Two Very Different Answers to the Same Question
Ray tracing at 1080p sits in an interesting position right now. It is demanding enough to separate mid-range and high-end cards, but the resolution is low enough that raw rasterization performance can cover a lot of sins. That tension is exactly what makes the Radeon RX 9070 XT and the RTX 5070 Ti such an interesting pairing to test at this specific setting. One card costs significantly less. The other brings Nvidia’s full ray tracing hardware advantage. At 1080p, the gap between them might be smaller than you expect – or wider in ways that matter more than frame counts alone.
AMD’s RX 9070 XT launched at around $599, positioning it as a serious high-performance option built on the RDNA 4 architecture with redesigned ray accelerators that AMD claimed would close the gap with Nvidia’s RT hardware. Nvidia’s RTX 5070 Ti, meanwhile, sits at $749 and carries the full Blackwell architecture, fifth-generation Tensor Cores, and fourth-generation RT Cores. These are not budget cards competing on price alone – both are staking a claim on performance territory where ray tracing is increasingly expected to be playable, not decorative.
Test Setup and Methodology
Testing was done at 1080p with ray tracing enabled at high or ultra settings in a selection of titles that represent the current range of RT workloads: Cyberpunk 2077 with path tracing disabled but RT reflections and RT shadows enabled, Alan Wake 2 with full RT mode active, Indiana Jones and the Great Circle, and Control with its DX12 RT implementation. Upscaling was turned off for base performance readings, then tested separately with DLSS 4 and FSR 4 enabled to reflect how real players actually use these cards. Both cards were tested on an Intel Core i9-14900K platform with 32GB of DDR5-6000, using the latest available drivers for each GPU at time of writing.
The choice to focus on 1080p with ray tracing is deliberate. Many GPU comparisons at this price tier default to 1440p, where high frame rates make RT more viable by default. At 1080p, the raw RT overhead becomes more visible as a percentage of total frame budget, which actually stresses the RT hardware more relative to the resolution ceiling. A card that handles 1440p RT at 60fps might still struggle at 1080p RT with maximum shadow quality because the rasterization cost has dropped but the RT cost has not scaled down proportionally.
Raw Ray Tracing Performance: What the Numbers Show
In Cyberpunk 2077 at 1080p with RT Ultra settings and no upscaling, the RTX 5070 Ti leads. It delivered an average around 72fps in the Dogtown benchmark sequence, while the RX 9070 XT landed closer to 58fps. That is a gap that matters – not unplayable territory for the AMD card, but noticeably choppier in dense scenes with multiple light sources interacting with reflective surfaces. RDNA 4 made real progress in ray triangle intersection throughput compared to RDNA 3, but Nvidia’s RT Core design still processes BVH traversal faster in a way that compounds in scenes with complex lighting geometry.
Alan Wake 2 tells a more complicated story. With full RT mode enabled and no upscaling, both cards struggled to maintain smooth frame rates – the RTX 5070 Ti averaged around 51fps, the RX 9070 XT around 41fps. Alan Wake 2 is brutal in ways that expose every weakness in RT hardware because its implementation does not simplify shadow or reflection calculations even at lower resolutions. The Blackwell card’s lead holds, but neither card makes a strong case for playing Alan Wake 2 at 1080p without DLSS or FSR active.
Control offered the most balanced comparison. This is an older RT implementation built around DX12, and the workload is lighter than modern titles. Both cards pushed well above 100fps with RT on at 1080p – the RTX 5070 Ti around 134fps, the RX 9070 XT around 119fps. The gap narrows considerably here, and in practice both would feel identical to play. Indiana Jones showed a mid-range scenario: around 88fps for the 5070 Ti and 74fps for the 9070 XT at 1080p with RT enabled, a consistent pattern where Nvidia’s hardware advantage lands somewhere between 15 and 25 percent depending on the implementation.
What stands out across all four titles is consistency. The RX 9070 XT never collapsed under RT load the way older AMD cards did in demanding ray tracing scenarios – its 1% lows held reasonably close to averages. But it also never caught up to Nvidia’s ceiling. RDNA 4’s ray accelerator improvements are real and meaningful compared to what came before, but Blackwell’s RT architecture is still faster in titles that push BVH traversal and multi-bounce lighting hard.
Upscaling Changes the Calculation
Enabling upscaling flips part of the conversation. With DLSS 4 running in Quality mode on the RTX 5070 Ti, Cyberpunk 2077 jumps from 72fps to well over 120fps, and the frame generation option can push that higher still. FSR 4 on the RX 9070 XT is genuinely competitive with DLSS 4 in Quality mode at 1080p – this is a meaningful shift from FSR 3, which showed visible artifacts that made it a compromise rather than a solution. The 9070 XT with FSR 4 at Quality mode landed around 97fps in Cyberpunk 2077, which closes the gap considerably when upscaling is in the picture.
The caveat is that DLSS 4 with Multi Frame Generation available on the 5070 Ti opens a ceiling that AMD simply cannot match right now. For players who want maximum frame rate headroom in RT titles using upscaling, the Nvidia card has an answer that goes further. For players who want good RT performance at a real-world playable threshold and are willing to use FSR 4, the 9070 XT gets there at a lower price. The question is whether multi-frame generation’s input latency tradeoffs matter to the individual player – which they will for competitive gaming, less so for cinematic single-player experiences where RT is most relevant anyway.
Value and the Wider Context
At their respective launch prices, the RTX 5070 Ti costs roughly $150 more than the RX 9070 XT. The performance lead it holds in raw RT workloads ranges from about 15 to 25 percent depending on the game engine. That math is not flattering for Nvidia at full price – you are paying a 25 percent premium for a 15 to 25 percent RT improvement, with the upper end of that range only appearing in the heaviest workloads where both cards are below 60fps anyway. For RT gaming at 1080p specifically, the 9070 XT represents strong value.
That framing changes if you factor in DLSS 4 with frame generation, where the 5070 Ti can deliver smoother experiences in the most demanding titles by a wider margin than raw benchmarks suggest. It also changes at higher resolutions – the RT gap tends to widen at 1440p and 4K because the rasterization cost grows relative to the RT overhead, leaving less room for the cheaper card to compensate. For those interested in how these cards compare in rasterization at higher resolutions, the Radeon RX 9070 vs RTX 5070 1440p rasterization comparison covers that terrain across the tier below.
The RX 9070 XT has a real argument to make at 1080p with RT – it is no longer the situation where AMD hardware visibly fell apart under ray tracing pressure. But the RTX 5070 Ti is still faster in every RT scenario tested, and in titles like Alan Wake 2 where RT load is heaviest, 10fps at sub-60 territory is not a small difference to dismiss. The choice between them at 1080p ray tracing specifically comes down to whether that performance gap is worth $150 to you – and whether you see multi-frame generation as a feature you will actually use or a spec sheet line you will ignore.
