ASUS DUAL RTX5070 OC Graphics Card Review UK 2026

I’ve spent the last few weeks pushing the ASUS DUAL RTX 5070 OC through its paces, and honestly? This is the sort of GPU that makes you question whether you actually need to spend flagship money anymore. NVIDIA’s 50-series launch has been… interesting, let’s say. The usual hype, the usual claims about revolutionary performance. But after benchmarking this card across everything from Cyberpunk 2077 to Counter-Strike 2, I’ve got some proper thoughts about where it sits in the current GPU landscape. If you’re staring at your aging RTX 3070 or RX 6700 XT wondering whether it’s finally time to upgrade, this review’s for you.

Now, here’s where things get interesting. That 192-bit bus is narrower than the RTX 4070’s 256-bit configuration, which had some people worried about memory bottlenecks. In practice? It’s mostly fine at 1440p, but you do notice it occasionally in 4K scenarios with maximum texture quality. More on that later.

The Time Spy score puts it roughly 25% ahead of the RTX 4060 Ti and about 15% behind the RTX 4070 Ti. Port Royal results show the ray tracing improvements from the newer RT cores, with a meaningful jump over last-gen equivalents. For content creators, the Blender performance is genuinely impressive, making this a viable option for GPU rendering work alongside gaming.

But synthetic benchmarks are just that. Synthetic. What matters is how this card actually performs in the games you’re going to play.

At 1080p, this card is frankly overkill for most titles. You’re hitting well over 100 FPS in demanding AAA games, and competitive shooters are pushing 300+ frames. If you’re primarily gaming at 1080p, save your money and buy something cheaper.

1440p is where the RTX 5070 really shines. Cyberpunk 2077 with ray tracing on Ultra (but no DLSS yet) delivered a very playable 94 FPS average. That’s proper smooth gaming in one of the most demanding titles out there. Starfield, which has been… let’s say ‘optimisation challenged’ since launch, managed 76 FPS average with occasional dips into the high 60s in New Atlantis. Still perfectly playable.

4K native is where things get more challenging. Cyberpunk 2077 dropped to 47 FPS, which is technically playable but not ideal. Starfield at 38 FPS felt choppy in busy areas. This is where DLSS becomes essential rather than optional.

With DLSS Quality mode enabled at 4K, performance jumped dramatically. Cyberpunk went from 47 to 89 FPS. Starfield climbed to 71 FPS. Alan Wake 2, which was struggling at 31 FPS native, hit a smooth 68 FPS with DLSS Quality. The image quality hit is minimal, honestly. I had to pixel-peep to spot the difference in most scenes.

So what’s the verdict on gaming performance? At 1440p, this card is brilliant. Proper high-refresh gaming in basically everything. At 4K, you’ll want DLSS enabled for demanding titles, but it’s absolutely capable of delivering smooth 60+ FPS experiences. Just don’t expect to max out every setting in 2026’s most demanding games at native 4K.

Portal RTX, which is basically a ray tracing torture test, ran at 41 FPS at 1440p with DLSS Quality and Frame Generation. Without those technologies? An unplayable slideshow at 18 FPS. This is the sort of scenario where NVIDIA’s AI upscaling tech justifies the price premium over AMD alternatives.

The Ray Reconstruction feature in DLSS 3.5 also improves image quality in ray traced scenes by using AI to denoise and reconstruct the lighting. In Alan Wake 2, this made a noticeable difference to shadow quality and reduced the grainy artifacts you sometimes get with standard ray tracing denoisers.

During testing, I monitored VRAM usage across all games. At 1440p Ultra, the highest I saw was 9.8GB in Hogwarts Legacy. At 4K Ultra, Resident Evil 4 Remake with the HD texture pack hit 11.2GB, and I noticed occasional texture streaming stutters when quickly turning the camera. Dropping texture quality one notch solved it entirely.

So is 12GB enough? For 1440p gaming, absolutely. For 4K, it’s adequate but not abundant. You’ll occasionally need to compromise on texture quality in the most demanding titles, which is a bit frustrating in this price bracket.

Gaming temperatures settled around 68°C for the GPU core, which is perfectly reasonable. The hotspot peaked at 77°C during extended gaming sessions, which is well within safe operating limits. Memory junction temps hit 72°C, which is fine for GDDR6X (these modules are rated to run much hotter than that).

The fans spin up to around 1,850 RPM under load, which brings us to noise levels.

