Sapphire Pulse RX 9060 XT 16GB Review: The Best 1440p Mid-Range GPU in 2025?

Picking a GPU in 2025 is less about finding the best card and more about not getting ripped off. The mid-range is a mess of overlapping price points, rebranded architectures, and cards that were genuinely good value eighteen months ago but aren't anymore. So when AMD launched the RX 9060 XT on their new RDNA 4 architecture, I wanted to know one thing: does it actually earn its place in the market, or is it just filling a slot?

The Sapphire Pulse variant is the one most people will end up buying. Sapphire has been making AMD cards for years and the Pulse line sits in that sensible middle ground between the stripped-back reference designs and the overbuilt, overpriced Nitro+ cards. I've been running this card for about a month now, through a proper rotation of games, some light streaming, and the usual day-to-day use. The 16GB GDDR6 version is what I've been testing, and that VRAM figure is going to be a significant part of the conversation.

Before I get into the numbers, it's worth being clear about where this card sits. This isn't a flagship. It's not trying to compete with the RX 9070 XT or the RTX 5070. It's a 1440p card aimed at people who want solid performance without spending north of £399.98. Whether it delivers that is what this review is actually about.

The 16GB of GDDR6 sits on a 128-bit memory bus, giving you 288 GB/s of bandwidth. That bus width is narrower than some competitors and it's something I'll come back to in the VRAM section, because bandwidth and capacity aren't the same thing. Display outputs are four in total: two DisplayPort 2.1 and two HDMI 2.1, which covers most setups including high refresh rate 1440p and 4K displays. The card connects via PCIe 5.0 x8, though it's backwards compatible with PCIe 4.0 and 3.0 systems without meaningful performance loss.

Physically, the Sapphire Pulse is a dual-slot, dual-fan card. It measures around 240mm in length, which means it'll fit in most mid-tower cases without any drama. There's a single 8-pin power connector, which is refreshingly straightforward compared to the 16-pin adapters on some Nvidia cards. The card doesn't sag noticeably in a standard motherboard slot, though if you're running a particularly heavy case orientation you might want a support bracket.

The AI accelerators are also new. RDNA 4 introduces dedicated AI compute hardware, which AMD calls the AI accelerators, and these are what power FidelityFX Super Resolution 4 (FSR 4) and the new AMD Neural Radiance Caching features. FSR 4 is a machine learning-based upscaler, a genuine step up from the spatial FSR 2 and FSR 3 implementations. The 9060 XT has 32 AI accelerators, one per Compute Unit. Whether that's enough to match Nvidia's Tensor core count for DLSS is a separate question, but it's a proper hardware implementation rather than a software workaround.

The shader array is organised into four shader engines, each with eight Compute Units. Clock speeds are high relative to the shader count, which is AMD's approach with RDNA 4: fewer CUs running faster and more efficiently rather than brute-forcing it with a massive die. The 4nm TSMC process helps here. Transistor density is up significantly over the 6nm node used in RDNA 3 mid-range parts, and you can see that in the power consumption figures. This is a 150W card doing work that would have needed 200W+ a generation ago.

The base clock is listed at 1607 MHz, but you'll essentially never see that in a gaming scenario. Base clocks on modern AMD cards are more of a floor for extreme thermal situations than a realistic operating point. What matters is where the card actually runs under load, and the answer here is solidly in the upper 2800s. That's a high clock speed for a mid-range card and it contributes to the snappy feel in lighter workloads where the card isn't fully saturated.

I did try pushing the card a bit further with AMD's Radeon Software overlay. There's headroom there, maybe another 50-80 MHz on the boost clock before you start needing to increase the power limit. But honestly, the factory OC is sensible and the gains from pushing further are marginal. If you're the type who wants to squeeze every last frame, the headroom exists. If you just want to plug it in and play, the out-of-box clocks are fine. The Radeon Software suite itself has improved a lot over the past couple of years, and the AMD Adrenalin driver package makes tuning straightforward if you want to go down that road.

The caveat is the 128-bit bus. More VRAM is good, but bandwidth matters too. At 288 GB/s, the 9060 XT has less bandwidth than some competing cards with wider buses. In practice, this shows up most in scenarios where you're pushing very high resolution textures or running at 4K. The card can handle 4K gaming, but it's not where it's most comfortable. At 1440p, the bandwidth is sufficient and the 16GB capacity means you're not going to run into texture streaming issues in any current title. At 1080p, it's overkill in the best possible way.

Looking ahead, 16GB future-proofs this card meaningfully. We're already seeing next-generation titles pushing VRAM harder, and the trend isn't reversing. A card you buy today needs to last two or three years to represent good value. The 9060 XT's VRAM situation in 2027 looks a lot better than an 8GB card bought at the same price point today. That's not a small thing. I've seen too many people regret buying 8GB cards that became bottlenecked within eighteen months, and the 16GB variant here addresses that concern directly. JEDEC's GDDR6 specification is mature and well-understood, so there are no surprises in terms of reliability.

