Two Mid-Range GPUs, One Very Specific Fight
AMD’s Radeon RX 9070 and Nvidia’s RTX 5060 Ti are both targeting the same buyer: someone with a 1440p monitor who wants solid ray tracing without spending flagship money. On paper, that sounds like a clean matchup. In practice, the gap between them at ray-traced workloads is wider than the price tags suggest.
The Hardware Reality Before the Benchmarks
The RX 9070 launches with 16GB of GDDR6 memory on a 256-bit bus, backed by AMD’s RDNA 4 architecture. RDNA 4 represents AMD’s most aggressive reworking of its ray tracing hardware yet, with doubled Ray Accelerators per compute unit compared to RDNA 3. That matters because previous AMD generations consistently fell behind Nvidia in ray-traced titles, not because of raw compute, but because the RT hardware simply couldn’t keep pace with Nvidia’s dedicated RT cores. RDNA 4 was built specifically to close that deficit.
The RTX 5060 Ti, running on Blackwell, carries either 8GB or 16GB of GDDR7 depending on the configuration – and that memory tier difference is worth calling out directly. Blackwell’s 4th-generation RT cores handle ray-box and ray-triangle intersection tests faster than any previous Nvidia mid-range design, and DLSS 4 with Multi Frame Generation adds a performance layer that AMD’s FSR 4 doesn’t yet match at the same level of image stability. Nvidia has spent three generations building a ray tracing pipeline that is genuinely hard to compete with at this price bracket.
Testing was done at 1440p across a set of titles with demanding ray tracing implementations: Cyberpunk 2077 with path tracing disabled but full RT Overdrive-adjacent settings active, Alan Wake 2, Black Myth: Wukong with RT reflections enabled, and Indiana Jones and the Great Circle. These aren’t synthetic benchmarks – they’re the games where ray tracing visibly changes what the player sees on screen. The test system used a Ryzen 7 7800X3D to keep CPU overhead out of the equation.
Frame generation was tested separately from native performance. Running native rasterization plus ray tracing without any upscaling or frame gen tells you what the GPU actually costs in raw compute terms. Then FSR 4 and DLSS 4 were layered in to show real-world playable numbers. That distinction matters because marketing tends to blur the line between native GPU performance and AI-assisted frame output.
1440p Ray Tracing: Where Each Card Wins and Loses
In Cyberpunk 2077 at 1440p with RT Ultra settings and no upscaling, the RTX 5060 Ti holds a consistent lead – roughly 15 to 20 percent faster in native frame rates depending on the scene. That gap has always existed between Nvidia and AMD at ray-traced workloads, but what’s changed with RDNA 4 is that the RX 9070 no longer falls apart under pressure. Previous RDNA 3 cards would stutter heavily in RT-heavy scenes because the hardware was doing too much in software fallback. The RX 9070 mostly holds its frame pacing together, even if the raw average is lower.
Alan Wake 2 is traditionally Nvidia’s strongest showcase, and the RTX 5060 Ti keeps that advantage at 1440p. Full ray tracing in Remedy’s title is demanding to a degree that makes the raw performance gap between these two cards feel larger than in any other game tested. The 5060 Ti averages noticeably higher frame rates here, and the 1% lows from the RX 9070 show some inconsistency that doesn’t fully disappear even with FSR 4 engaged. This is the one scenario where AMD’s improved RT hardware still hasn’t caught up.
Black Myth: Wukong tells a different story. With RT reflections active at 1440p, the RX 9070 closes the gap significantly – in some benchmark sequences, the difference between the two cards drops to within 5 to 8 percent. The game’s particular implementation of ray traced reflections appears to favor AMD’s newer RT pipeline more than the shader-heavy Cyberpunk workload does. This suggests the “AMD is bad at RT” narrative that carried over from RDNA 3 needs some qualification at this point. RDNA 4 is selectively competitive, not universally behind.
With DLSS 4 enabled on the RTX 5060 Ti and FSR 4 on the RX 9070 – both at Quality mode, 1440p output – the performance picture shifts in ways that favor Nvidia more than the native numbers alone do. DLSS 4’s Transformer model produces sharper, more stable output than FSR 4 in fast-moving RT scenes, and Multi Frame Generation adds frames that feel more natural than previous implementations. The RTX 5060 Ti with DLSS 4 at Quality mode reaches frame rates in Cyberpunk 2077 that the RX 9070 with FSR 4 simply can’t match. If upscaling is part of how you plan to play, that gap widens rather than narrows.
The memory situation on the 8GB RTX 5060 Ti variant is a genuine concern for 1440p ray tracing specifically. RT workloads push VRAM consumption significantly higher than standard rasterization, and 8GB at 1440p with RT enabled in Alan Wake 2 or Cyberpunk 2077 leads to stuttering that the 16GB RX 9070 avoids entirely. Anyone considering the 8GB 5060 Ti for a ray tracing-focused build should factor that in. The 16GB 5060 Ti variant removes that problem but also costs more, at which point the value comparison with the RX 9070 gets tighter than Nvidia’s ray tracing advantage might justify. For more on how AMD’s current GPU lineup stacks up against Nvidia’s higher-tier cards, the Radeon RX 9070 XT vs RTX 5070 4K rasterization comparison covers that territory in detail.
The Honest Verdict on a Messy Matchup
The RTX 5060 Ti is the faster ray tracing card at 1440p. That’s not close enough to debate – Nvidia’s RT architecture, combined with DLSS 4, gives it a real and consistent advantage across the majority of titles that use ray tracing meaningfully. If ray tracing performance is the primary reason you’re buying a GPU right now, the 5060 Ti in 16GB form is the cleaner choice.
What the RX 9070 does is force a harder question about how much that RT advantage is worth paying for. The 16GB RX 9070 undercuts the 16GB RTX 5060 Ti at most retail price points, offers equal or better rasterization performance in non-RT titles, and eliminates the VRAM anxiety that haunts the 8GB 5060 Ti. AMD’s ray tracing deficit is real but no longer disqualifying – and the card it’s being compared against has a known weakness that AMD doesn’t share.
