ComputerBase’s hardware-measured comparison, as summarized by The FPS Review, used NVIDIA’s LDAT sensor across seven games with an RTX 5060 Ti 16GB, Radeon RX 9060 XT 16GB, and Intel Arc Pro B70. The test avoided each vendor’s own telemetry, which matters: in-game latency counters are calculated differently and cannot be treated as interchangeable measurements of click-to-photon delay.
The result is more useful than the familiar vendor promise that lower latency is always better. Reflex, Anti-Lag 2, and XeLL can all improve responsiveness when the GPU is the limiting factor, but the review found uneven game support and uneven costs. XeLL reportedly reduced latency in every tested title but caused the harshest deterioration in percentile performance in some games; Reflex was generally better behaved; Anti-Lag 2 had the mildest frame-pacing penalty but also the smallest average latency reduction.
That does not establish a universal law that every latency-reduction mode harms frame pacing. It establishes something more practical: these technologies manipulate CPU-to-GPU queueing, so they can alter the timing consistency of delivered frames. A setting that lowers measured latency while making 1 percent lows less stable may be a poor choice in a competitive shooter, particularly if the game already has an inconsistent frame-time profile.
Reflex Still Has the Deployment Advantage
The larger split is availability. ComputerBase has previously documented that NVIDIA Reflex is tightly coupled to DLSS Frame Generation integrations, while AMD’s comparable technology has not been a mandatory companion to FSR Frame Generation. AMD’s own software direction has also been unsettled: ComputerBase reported in March that Anti-Lag 2 was being repositioned as FSR Latency Reduction 2.0.
For Radeon owners, that means “the game supports FSR” does not mean it exposes the latency control needed to offset frame generation’s responsiveness penalty. For Intel users, XeLL support remains still less common, even though newer versions can potentially operate across GPU brands.
The practical recommendation is therefore less glamorous than toggling the newest option: test the mode in the specific game. Enable the latency reducer, watch frame times and 1 percent lows, and judge mouse response in the situations that matter—dense combat, fast camera pans, and GPU-bound scenes. A latency chart without a frame-time chart is incomplete evidence.
Radeon RX 9070 XT’s Extra Performance Is a Pricing Question
Hardware Unboxed’s 52-game Radeon RX 9070 versus RX 9070 XT comparison puts the XT about 12 percent ahead on average, according to The FPS Review. That is a believable separation for two products based on the same Navi 48 family with a specification and clock-speed gap, but it is not wide enough to decide a purchase by itself.
A 12 percent lead is valuable only if the street-price premium is smaller than the performance premium, or if the extra headroom crosses a specific threshold: maintaining 60 FPS at 4K, holding a high-refresh target at 1440p, or avoiding a settings reduction in a demanding game. The recent US GPU market has been volatile enough that launch MSRPs are a poor guide to this calculation.
Buyers should compare actual checkout prices, not model names. If the XT costs roughly 10 to 12 percent more than a comparable RX 9070 board, the faster card is the straightforward choice. If its premium moves materially above that level, the non-XT becomes the more rational version of the same GPU.
Samsung’s 6K Monitor Exposes a Display-Pipeline Limit
Hardware Unboxed also tested the Samsung Odyssey G8 G80HS, the 32-inch 6K 165Hz display Samsung calls the industry’s first 6K gaming monitor. Samsung specifies a 6,144-by-3,456 panel with a 165Hz native mode and a 3K/330Hz dual mode, and the pixel math explains why current graphics hardware struggles: native 6K renders roughly 2.6 times as many pixels as 4K.
The important constraint is not only GPU throughput. The FPS Review notes that GeForce RTX 40-series cards cannot drive the monitor at 6K and 165Hz because they lack the necessary DisplayPort 2.1 output capability. An RTX 4090 owner can have enormous raster performance and still be limited by the display engine.
Hardware Unboxed’s conclusion that even an RTX 5090 cannot consistently run current games at maximum settings at 6K is unsurprising. The useful purchasing conclusion is sharper: the G80HS is a premium desktop and creator display that can also game well in its lower-resolution dual mode. It is not evidence that 6K high-refresh PC gaming has arrived as a practical native-resolution target.
Cooling, Cases, and Power Hardware Are Becoming More Specialized
be quiet!’s Light Loop IO LCD 360 is a conventional liquid-cooling platform with a modern control-and-display package wrapped around it. Reviews from OC3D, Club386, and KitGuru, collected by The FPS Review, broadly found performance in line with the existing Light Loop range rather than a generational thermal leap.