FSR 4 is the upscaling story here, and it's genuinely good. This is AMD's machine learning-based upscaler, and it requires RDNA 4 hardware to run. At Quality preset, the image quality is noticeably better than FSR 3.1, with less ghosting and sharper fine detail. It's not quite DLSS 4 with Multi Frame Generation, but it's competitive with DLSS 3.5 in most scenarios. The important thing is that FSR 4 is available in a growing list of titles, and AMD has been pushing developer adoption hard. AMD's GPUOpen FSR documentation shows the current list of supported games, which is expanding regularly.

For games that don't support FSR 4 yet, FSR 3.1 and FSR 2 are still available and the 9060 XT runs them fine. XeSS also works on AMD hardware, though without the Intel-specific optimisations. DLSS is Nvidia-only, full stop, so if you're heavily invested in DLSS titles, that's a genuine consideration. But the gap between FSR 4 and DLSS 4 is smaller than it's ever been, and for most gaming scenarios at 1440p with Quality or Balanced preset, FSR 4 produces results you'd be happy with. Frame Generation is also supported via FSR 3, adding interpolated frames on top of the native rendering, which helps in CPU-bound scenarios particularly.

In practical terms, I tested streaming via OBS with AMD's AMF encoder at 1440p, 6000 kbps, AV1. The results were solid. No dropped frames, clean image quality, and the encoding overhead on the GPU was minimal enough that gaming performance wasn't meaningfully affected. If you're a streamer who's been relying on CPU encoding because your old GPU's encoder was rubbish, this is a proper upgrade. The AMD Advanced Media Framework handles the encoding pipeline and it's well-supported in OBS, Streamlabs, and most major streaming tools.

Decode support is equally strong. AV1 decode acceleration means YouTube 4K AV1 streams, which are increasingly common, play back without taxing the CPU. If you're running a Ryzen system where the CPU doesn't have strong integrated decode, offloading to the GPU matters. H.264, H.265, and VP9 are all hardware-accelerated as well. For content creators doing light video editing, the hardware decode acceleration in Premiere Pro and DaVinci Resolve will be useful, though this card isn't positioned as a workstation GPU and I wouldn't buy it primarily for that purpose.

For PSU recommendations, a 550W unit is plenty for a mid-range system with this card. If you're running a Ryzen 7 or Core i5 class CPU, 550W gives you comfortable headroom. You don't need a 750W PSU for this build, which is worth knowing because PSU costs add up. The lack of a 16-pin connector is also a practical benefit: no adapters, no concerns about the connector seating properly, no drama. I've had enough conversations with people who've had issues with 16-pin adapters on Nvidia cards to appreciate the simplicity here.

The efficiency story extends to idle power as well. At desktop, the card drops to around 5-8W in its low-power state, which is normal for modern GPUs with good power gating. The fan stops completely at idle and light loads (more on that in the acoustics section), so the card is genuinely quiet when you're not gaming. Over the course of a month of testing, my electricity usage didn't spike noticeably compared to running a 200W+ card. At UK electricity prices, that difference adds up over a year of gaming.

I ran a sustained stress test for about an hour to check for thermal throttling. The card maintained its boost clocks throughout without dropping below 2800 MHz, which tells you the cooling is adequate for the TDP. In a well-ventilated mid-tower case, you're not going to have thermal issues with this card. I did test it in a smaller mATX case with restricted airflow, and temperatures climbed to around 80°C GPU and 95°C hotspot under load. Still within spec, but worth noting if you're building in a compact case with poor airflow.

Compared to higher-TDP cards, the thermal situation here is genuinely easier to manage. A 300W GPU in a tight case is a real problem. A 150W GPU is much more forgiving. If you're upgrading from an older card and your case cooling was marginal before, the 9060 XT is unlikely to make things worse and may actually run cooler than what you're replacing. The heatsink doesn't extend beyond the PCB, so there's no awkward overhang into adjacent slots either.

Under gaming load, the fans spin up to around 1400-1600 RPM at the temperatures I described above. At that speed, the noise level is low. I measured approximately 34-36 dB(A) at 30cm from the card, which is quiet enough that you'll hear your case fans before you hear the GPU. The fan character is smooth, no bearing whine, no rattling. Sapphire's fan quality on the Pulse line has been consistently decent in my experience and this card continues that trend.

At maximum fan speed (which you'd only hit if you manually set the fan curve or if the card was severely thermally stressed), the noise climbs to around 42-44 dB(A). That's audible but not unpleasant. In normal gaming scenarios, you won't be anywhere near that. For comparison, I've tested cards that hit 48+ dB(A) at normal gaming loads, and those are genuinely distracting. The 9060 XT is not one of those cards. If acoustic performance matters to you, this is a good choice at the price.

