AMD Radeon RX 6800 vs NVIDIA GeForce RTX 3090 Comparison

FAQ

Q: Which GPU has the higher average benchmark score?

A: The AMD Radeon RX 6800 holds a higher average benchmark score of 30,095, compared to the NVIDIA GeForce RTX 3090’s 27,565. This places the RX 6800 at the 75th percentile of all GPUs, while the RTX 3090 sits at the 73rd percentile.

Q: How do the two cards compare in the 3DMark Steel Nomad DX12 test?

A: The NVIDIA GeForce RTX 3090 wins decisively, scoring 5,118 against the RX 6800’s 3,188. That is a 37.7% lead for the RTX 3090, making it the clear performance king in this modern DX12 workload.

Q: Where does the AMD RX 6800 outperform the RTX 3090?

A: The RX 6800 wins the Geekbench Vulkan test with a score of 115,107, which is 113.4% higher than the RTX 3090’s 53,927. This is the only head-to-head benchmark where AMD takes the win.

Q: What are the closest rivals to each card according to the data?

A: The RX 6800’s nearest rival is the NVIDIA GeForce RTX 3070 Ti, with a 0.5% score difference. The RTX 3090’s nearest rival is the NVIDIA GeForce RTX 4070 Mobile, also with a 0.5% delta, though its average score is slightly lower.

Q: What is the memory configuration difference between the two?

A: The RTX 3090 comes with 24 GB of GDDR6X memory on a 384-bit bus, delivering 936.2 GB/s of bandwidth. The RX 6800 has 16 GB of GDDR6 on a 256-bit bus, providing 512.0 GB/s of bandwidth.

Q: Which card has a higher transistor count and die size?

A: The RTX 3090 uses 28,300 million transistors on a 628 mm² die, while the RX 6800 uses 26,800 million transistors on a 520 mm² die. Despite the larger die, the RTX 3090 has a lower transistor density of 45.1M / mm² compared to the RX 6800’s 51.5M / mm².

Architecture Differences

The two GPUs represent fundamentally different architectural philosophies from their respective manufacturers. AMD’s Radeon RX 6800 is built on the RDNA 2.0 architecture, fabricated on a 7 nm process at TSMC. NVIDIA’s GeForce RTX 3090 uses the Ampere architecture, manufactured on an 8 nm process at Samsung. The process difference is notable: AMD’s smaller node allows for a higher transistor density of 51.5M / mm², whereas NVIDIA’s larger die has a density of 45.1M / mm².

The compute resources diverge sharply. The RX 6800 packs 3,840 shading units, 240 TMUs, and 96 ROPs, along with 60 dedicated ray tracing cores. The RTX 3090 is far larger in this regard, featuring 10,496 shading units, 328 TMUs, and 112 ROPs, plus 82 RT cores and 328 tensor cores. The RTX 3090’s tensor cores are a unique feature for AI acceleration, which the RX 6800 lacks entirely.

Memory architecture also separates the two. The RX 6800 uses 16 GB of GDDR6 on a 256-bit bus, while the RTX 3090 uses 24 GB of GDDR6X on a 384-bit bus. This gives the RTX 3090 a bandwidth advantage of 936.2 GB/s versus 512.0 GB/s. Clock behavior differs as well: the RX 6800 has a higher base clock of 1700 MHz and boost clock of 2105 MHz, while the RTX 3090 runs lower at 1395 MHz base and 1695 MHz boost. The RX 6800 also has a game clock of 1815 MHz, a figure the RTX 3090 does not specify.

Power and physical design reflect the performance disparity. The RX 6800 is a dual-slot card with a 250 W TDP, requiring a 600 W PSU and two 8-pin connectors. The RTX 3090 is a triple-slot behemoth with a 350 W TDP, a suggested 750 W PSU, and a single 12-pin connector. The physical dimensions differ accordingly: the RX 6800 measures 267 mm in length, while the RTX 3090 stretches to 336 mm.

Both cards support PCIe 4.0 x16 and share the same API feature set, including DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. Display outputs vary slightly: the RX 6800 offers 1x HDMI 2.1, 2x DisplayPort 1.4a, and 1x USB Type-C, while the RTX 3090 provides 1x HDMI 2.1 and 3x DisplayPort 1.4a.

Head-to-Head Benchmarks

The head-to-head data across ten benchmarks is lopsided: the RTX 3090 wins nine of ten tests, with the RX 6800 taking a single victory. The largest margin of victory belongs to NVIDIA in the Geekbench OpenCL test, where the RTX 3090 scores 172,758 against the RX 6800’s 24,508 — a staggering 85.8% lead. This indicates a massive advantage in general-purpose compute workloads that favor NVIDIA’s architecture.

In the 3DMark Steel Nomad DX12 test, the RTX 3090 scores 5,118 versus 3,188 for the RX 6800, a 37.7% difference. This is the most modern gaming-oriented test in the suite, and the RTX 3090 dominates. The Passmark G3D test shows a similar trend, with the RTX 3090 scoring 26,645 against 22,067, a 17.2% lead. In GPU compute via Passmark, the RTX 3090 wins again with 15,356 versus 10,864, a 29.3% margin.

The RTX 3090’s wins continue across legacy DirectX tests. In Passmark DirectX 10, it scores 182 versus 128, a 29.7% lead. DirectX 11 shows a narrower margin: 220 versus 214, just a 2.7% difference. DirectX 12 gives the RTX 3090 a 110 to 89 win, a 19.1% gap. DirectX 9 is close, with the RTX 3090 at 268 versus 257, a 4.1% edge. Even in the 2D test, Passmark G2D, the RTX 3090 leads 1,063 to 990, a 6.9% difference.