That is not a flaw, but buyers should recognize what the premium buys: a 2.1-inch IPS display, cleaner wiring through the new IO Controller, and Light Wings LX fans. Club386’s Ryzen 9 9950X3D testing also reinforces a more basic point: a 360mm AIO remains adequate for a high-end desktop CPU, but users running an unlocked 16-core processor at sustained heavy load should not expect a screen-equipped 360mm model to outperform larger 420mm coolers simply because it costs more.
InWin’s Covalent targets a different problem. Club386’s review describes an 81-liter full tower with SSI-EEB support, room for substantial drive storage, and enough airflow for a workstation-class configuration. That makes it relevant to a narrow but growing group: users assembling local AI, rendering, storage, or multi-expansion-card machines that no longer fit comfortably in enthusiast mid-towers.
The week’s two power-supply reviews underline how divergent high-wattage PC builds have become. Igor’s Lab found MSI’s 1,600W MPG Ai1600TS PCIE5 technically capable, but its most notable feature is GPU Safeguard+, which monitors distribution across its two native 12V-2×6 connectors and can trigger alerts or reduce a supported GPU’s power limit through MSI software.
That is a meaningful safety feature, but also a tacit acknowledgement that the 12V-2×6 connector remains an area where users need monitoring rather than blind confidence. The FSP DAGGER PM 1200W reviewed by Vortez is arguably the more broadly useful product: a Platinum-rated SFX-L supply with a native 12V-2×6 lead and enough capacity for a compact RTX 5080 or RTX 5090 system. Small-form-factor builders need high output density far more often than ordinary gaming PCs need 1,600W.
Lexar’s Single-Package SSD Is a Manufacturing Change, Not a Speed Breakthrough
Lexar’s Play X 1TB is the most unusual storage product in the roundup because its controller, NAND, and associated components are integrated into one package rather than distributed across a conventional populated M.2 PCB. Longsys, Lexar’s parent company, says the mSSD packaging approach can reduce manufacturing cost and solder-joint failure risk.
ComputerBase’s testing, summarized by The FPS Review, found that the concept works and reaches Lexar’s claimed sequential speeds, but the Play X does not lead the PCIe 4.0 performance field. Its random-read performance, copy behavior, and latency keep it below stronger conventional drives using similar controller hardware.
The appeal is therefore form factor and power efficiency. The module can fit compact systems and ships with an adapter for a standard M.2 2280 slot, making it potentially useful for people moving storage between handhelds, mini PCs, and desktops. Its long-term serviceability is the catch: with every major component enclosed in a single package, an internal failure means replacement rather than board-level diagnosis or component salvage.
Value Boards and Game Support Carry Their Own Caveats
Tom’s Hardware found Maxsun’s Terminator B850M Pro II compelling only near its $199 target, despite an unusually rich AM5 feature list that includes Wi-Fi 7, 2.5GbE, and multiple M.2 slots. TweakTown reached a more favorable conclusion on MSI’s $199 Z890 Gaming Plus WiFi6E, praising its expansion options and VRM thermals while recognizing that Intel’s LGA 1851 platform has no obvious upgrade runway.
The shared lesson is that inexpensive motherboards have become more capable, but buyers need to look past the headline I/O list. M.2 placement beneath a graphics card, USB speed ceilings, audio compromises, and socket longevity are not minor details once a build is assembled.
ComputerBase’s Hell Let Loose: Vietnam testing points to a similarly practical platform split. The Unreal Engine 5.5.4 game reportedly ships with DLSS support but without FSR or XeSS at launch, leaving Radeon and Arc users dependent on TSR until promised support arrives. Its analysis also found that hardware Lumen has a gameplay implication: disabling it does not remove Lumen but substitutes software Lumen, which can make jungle and indoor scenes darker enough to affect enemy visibility.
G.SKILL’s Trident Z5 NeoX RGB DDR5-6000 EXPO ULL kit brings the Weekender back to a familiar AM5 reality. Tom’s Hardware and KitGuru both found that AMD’s tighter EXPO ULL sub-timings can offer small gains on conventional Ryzen processors, but neither found a meaningful gaming benefit on X3D chips. For Ryzen X3D owners, paying extra for a ULL profile is difficult to justify unless the kit is priced essentially the same as a standard low-latency DDR5-6000 alternative.
The common thread across this week’s hardware is that the advertised feature is rarely the deciding factor. The better purchase depends on whether the feature changes the limitation in your system: frame-time consistency rather than a latency checkbox, DisplayPort capability rather than raw shader power, connector monitoring rather than wattage, or memory tuning that actually benefits the CPU in the socket.