At 1440p native: Cyberpunk 2077 Ultra averaged 68 FPS with RT Medium enabled, dropping to around 55 FPS with RT High. With FSR 4 Quality, those numbers jump to 94 FPS and 76 FPS respectively, which is very playable. Alan Wake 2 at High settings averaged 71 FPS native, 98 FPS with FSR 4 Quality. Call of Duty: Black Ops 6 at High settings hit 142 FPS average, which is excellent for competitive play. Baldur's Gate 3 at Ultra averaged 89 FPS. At 1080p, all of these numbers go up by roughly 25-35%, so you're looking at very high framerates in most titles.

At 4K, the picture is more mixed. Cyberpunk 2077 Ultra at 4K native is around 38 FPS, which needs FSR 4 to be usable. Alan Wake 2 at 4K High is around 42 FPS native. With FSR 4 Quality at 4K, you're rendering at 1440p effectively and the results are good, but you're relying on upscaling to make 4K work. This isn't a 4K card in the traditional sense. It can do 4K with upscaling, but if you want native 4K performance you need to look at the RX 9070 XT or RTX 5070 tier. For 1440p gaming, though, this card is properly good. It's not struggling, it's not marginal, it's genuinely comfortable at that resolution.

Against the RTX 4060 Ti 16GB: the 9060 XT is faster in rasterisation performance by roughly 10-15% at 1440p, depending on the title. The 4060 Ti has DLSS 3 with Frame Generation, which is a genuine advantage in supported titles. But FSR 4 has closed the gap significantly, and the 9060 XT's lower power consumption and typically lower price make it the more sensible buy in most cases. The 4060 Ti's 128-bit bus was criticised when it launched, and the 9060 XT has the same bus width, so that's a wash.

Against the RX 7700 XT: the 9060 XT is faster by around 15-20% at 1440p, has better ray tracing performance, FSR 4 support (the 7700 XT is limited to FSR 3.1), and lower power consumption. The 7700 XT launched at a higher price than the 9060 XT typically sells for now. If you're deciding between a used 7700 XT and a new 9060 XT at similar prices, the 9060 XT is the better buy. The architecture jump is real.

The question of whether the 9060 XT 16GB is better than a used RTX 3070 comes up a lot. The 3070 has an 8GB VRAM limit, which is increasingly problematic. In raw rasterisation performance they're broadly similar, but the 9060 XT wins on VRAM, power efficiency, upscaling quality, and ray tracing. The 3070 wins on nothing in 2025 except potentially price if you find one cheap. I wouldn't buy a used 3070 over a new 9060 XT.

Is it low end? No. It's solidly mid-range, aimed squarely at 1440p gaming. Is it a 4K card? Not really, not without leaning on upscaling. Can it run 1440p well? Yes, properly well. Is the Sapphire Pulse a good choice specifically? Yes. It's one of the better cooler implementations at this price tier, the build quality is solid, and Sapphire's driver support track record is fine. The single 8-pin connector and 150W TGP make it easy to fit into existing systems without a PSU upgrade in most cases.

The main reasons to look elsewhere: if you're heavily invested in DLSS and play a lot of titles where DLSS 4 with Multi Frame Generation makes a big difference, the Nvidia side of the market is worth considering. If you want native 4K performance without upscaling, you need to spend more. And if you find the RTX 4060 Ti 16GB at a significantly lower price, it's still a reasonable card. But at its current price, the 9060 XT 16GB is the practical choice for 1440p gaming in 2025. It does what it needs to do, it does it quietly, and it won't leave you wishing you'd spent more within the next two years.

Score: 8/10. Recommended for 1440p gaming builds where value and efficiency matter. Not the card for 4K purists or DLSS devotees, but for everyone else, it's a proper option.

• You primarily play titles where DLSS 4 with Multi Frame Generation is a major factor
• You want native 4K performance without relying on upscaling
• Your budget stretches to the RX 9070 XT, which is a meaningful step up
• You find a used RTX 4070 at a price that undercuts this card significantly

• 128-bit memory bus limits bandwidth, noticeable at 4K
• Not a native 4K card without upscaling assistance
• DLSS is unavailable, FSR 4 game support still growing
• Ray tracing improved but still behind Nvidia at equivalent price points
• Sapphire Pulse cooler is adequate but not exceptional compared to premium alternatives

Review unit tested over approximately one month on a system running a Ryzen 7 7700X, 32GB DDR5-6000, and a 650W 80+ Gold PSU. All gaming benchmarks taken at 1440p unless stated otherwise, with AMD drivers version 25.x series.

This article contains affiliate links. If you purchase through these links, vividrepairs.co.uk may earn a small commission at no extra cost to you. This does not influence our editorial recommendations.