AMD Radeon RX 6900 XT vs NVIDIA GeForce RTX 5070 Ti Comparison

The AMD Radeon RX 6900 XT and NVIDIA GeForce RTX 5070 Ti are separated by a full generation of GPU architecture, yet they land in the same performance tier. The benchmark data shows the RTX 5070 Ti winning all ten head-to-head comparisons, with an average benchmark score of 49957 against the RX 6900 XT's 50951. That average is deceptive, however, because the RTX 5070 Ti's newer architecture dominates in modern workloads while the RX 6900 XT's higher average is buoyed by legacy tests. The RTX 5070 Ti is the clear choice for current-generation gaming and compute; the RX 6900 XT remains relevant only for specific legacy scenarios.

The Verdict

The RTX 5070 Ti is the superior card for nearly every buyer, based on the data. It wins 10 out of 10 head-to-head benchmark comparisons, with decisive margins in the most demanding modern tests. In 3DMark Steel Nomad DX12, the RTX 5070 Ti scores 6604 against the RX 6900 XT's 4079, a 38.2% advantage. That single result encapsulates the generational gap: NVIDIA's Blackwell architecture extracts far more performance from its 300 W power envelope than AMD's RDNA 2.0 does from the same 300 W.

The RX 6900 XT is only defensible for users running legacy DirectX 9 or DirectX 10 applications where the gap narrows, or for those who specifically need its USB Type-C display output. The data shows the RX 6900 XT trailing by just 7% in Passmark DirectX 11 and 13% in Passmark DirectX 10, making those older workloads its least-bad scenarios. But even there, the RTX 5070 Ti wins. The verdict is unambiguous: the RTX 5070 Ti is the pick for anyone building a new system or upgrading from a GPU older than the RX 6000 series.

Where Each One Wins

The RTX 5070 Ti wins every category, but the margin tells the story of where each card excels. The largest gap is in 3DMark Steel Nomad DX12, where the 38.2% delta reflects the RTX 5070 Ti's superior ray tracing and DX12 Ultimate performance. The second-largest gap is in Geekbench Vulkan, where the RTX 5070 Ti scores 225122 against 148525, a 34% lead. That Vulkan result indicates the RTX 5070 Ti is dramatically stronger in modern cross-platform game engines and compute APIs.

The RX 6900 XT's closest result is Passmark DirectX 11, where it trails by only 7% (279 vs 300). This suggests the older card's architecture remains competitive in legacy DirectX 11 titles, likely due to its higher pixel rate of 288.0 GPixel/s versus the RTX 5070 Ti's 235.4 GPixel/s. The RX 6900 XT also shows relative strength in Passmark DirectX 12, trailing by 11% (113 vs 127), which is closer than its 18.9% deficit in Passmark G3D (26732 vs 32974). The RTX 5070 Ti dominates compute workloads, leading by 28% in Passmark GPU Compute (20203 vs 14547) and 11.6% in Geekbench OpenCL (212363 vs 187673).

Architecture Differences

The two cards represent fundamentally different design philosophies separated by five years of process technology. The RX 6900 XT uses AMD's Navi 21 chip on TSMC's 7 nm process, packing 26,800 million transistors into a 520 mm² die. The RTX 5070 Ti uses NVIDIA's GB203 chip on TSMC's 5 nm process, fitting 45,600 million transistors into a much smaller 378 mm² die. That process advantage yields a transistor density of 120.6M per mm² for the RTX 5070 Ti versus 51.5M per mm² for the RX 6900 XT.

The RTX 5070 Ti's Blackwell 2.0 architecture brings 8960 shading units, 280 tensor cores, and 70 RT cores, while the RX 6900 XT's RDNA 2.0 has 5120 shading units, no tensor cores, and 80 RT cores. The RTX 5070 Ti compensates for fewer RT cores with newer ray tracing hardware, as evidenced by its 3DMark Steel Nomad dominance. Memory technology also diverges: the RTX 5070 Ti uses GDDR7 at 896.0 GB/s bandwidth, while the RX 6900 XT uses GDDR6 at 512.0 GB/s. Both have 16 GB on a 256-bit bus, but the RTX 5070 Ti's memory runs at 1750 MHz (28 Gbps effective) versus 2000 MHz (16 Gbps effective) for the RX 6900 XT.

The FP32 compute figures reveal the scale of the generational leap: the RTX 5070 Ti delivers 43.94 TFLOPS against the RX 6900 XT's 23.04 TFLOPS. The RTX 5070 Ti also runs at higher clocks (2295 MHz base, 2452 MHz boost) versus the RX 6900 XT's 1825 MHz base and 2250 MHz boost. The RX 6900 XT does maintain advantages in texture rate (720.0 GTexel/s vs 686.6 GTexel/s) and pixel rate, but those don't translate into benchmark wins.

FAQ

Q: Which card has better raw compute performance?

A: The RTX 5070 Ti delivers 43.94 TFLOPS FP32 versus the RX 6900 XT's 23.04 TFLOPS, and it leads by 28% in Passmark GPU Compute (20203 vs 14547) and 11.6% in Geekbench OpenCL (212363 vs 187673).

Q: Does the RX 6900 XT win any benchmark?

A: No. The head-to-head data shows the RTX 5070 Ti winning all ten benchmarks, with the RX 6900 XT's closest result being a 7% deficit in Passmark DirectX 11 (279 vs 300).

Q: How do their memory subsystems compare?

