. An x2, x3 or x4 label is a mode description, not a guaranteed multiplication of measured performance.
- Reduce expensive settings when needed. The reported native results suggest that lowering the preset may be more dependable than relying entirely on generated frames.
- Track stability separately. Record crashes and hangs independently from performance changes so that one issue is not mistaken for the other.
What the launch evidence supports
The strongest conclusion is that Zero Company’s maximum settings are unusually demanding and that EA designed its recommended 1440p/60 target around upscaling. Noticias3D’s reported 48 FPS, 51 FPS and 43 FPS figures reinforce that conclusion, although the underlying TechPowerUp charts and full test configuration remain unverified.
The Arc B580 result warrants further controlled testing, but its cause is unknown. Intel documents support for XeSS 3 Multi Frame Generation, while the secondary benchmark report detected no performance increase. EA’s separate Resizable BAR recommendation should be followed for general Arc performance, not retrofitted as an unsupported explanation for that anomaly.
For Windows players, the practical response is to update the graphics software, configure platform requirements, begin with upscaling, and measure frame generation separately. Native maximum settings remain useful for stress testing, but they are not a realistic shorthand for what EA’s recommended specification is intended to deliver.