AMD Radeon RX 6900 XT vs NVIDIA GeForce RTX 4080 SUPER Comparison

AMD
RADEON

AMD Radeon RX 6900 XT

CORE STATE Navi 21
VRAM 16 GB
CLOCK SPEED 2250 MHz
TDP 300 W
BUS WIDTH 256 bit
ARCHITECTURE RDNA 2.0
nm
PROCESS 7 nm
LAUNCH DATE 2020
VS
NVIDIA
GEFORCE

GeForce RTX 4080 SUPER

CORE STATE AD103
VRAM 16 GB
CLOCK SPEED 2550 MHz
TDP 320 W
BUS WIDTH 256 bit
ARCHITECTURE Ada Lovelace
nm
PROCESS 5 nm
LAUNCH DATE 2024

PERFORMANCE BENCHMARKS

3dmark_3dmark_steel_nomad_dx12
4,079
6,600
geekbench_metal
177,021
N/A
geekbench_opencl
187,673
219,065
geekbench_vulkan
148,525
260,075
passmark_directx_10
167
193
passmark_directx_11
279
301
passmark_directx_12
113
134
passmark_directx_9
268
381
passmark_g2d
1,055
1,270
passmark_g3d
26,732
34,245
passmark_gpu_compute
14,547
19,822

Analysis: AMD Radeon RX 6900 XT vs NVIDIA GeForce RTX 4080 SUPER

The benchmark data is unambiguous: the NVIDIA GeForce RTX 4080 SUPER defeats the AMD Radeon RX 6900 XT in every single head-to-head test recorded, with overall average scores of 53610 versus 53489, a delta of only 0.2%. While the aggregate margin is razor-thin, the individual benchmark deltas reveal a much more complex story—the RTX 4080 SUPER’s victories range from a modest 7.9% to a dominant 61.8%, indicating that its advantage is not uniform but heavily concentrated in specific workloads. Both cards share the same 88th percentile ranking among all GPUs, yet the pattern of wins suggests they are not interchangeable for every task.

Head-to-Head Benchmarks

The single largest margin of victory for the RTX 4080 SUPER comes in the 3DMark Steel Nomad DX12 test, where it scores 6600 against the RX 6900 XT’s 4079, a staggering 61.8% delta. This is the definitive statement of modern DirectX 12 rasterization performance—the NVIDIA card is in a different class entirely in this workload. The gap is so large that it cannot be explained by driver optimization alone; it points to a fundamental architectural superiority in how the AD103 chip handles the test’s geometry and shading demands.

Compute performance shows a similar but less extreme pattern. In Geekbench OpenCL, the RTX 4080 SUPER posts 246994 versus 167057, a 47.9% advantage, while in Geekbench Vulkan it leads 226163 to 166636, a 35.7% delta. These are not marginal improvements; they represent nearly half again as much compute throughput in synthetic API tests. The PassMark GPU Compute test reinforces this, with the NVIDIA card scoring 19822 against 14547, a 36.3% win. The data consistently shows the RTX 4080 SUPER executing parallel workloads with far greater efficiency.

Legacy DirectX performance tells a mixed story. In PassMark DirectX 9, the RTX 4080 SUPER wins 381 to 268, a 42.2% delta, which is surprisingly large for an older API. However, the DirectX 11 test narrows dramatically: 301 versus 279, only a 7.9% difference. DirectX 12 sees the gap widen again to 18.6% (134 vs 113), while DirectX 10 sits at 15.6% (193 vs 167). The pattern suggests the RX 6900 XT is most competitive in older, less demanding API paths, but loses ground as the API becomes more modern or more compute-intensive.

The 2D and general 3D tests further confirm NVIDIA’s lead. PassMark G2D shows 1270 versus 1055, a 20.4% delta, while PassMark G3D records 34245 versus 26732, a 28.1% advantage. Across all ten head-to-head benchmarks, the RTX 4080 SUPER wins every single one, with an average delta that skews heavily toward the larger compute and DX12 results. The RX 6900 XT never once flips the result in its favor.

Where Each One Wins

Based strictly on the benchmark data, the RTX 4080 SUPER wins in every measurable category. Its largest advantages are in 3DMark Steel Nomad DX12 (61.8% ahead), Geekbench OpenCL (47.9%), and PassMark DirectX 9 (42.2%). These are the workloads where the NVIDIA card’s architecture is most decisively superior—modern DX12 game engines, general-purpose compute through OpenCL, and even legacy DirectX 9 titles. For users targeting current-generation AAA games or GPU-accelerated compute tasks, the RTX 4080 SUPER is the clear choice.

The RX 6900 XT, despite losing every test, shows its smallest deficits in PassMark DirectX 11 (7.9%) and PassMark DirectX 10 (15.6%). This indicates that in older DirectX 11 applications, the AMD card is nearly competitive, trailing by less than 8%. Users with a library of pre-2015 games or applications locked to DX11 may find the RX 6900 XT’s performance acceptable, as the gap there is far smaller than in any other test. However, this is a narrow niche—it does not win any benchmark, it merely loses by a smaller margin.

