Two Cards, One Target, Very Different Strategies
The mid-range GPU market at 1080p has always been where most PC gaming actually happens, and right now two cards are fighting hard for that space: Intel’s Arc B770 and Nvidia’s RTX 5060 Ti. One is Intel’s most aggressive push yet into discrete graphics, priced to disrupt. The other is Nvidia leaning on its established architecture advantage and ecosystem lock-in through DLSS. Neither card is cheap by historical mid-range standards, but both are being positioned as the sensible buy for 1080p gaming in 2025.
Rasterization performance – traditional rendering without ray tracing or upscaling tricks – is the fairest test of raw GPU muscle. It strips away the software advantages that Nvidia routinely uses to pad benchmark numbers and forces both cards to compete on shader throughput, memory bandwidth, and driver efficiency. What the benchmarks show is a more complicated picture than either company’s marketing would suggest.
Specifications and Pricing
The Arc B770 is built on Intel’s Xe2 architecture, the same foundation that made the B580 a genuine surprise when it launched. It carries 20 Xe2 cores, 20GB of GDDR6 memory on a 160-bit bus, and a 200W TDP. Intel priced it at $349 at launch, which was a deliberate shot at Nvidia’s mid-range lineup. The memory capacity alone is notable – 20GB at this price point is something neither Nvidia nor AMD offers anywhere close to this tier.
The RTX 5060 Ti uses Nvidia’s Blackwell architecture and comes in two configurations: 8GB and 16GB GDDR7 variants, priced at $379 and $429 respectively. The 16GB version is the fair comparison point against the B770. Blackwell brings higher shader efficiency per watt compared to Ada Lovelace, and Nvidia’s driver stack is the most mature in the industry. On paper, the clock speeds and compute units favor Nvidia, but raw specs have a habit of not telling the full story with Intel’s XeSS and shader performance improvements in Xe2.
1080p Rasterization Benchmarks: Game by Game
In Cyberpunk 2077 at 1080p Ultra settings with rasterization only, the RTX 5060 Ti 16GB pulls ahead by roughly 12 to 15 percent in average frame rates. CD Projekt Red’s engine has historically favored Nvidia hardware due to deep optimization work done alongside Nvidia’s developer relations team, so this gap is not entirely a pure architecture win. The B770 still delivers well above 60fps averages, landing in playable, comfortable territory, but the 5060 Ti leads clearly here.
Alan Wake 2 tells a similar story in rasterization-only mode – Nvidia leads by around 10 percent at 1080p Ultra. Remedy’s engine is another case where Nvidia’s toolchain partnerships create a built-in edge. The B770 holds its own, but Intel’s driver maturity in titles that push complex geometry and large scene rendering still trails Nvidia’s years of accumulated optimization.
Where Intel closes the gap is in titles that are less Nvidia-optimized. In Baldur’s Gate 3, Elden Ring, and Spider-Man: Miles Morales at 1080p, the B770 trades blows almost evenly with the RTX 5060 Ti 8GB variant and comes within 5 to 7 percent of the 16GB model. The memory advantage on the B770 shows up most in open-world titles with dense asset streaming, where the 8GB RTX 5060 Ti can occasionally stutter under heavy load. Intel’s 20GB buffer makes those scenarios a non-issue.
The one consistent area where the B770 genuinely surprises is compute-heavy workloads embedded in rasterization pipelines. Titles using extensive physics simulation or complex particle systems run closer to parity than shader-heavy benchmarks suggest. Intel’s Xe2 architecture has strong compute throughput, and that shows up in specific workload profiles even when raw frame rates favor Nvidia.
Driver Stability and Real-World Experience
Benchmark numbers capture one dimension of GPU performance; driver stability captures another. Intel’s Arc drivers have improved dramatically since the disastrous early B580 launch period, but edge cases still surface more often than they do with Nvidia’s GeForce drivers. Some users report occasional stuttering in DirectX 11 titles, and support for older game libraries remains inconsistent. For someone building a system primarily around recent releases, this matters less. For someone with a library spanning a decade, it matters quite a bit.
Nvidia’s driver ecosystem is simply more mature, and the RTX 5060 Ti benefits from that. Frame pacing is tighter, compatibility is broader, and Nvidia’s overlay tools, Reflex, and broadcast features add real utility beyond raw frame rates. The B770 offers XeSS, which has improved to genuine quality at higher presets, but Nvidia’s feature breadth remains wider.
Value Calculation at 1080p
The RTX 5060 Ti 8GB at $379 is a bad buy specifically because of the memory ceiling. At 1080p today it is mostly fine, but texture budgets in modern games are climbing and 8GB will become a real constraint faster than the card’s performance ceiling suggests. The 16GB model at $429 is the correct comparison point, and at that price the B770 at $349 makes a strong argument purely on value per frame.
If average frame rates are the only metric, Nvidia wins most benchmarks. But the B770’s $80 price advantage over the 16GB 5060 Ti and its 20GB memory buffer make it a more future-conscious buy for pure 1080p rasterization gaming. The performance gap in Nvidia-optimized titles is real but rarely dramatic enough to ruin gameplay. In titles without deep Nvidia partnerships, the gap shrinks to near noise.
The uncomfortable truth for Intel is that even at a $80 discount, the B770 needs to be meaningfully faster – not just competitive – to shift buying habits away from Nvidia’s ecosystem. Gamers who already own RTX cards, who use DLSS regularly, or who value the full Nvidia software suite have little reason to switch. The B770 makes its best case to first-time builders, budget-conscious upgraders, and anyone who specifically needs that 20GB buffer. For the 1080p gamer who just wants the most frames per dollar with rasterization as the primary workload, the B770 is harder to dismiss than Nvidia would prefer.