The sole AMD victory is in Geekbench Vulkan, where the RX 6800 scores 115,107 against the RTX 3090’s 53,927. That is a 113.4% lead — more than doubling NVIDIA’s score. This suggests that in Vulkan-based workloads, the RX 6800’s architecture is exceptionally efficient, even against a far more powerful GPU on paper. The delta is so large that it skews the overall average: despite losing nine tests, the RX 6800 still has a higher average benchmark score of 30,095 versus 27,565.

It is worth noting the RTX 3090’s nearest rivals include the RTX 4070 Mobile and RX 6700 XT, both around 0.5% away in score. The RX 6800’s nearest rivals are the RTX 3070 Ti and RTX 5070 Mobile, also within 0.6%. This places the two cards in different competitive brackets despite the RX 6800’s higher average.

The Verdict

The data paints a clear picture for different use cases. If the priority is raw compute performance and modern DX12 gaming, the NVIDIA GeForce RTX 3090 is the superior choice. It wins the 3DMark Steel Nomad test by 37.7%, the Passmark G3D test by 17.2%, and the Passmark GPU compute test by 29.3%. Its 35.58 TFLOPS of FP32 performance, more than double the RX 6800’s 16.17 TFLOPS, is reflected in these benchmark margins. For creators or users running OpenCL-heavy applications, the RTX 3090’s 85.8% lead in Geekbench OpenCL makes it the only rational pick.

However, the RX 6800 is not without its strengths. Its 113.4% victory in Geekbench Vulkan is a standout result, indicating that Vulkan-based games or applications could see exceptional performance on AMD’s architecture. The RX 6800 also has a higher average benchmark score overall, 30,095 versus 27,565, and a better percentile ranking at 75 versus 73. This suggests that across a broad mix of workloads, the RX 6800 is more consistent, even if it loses in raw peak performance.

From a physical standpoint, the RX 6800 is the more practical card. It is a dual-slot design at 267 mm length, with a 250 W TDP and a 600 W PSU requirement. The RTX 3090 is a triple-slot card at 336 mm, drawing 350 W and needing a 750 W PSU. For smaller cases or lower-power builds, the RX 6800 is far easier to accommodate. The launch MSRP also reflects the tier gap: the RX 6800 was 579 USD, while the RTX 3090 launched at 1,499 USD.

In terms of memory, the RTX 3090’s 24 GB of GDDR6X with 936.2 GB/s bandwidth is objectively superior for high-resolution textures and large datasets. The RX 6800’s 16 GB of GDDR6 at 512.0 GB/s is adequate but half the bandwidth. For users who need maximum memory capacity, the RTX 3090 is the clear winner.

The verdict depends on workload. The RTX 3090 is the performance flagship, winning nine of ten benchmarks and offering 82 RT cores and 328 tensor cores for future-proofing in ray tracing and AI tasks. The RX 6800 is the efficiency pick with a higher average score, a Vulkan advantage, and a much smaller physical footprint. Neither card is universally better; the data supports the RTX 3090 for compute and DX12 gaming, and the RX 6800 for Vulkan-heavy environments and users prioritizing a smaller, lower-power build.

Specification Differences

The specifications that differ between the two cards are extensive. The process node is 7 nm for the RX 6800 versus 8 nm for the RTX 3090, with different foundries (TSMC versus Samsung). Transistor counts are 26,800 million for AMD and 28,300 million for NVIDIA, with die sizes of 520 mm² and 628 mm², respectively. Transistor density is 51.5M / mm² versus 45.1M / mm².

Clock speeds differ in base and boost: the RX 6800 runs at 1700 MHz base and 2105 MHz boost, while the RTX 3090 runs at 1395 MHz base and 1695 MHz boost. The RX 6800 has a game clock of 1815 MHz, which the RTX 3090 does not list. Memory clocks are 2000 MHz (16 Gbps effective) for AMD and 1219 MHz (19.5 Gbps effective) for NVIDIA.

Memory specifications diverge completely: 16 GB GDDR6 versus 24 GB GDDR6X, 256-bit versus 384-bit bus, and 512.0 GB/s versus 936.2 GB/s bandwidth. Shading units are 3,840 versus 10,496; TMUs are 240 versus 328; ROPs are 96 versus 112. Ray tracing cores are 60 versus 82, and the RTX 3090 adds 328 tensor cores while the RX 6800 has none.

Pixel rate is 202.1 GPixel/s for AMD versus 189.8 GPixel/s for NVIDIA, a rare AMD win. Texture rate is 505.2 GTexel/s versus 556.0 GTexel/s. FP32 performance is 16.17 TFLOPS versus 35.58 TFLOPS, and FP16 is 32.33 TFLOPS (2:1) versus 35.58 TFLOPS (1:1). TDP is 250 W versus 350 W, with slot widths of dual-slot versus triple-slot. Power connectors are 2x 8-pin versus 1x 12-pin, and suggested PSU is 600 W versus 750 W.

Dimensions differ: length 267 mm versus 336 mm, height 120 mm versus 140 mm, and width 40 mm versus 61 mm. Display outputs are 1x HDMI 2.1, 2x DisplayPort 1.4a, 1x USB Type-C for AMD, versus 1x HDMI 2.1 and 3x DisplayPort 1.4a for NVIDIA. Release dates are 2020-10-27 for AMD and 2020-08-31 for NVIDIA. The launch MSRP is 579 USD versus 1,499 USD.