The Geekbench Vulkan result (35.7% delta) shows that even in a cross-platform, low-level API, the RTX 4080 SUPER maintains a solid lead. The PassMark G3D result (28.1% delta) confirms that general 3D rendering favors NVIDIA. In no test does the RX 6900 XT outperform, meaning there is no workload category where the data supports choosing the AMD card on raw performance grounds.

Architecture Differences

The two GPUs are built on fundamentally different architectures and manufacturing processes. The RTX 4080 SUPER uses the AD103 chip on NVIDIA’s Ada Lovelace architecture, fabricated on a 5 nm process at TSMC with 45,900 million transistors in a 379 mm² die, yielding a transistor density of 121.1M per mm². The RX 6900 XT uses the Navi 21 chip on AMD’s RDNA 2.0 architecture, fabricated on a 7 nm process at TSMC with 26,800 million transistors in a much larger 520 mm² die, giving a density of only 51.5M per mm². The NVIDIA chip is smaller, denser, and packs 71% more transistors.

Memory configurations diverge sharply. Both cards have 16 GB, but the RTX 4080 SUPER uses GDDR6X at 23 Gbps effective, producing 736.3 GB/s of bandwidth over a 256-bit bus. The RX 6900 XT uses GDDR6 at 16 Gbps effective, yielding 512.0 GB/s over the same 256-bit bus. That is a 43.8% bandwidth advantage for NVIDIA, which directly explains its dominance in bandwidth-sensitive compute and DX12 tests.

Compute resources are radically different in count. The RTX 4080 SUPER has 10240 shading units, 320 TMUs, and 112 ROPs, plus 80 RT cores and 320 tensor cores. The RX 6900 XT has 5120 shading units, 320 TMUs, and 128 ROPs, plus 80 RT cores and no tensor cores. NVIDIA doubles the shading units and adds tensor cores, while AMD has more ROPs. This translates into FP32 performance of 52.22 TFLOPS for NVIDIA versus 23.04 TFLOPS for AMD—a 126.6% raw compute advantage. Clock speeds also differ: the RTX 4080 SUPER boosts to 2550 MHz, while the RX 6900 XT boosts to 2250 MHz.

The Verdict

The data is conclusive: the NVIDIA GeForce RTX 4080 SUPER is the superior GPU in every benchmark recorded, winning all ten head-to-head tests. Its overall average score of 53610 is 0.2% higher than the RX 6900 XT’s 53489, but the individual deltas show this is not a close contest in most workloads. The RTX 4080 SUPER is the only choice for users prioritizing modern DX12 gaming (61.8% ahead in 3DMark Steel Nomad), compute workloads (47.9% ahead in OpenCL), or even legacy DX9 titles (42.2% ahead). The tensor cores, doubled shading units, and significantly higher memory bandwidth make it categorically faster in every tested scenario.

The AMD Radeon RX 6900 XT, while losing every test, is not without merit. Its smallest deficit is 7.9% in DirectX 11, and it matches the RTX 4080 SUPER in 88th percentile ranking. For a user exclusively running DX11 applications, the RX 6900 XT offers near-parity performance. However, the data provides no scenario where the RX 6900 XT wins, so any claim of AMD superiority for a specific use case would be unsupported by these results. The RTX 4080 SUPER is the recommended pick for all workloads represented in the benchmark suite.

FAQ

Q: Which GPU has the higher average benchmark score?

A: The NVIDIA GeForce RTX 4080 SUPER has an average benchmark score of 53610, which is 0.2% higher than the AMD Radeon RX 6900 XT’s 53489.

Q: What is the largest performance gap between the two cards?

A: The largest gap is in 3DMark Steel Nomad DX12, where the RTX 4080 SUPER scores 6600 against the RX 6900 XT’s 4079, a 61.8% delta.

Q: In which test is the RX 6900 XT most competitive?

A: The RX 6900 XT is most competitive in PassMark DirectX 11, where it trails by only 7.9% (279 vs 301).

Q: How do the memory bandwidths compare?

A: The RTX 4080 SUPER has 736.3 GB/s of bandwidth with GDDR6X, while the RX 6900 XT has 512.0 GB/s with GDDR6, both on a 256-bit bus.

Q: Do both cards have the same number of ray tracing cores?

A: Yes, both the RTX 4080 SUPER and the RX 6900 XT have 80 RT cores, but the RTX 4080 SUPER also has 320 tensor cores while the RX 6900 XT has none.

Q: What is the FP32 compute performance difference?

A: The RTX 4080 SUPER delivers 52.22 TFLOPS, while the RX 6900 XT delivers 23.04 TFLOPS, a 126.6% advantage for NVIDIA.