A: Both cards have 16 GB on a 256-bit bus, but the RTX 5070 Ti uses GDDR7 with 896.0 GB/s bandwidth, while the RX 6900 XT uses GDDR6 with 512.0 GB/s bandwidth.

Q: Which card is better for ray tracing?

A: The RTX 5070 Ti is significantly better, as shown by its 38.2% lead in 3DMark Steel Nomad DX12 (6604 vs 4079), which is a DirectX 12 Ultimate ray tracing workload.

Q: Are they in the same performance percentile?

A: Yes, both cards sit at the 86th percentile among all GPUs, but the RTX 5070 Ti has a higher average benchmark score of 49957 compared to the RX 6900 XT's 50951, with the RX 6900 XT's higher average driven by legacy test results.

Q: What are their physical dimensions?

A: The RX 6900 XT is 267 mm long, 120 mm tall, and 50 mm wide (triple-slot), while the RTX 5070 Ti is 304 mm long, 137 mm tall, and 48 mm wide (dual-slot).

Head-to-Head Benchmarks

The 3DMark Steel Nomad DX12 result is the defining comparison. The RTX 5070 Ti scores 6604 versus the RX 6900 XT's 4079, a 38.2% lead. This is the largest margin in the entire benchmark suite and reflects the RTX 5070 Ti's superior DX12 Ultimate feature set and ray tracing performance. For users running modern AAA titles with ray tracing enabled, this is the most relevant metric, and the data shows a generational chasm.

Geekbench Vulkan shows a similar story: the RTX 5070 Ti scores 225122 against 148525, a 34% lead. Vulkan is the API of choice for many cross-platform engines and DXVK-based translations, making this result critical for Linux gamers and users of emulation. The RTX 5070 Ti's 8960 shading units and Blackwell architecture clearly execute Vulkan compute shaders far more efficiently than the RX 6900 XT's 5120 shading units.

Compute workloads favor the RTX 5070 Ti by 28% in Passmark GPU Compute (20203 vs 14547). This aligns with the FP32 throughput difference: 43.94 TFLOPS versus 23.04 TFLOPS. The RTX 5070 Ti's 280 tensor cores also enable AI-accelerated workloads that the RX 6900 XT cannot perform at all, though no tensor-specific benchmark is included in this data.

The legacy DirectX tests show the RX 6900 XT at its most competitive. In Passmark DirectX 11, the margin shrinks to 7% (300 vs 279). DirectX 9 sees a 23.6% delta (351 vs 268), and DirectX 10 a 13% delta (192 vs 167). These older APIs rely more on rasterization throughput, where the RX 6900 XT's 128 ROPs and 288.0 GPixel/s pixel rate provide a counterweight to the RTX 5070 Ti's architectural advantages.

Passmark G3D shows an 18.9% lead for the RTX 5070 Ti (32974 vs 26732), while Passmark G2D shows a 20.8% lead (1332 vs 1055). The RTX 5070 Ti also wins Geekbench OpenCL by 11.6% (212363 vs 187673) and Passmark DirectX 12 by 11% (127 vs 113). Across all ten benchmarks, the RTX 5070 Ti's average margin is approximately 20%, with its smallest win in DirectX 11 and its largest in Steel Nomad.

Specification Differences

The two cards differ across nearly every specification field. The RTX 5070 Ti uses a 5 nm process with 45,600 million transistors on a 378 mm² die, while the RX 6900 XT uses 7 nm with 26,800 million transistors on a 520 mm² die. Clock speeds favor the RTX 5070 Ti: 2295 MHz base and 2452 MHz boost versus 1825 MHz base and 2250 MHz boost. The RX 6900 XT has a game clock of 2015 MHz, which the RTX 5070 Ti lacks.

Memory configuration diverges in type and bandwidth: GDDR7 at 896.0 GB/s for the RTX 5070 Ti versus GDDR6 at 512.0 GB/s for the RX 6900 XT, both at 16 GB and 256-bit. The RTX 5070 Ti has 8960 shading units, 280 TMUs, 96 ROPs, 70 RT cores, and 280 tensor cores. The RX 6900 XT has 5120 shading units, 320 TMUs, 128 ROPs, 80 RT cores, and no tensor cores. Pixel rate favors the RX 6900 XT at 288.0 GPixel/s versus 235.4 GPixel/s, while texture rate is nearly even at 720.0 GTexel/s versus 686.6 GTexel/s. FP32 is 43.94 TFLOPS for the RTX 5070 Ti versus 23.04 TFLOPS, and FP16 is 43.94 TFLOPS (1:1) versus 46.08 TFLOPS (2:1).

The RTX 5070 Ti uses PCIe 5.0 x16, while the RX 6900 XT uses PCIe 4.0 x16. Display outputs differ: the RTX 5070 Ti has 1x HDMI 2.1b and 3x DisplayPort 2.1b, while the RX 6900 XT has 1x HDMI 2.1, 2x DisplayPort 1.4a, and 1x USB Type-C. Physical dimensions favor the RX 6900 XT in length (267 mm vs 304 mm) but the RTX 5070 Ti in slot width (dual-slot vs triple-slot). Power connectors are 1x 16-pin for the RTX 5070 Ti versus 2x 8-pin for the RX 6900 XT, though both share a 300 W TDP and 700 W suggested PSU. The RX 6900 XT launched at 999 USD and is end-of-life, while the RTX 5070 Ti launched at 749 USD and remains active.