Two Cards, One Resolution, One Clear Winner?
AMD’s Radeon RX 9070 and Nvidia’s RTX 5070 are priced within striking distance of each other, target the same enthusiast-adjacent buyer, and both launched in early 2025 with significant marketing fanfare. At 1080p rasterization, the gap between them tells a story that neither company’s PR materials fully capture.
The Setup: Hardware, Context, and What We Are Actually Testing
Both cards sit in the upper-mid tier of their respective lineups. The RX 9070 is built on AMD’s RDNA 4 architecture, using a 4nm TSMC process node, with 16GB of GDDR6 memory and 56 compute units. The RTX 5070 runs on Nvidia’s Blackwell architecture, also on a 4nm node, with 12GB of GDDR7 and 48 shader multiprocessors. On paper, the memory bandwidth and capacity advantage goes to AMD, while Nvidia holds architectural efficiency gains in raw shader throughput.
Testing at 1080p is a different exercise than testing at 4K. At higher resolutions, the GPU becomes the undisputed bottleneck, making comparisons cleaner and more GPU-dependent. At 1080p, CPU performance, memory latency, and driver overhead all carry meaningful weight. For this comparison, the test system used a Ryzen 7 9800X3D to keep CPU bottlenecking as minimal as possible, paired with 32GB DDR5 at 6000MHz.
Rasterization only – no ray tracing, no upscaling, no frame generation. FSR 4 and DLSS 4 are both genuinely impressive technologies, but they obscure the underlying hardware performance. What does each card do with raw, traditional rendering pipelines at the resolution most competitive PC gamers are actually playing at? That is the only question being answered here.
The game selection spans recent titles across multiple genres: Cyberpunk 2077, Alan Wake 2, Black Myth: Wukong, Counter-Strike 2, Forza Horizon 5, and Hogwarts Legacy. These titles range from GPU-bound cinematic experiences to competitive shooters where frame rates and latency matter far more than visual fidelity. Together they give a realistic picture of what a typical gaming library demands from mid-to-high-end hardware in 2025.
Benchmark Results: Where the RX 9070 Surprises and Where the RTX 5070 Holds Ground
In Cyberpunk 2077 at 1080p Ultra settings with rasterization only, the two cards trade blows within a margin that is effectively meaningless at this resolution – both cards are partially CPU-limited, and frame rates push well above 120fps for both. The RTX 5070 edges ahead by around five to eight percent in most passes, which translates to perhaps four or five extra frames per second in a range where the human eye is not going to notice the difference.
Black Myth: Wukong is where RDNA 4 starts showing its rasterization muscle more clearly. AMD’s architectural improvements in compute unit efficiency on RDNA 4 are not subtle – they represent a genuine generational leap over RDNA 3. At 1080p with High settings and no upscaling, the RX 9070 consistently matches the RTX 5070 and occasionally pulls marginally ahead. Given that the RX 9070 typically retails for less than the RTX 5070 at MSRP, this result carries real-world weight for buyers watching their budgets.
Counter-Strike 2 flips the narrative. At 1080p with competitive settings, Nvidia’s driver stack and frame-pacing consistency have historically been tighter for esports titles, and the RTX 5070 demonstrates that here. Average frame rates for both cards are absurdly high – well beyond what any monitor can display – but the 1% and 0.1% lows favor Nvidia noticeably. For a player running a 360Hz monitor and playing at a high competitive level, that consistency gap is meaningful.
Forza Horizon 5 and Hogwarts Legacy both run comfortably for both cards at maximum settings. Neither title exposes a meaningful weakness in either GPU at 1080p, and both push comfortably past 144fps average. Alan Wake 2, even without ray tracing, remains one of the most demanding rasterization workloads in gaming, and here the RTX 5070 holds a consistent lead of around eight to twelve percent across multiple test runs. Remedy’s engine clearly favors Nvidia’s architecture, though part of that advantage may be driver optimization rather than raw silicon performance.
Across the full suite, the RTX 5070 leads by roughly six to ten percent on average in rasterization-only workloads at 1080p. That is a real performance gap, but a smaller one than Nvidia’s pricing premium might suggest buyers are getting. The RX 9070 is not losing badly – it is losing by a margin that most players would genuinely not notice in everyday gaming sessions, and winning or drawing in specific workloads that happen to align with popular titles.
Pricing, Value, and the 1080p Context That Changes Everything
At MSRP, the RTX 5070 carries a roughly $100 to $150 premium over the RX 9070 depending on region and specific AIB model. For a six to ten percent average performance advantage at 1080p – a resolution where both cards are frequently CPU-bound and both deliver very high frame rates – that pricing structure is hard to defend on performance alone. Nvidia’s value proposition at this tier leans heavily on DLSS 4, Multi Frame Generation, and the broader ecosystem of Nvidia-exclusive features. If a buyer is committed to native rasterization and 1080p, the math increasingly favors AMD, particularly for players targeting 144Hz rather than 240Hz or above.
There is one factor that cuts the other way. AMD’s driver history is not spotless, and some players have experienced stability issues and inconsistent frame pacing with RDNA 4 cards in specific titles during the early months post-launch. Nvidia’s drivers tend to be more mature at launch, and for buyers who never want to troubleshoot a graphics issue, that reliability premium is real. Whether it is worth $100 to $150 is a personal calculation – but it is a calculation that belongs in this comparison, not buried in footnotes.