At idle, the fans stop completely thanks to the 0 RPM mode, which is lovely for desktop work. Under gaming load, the card sits at around 37 dB measured from 50cm away. That’s audible if you’re not wearing headphones, but it’s not annoying. The fan profile is reasonably well-tuned, without the aggressive ramp-up behaviour that makes some cards sound like they’re preparing for takeoff.

During stress testing (FurMark, which pushes power draw higher than any real game), noise peaked at 42 dB. That’s getting into ‘properly audible’ territory, but you’ll never hit these levels in actual gaming scenarios.

No coil whine on my sample, which is always a relief. That’s not guaranteed unit-to-unit, but ASUS generally does a decent job with component quality.

NVIDIA recommends a 650W PSU for system builds with the RTX 5070, which is sensible. I tested with a Corsair RM750x (80+ Gold) and saw total system power draw peak at around 420W during gaming with the Ryzen 7 7800X3D. A quality 650W unit would handle this fine, but I’d personally recommend 750W for headroom and efficiency. The 16-pin connector handles transient spikes without issue, despite all the drama around melting connectors with previous generations.

Idle power draw is pleasantly low at 12W thanks to decent power management. The card drops into low-power states properly when you’re just browsing or watching videos.

Gaming power draw averaged 218W across my test suite, which is pretty much bang on the rated TDP. I did see brief spikes to 245W during scene transitions and loading screens, which is normal behaviour. Make sure your PSU can handle these transients without tripping over-current protection.

The card uses a single 16-pin 12VHPWR connector, with an adapter cable included for traditional 8-pin PSU cables. The connector seated firmly with a satisfying click, and I had zero issues during several weeks of testing. Just make sure you push it in properly until you hear the click.

Display outputs are three DisplayPort 1.4a ports and one HDMI 2.1 port. That’s the standard configuration, and it’s perfectly adequate for multi-monitor setups.

I tested streaming to Twitch at 1080p60 using OBS while playing Cyberpunk 2077 at 1440p. The performance hit was minimal (maybe 3-4 FPS), and the stream quality with AV1 at 6,000 kbps looked noticeably better than H.264 at the same bitrate.

For video editing, the 12GB VRAM handles 4K timelines in DaVinci Resolve without issues. Rendering a 10-minute 4K video with effects and colour grading took about 8 minutes, which is respectable. Not as fast as the RTX 4080, but perfectly usable for hobbyist creators.

Against the AMD RX 7800 XT, the RTX 5070 trades blows in rasterization performance (NVIDIA edges slightly ahead) but pulls significantly ahead in ray tracing scenarios. The 7800 XT’s 16GB of VRAM is genuinely appealing for 4K gaming, though. If you don’t care about ray tracing or DLSS, the AMD card offers excellent value.

Compared to the RTX 4070, the 5070 is about 8-10% faster in most games thanks to the architectural improvements and higher boost clocks. The price difference isn’t massive, so I’d recommend the newer card unless you find a particularly good deal on last-gen stock.

At the upper mid-range tier, you’re getting genuinely excellent 1440p performance and viable 4K gaming with upscaling enabled. This is the sweet spot for most enthusiast gamers who want high-refresh 1440p or smooth 4K experiences without spending flagship money. The tier below (mid-range, roughly £350-500) will handle 1440p but struggles at 4K. The tier above (high-end, £500-750+) offers more headroom for native 4K but with diminishing returns on value. This RTX 5070 sits right at that inflection point where you’re getting most of the performance without the exponential price increase.

The RTX 5070 delivers about 85% of RTX 4080 performance whilst costing significantly less. That’s solid value in the GPU market, where the price-to-performance curve gets brutal at the high end. For 1440p gaming, it’s arguably overkill. For 4K, it’s adequate with DLSS. That makes it ideal for gamers who want one card that can handle both resolutions depending on the game and their monitor setup.

Compared to previous generations at launch, the pricing is… fine. Not amazing, not terrible. We’re not back to the scalper-era madness of 2021-2022, but we’re not at the pre-mining-boom value levels either. It’s competitive with AMD’s offerings at this performance tier, which is what matters most.

For 4K gaming, it’s more complicated. You’ll get smooth 60+ FPS experiences in most games with DLSS Quality enabled, but native 4K performance requires settings compromises. The 12GB VRAM is adequate now but might feel restrictive in 2-3 years as texture quality continues to increase.

The price positioning in the upper mid-range bracket feels about right. It’s not the value champion of the generation, but it’s competitive with AMD alternatives whilst offering superior ray tracing and AI upscaling technology. If you care about those features (and in 2026, you probably should), the NVIDIA premium is justified.