Specification Differences

| Specification | NVIDIA GeForce RTX 4080 SUPER | AMD Radeon RX 6900 XT |

|---|---|---|

| Process Node | 5 nm | 7 nm |

| Transistors | 45,900 million | 26,800 million |

| Die Size | 379 mm² | 520 mm² |

| Transistor Density | 121.1M / mm² | 51.5M / mm² |

| Base Clock | 2295 MHz | 1825 MHz |

| Boost Clock | 2550 MHz | 2250 MHz |

| Game Clock | N/A | 2015 MHz |

| Memory Type | GDDR6X | GDDR6 |

| Memory Speed | 23 Gbps effective | 16 Gbps effective |

| Memory Bandwidth | 736.3 GB/s | 512.0 GB/s |

| Shading Units | 10240 | 5120 |

| ROPs | 112 | 128 |

| Tensor Cores | 320 | None |

| Pixel Rate | 285.6 GPixel/s | 288.0 GPixel/s |

| Texture Rate | 816.0 GTexel/s | 720.0 GTexel/s |

| FP32 Performance | 52.22 TFLOPS | 23.04 TFLOPS |

| FP16 Performance | 52.22 TFLOPS (1:1) | 46.08 TFLOPS (2:1) |

| TDP | 320 W | 300 W |

| Power Connectors | 1x 16-pin | 2x 8-pin |

| Display Outputs | 1x HDMI 2.1, 3x DisplayPort 1.4a | 1x HDMI 2.1, 2x DisplayPort 1.4a, 1x USB Type-C |

| Length | 310 mm (12.2 inches) | 267 mm (10.5 inches) |

| Release Date | 2024-01-07 | 2020-10-27 |

DETAILED SPECIFICATIONS

SPECIFICATION
RX 6900 XT
RTX 4080 SUPER
Core Specs
Shading Units
5,120
10,240 +100.0%
Shaders
5,120
10,240 +100.0%
TMUs
320
320 0.0%
ROPs
128
112 -12.5%
Compute Units
80
—
SM Count
—
80
Clocks
Base Clock
1825 MHz
2295 MHz
Boost Clock
2250 MHz
2550 MHz
Game Clock
2015 MHz
—
Memory Clock
2000 MHz 16 Gbps effective
1438 MHz 23 Gbps effective
Memory
Memory Size
16 GB
16 GB
VRAM (MB)
16,384
16,384 0.0%
Memory Type
GDDR6
GDDR6X
Memory Bus
256 bit
256 bit
Bandwidth
512.0 GB/s
736.3 GB/s
Cache
L1 Cache
128 KB per Array
128 KB (per SM)
L2 Cache
4 MB
64 MB
L3 Cache
128 MB
—
L0 Cache
32 KB per WGP
—
Performance
Pixel Rate
288.0 GPixel/s
285.6 GPixel/s
Texture Rate
720.0 GTexel/s
816.0 GTexel/s
FP32 (TFLOPS)
23.04 TFLOPS
52.22 TFLOPS
FP64 (TFLOPS)
1,440.0 GFLOPS (1:16)
816.0 GFLOPS (1:64)
FP16 (TFLOPS)
46.08 TFLOPS (2:1)
52.22 TFLOPS (1:1)
AI/RT
RT Cores
80
80 0.0%
Tensor Cores
—
320
Power
TDP
300 W
320 W
TDP (W)
300
320 +6.7%
Suggested PSU
700 W
700 W
Power Connectors
2x 8-pin
1x 16-pin
Architecture
Architecture
RDNA 2.0
Ada Lovelace
GPU Name
Navi 21
AD103
Generation
Navi II (RX 6000)
GeForce 40
Process Size
7 nm
5 nm
Transistors
26,800 million
45,900 million
Die Size
520 mm²
379 mm²
Foundry
TSMC
TSMC
Density
51.5M / mm²
121.1M / mm²
API Support
DirectX
12 Ultimate (12_2)
12 Ultimate (12_2)
OpenGL
4.6
4.6
Vulkan
1.4
1.4
OpenCL
2.1
3.0
CUDA
—
8.9
Shader Model
6.8
6.9
Physical
Slot Width
Triple-slot
Triple-slot
Length
267 mm 10.5 inches
310 mm 12.2 inches
Height
120 mm 4.7 inches
140 mm 5.5 inches
Outputs
1x HDMI 2.12x DisplayPort 1.4a1x USB Type-C
1x HDMI 2.13x DisplayPort 1.4a
Bus Interface
PCIe 4.0 x16
PCIe 4.0 x16
Other
Launch Price
999 USD
999 USD
Production
End-of-life
End-of-life
Predecessor
Navi
GeForce 30
Successor
Navi III
GeForce 50
View Radeon RX 6900 XT Details View GeForce RTX 4080 SUPER Details