AMD Radeon RX 7800 XT vs NVIDIA GeForce GTX 680 Comparison

FAQ

Q: How do the two cards compare in raw compute performance?

A: The AMD Radeon RX 7800 XT delivers 37.32 TFLOPS of FP32 compute, while the NVIDIA GeForce GTX 680 delivers 3.250 TFLOPS. The RX 7800 XT is over 11 times faster in theoretical floating-point throughput, a gap that reflects over a decade of architectural progress.

Q: What is the memory configuration difference?

A: The GTX 680 uses 2 GB of GDDR5 on a 256-bit bus with 192.3 GB/s of bandwidth. The RX 7800 XT uses 16 GB of GDDR6 on the same 256-bit bus but with 624.1 GB/s of bandwidth, more than three times the bandwidth of the older card.

Q: Which card has a higher average benchmark score in the database?

A: The GTX 680 has an average benchmark score of 14150 across its tested workloads, while the RX 7800 XT has an average of 11627. This is counterintuitive given the RX 7800 XT's newer architecture, but the GTX 680's scores come from a different, smaller benchmark set.

Q: How do the two cards fare in the direct head-to-head tests?

A: The RX 7800 XT wins both recorded head-to-head tests. In Geekbench OpenCL, it scores 18290 versus 16906 (a 7.6% margin). In Geekbench Vulkan, it scores 54228 versus 17646, a massive 67.5% advantage.

Q: What are the transistor density differences?

A: The GTX 680 packs 3,540 million transistors on a 294 mm² die, yielding 12.0M transistors per mm². The RX 7800 XT packs 28,100 million transistors on a 346 mm² die, yielding 81.2M per mm², a density nearly seven times higher due to the 5 nm process.

Q: What are the power requirements?

A: The GTX 680 has a 195 W TDP with a suggested 450 W power supply and dual 6-pin connectors. The RX 7800 XT has a 263 W TDP with a suggested 600 W power supply and dual 8-pin connectors.

Architecture Differences

The two cards represent a fundamental generational leap in GPU design. The GTX 680 is built on NVIDIA's Kepler architecture using TSMC's 28 nm process, while the RX 7800 XT uses AMD's RDNA 3.0 architecture on TSMC's 5 nm node. The process shrink alone enables dramatic changes: the RX 7800 XT houses 28,100 million transistors versus 3,540 million, and while its die is only slightly larger at 346 mm² versus 294 mm², the transistor density jumps from 12.0M to 81.2M per mm².

The execution units tell a similar story. The GTX 680 has 1,536 shading units, 128 texture mapping units, and 32 ROPs. The RX 7800 XT has 3,840 shading units, 240 TMUs, and 96 ROPs, roughly 2.5 times the shading units and three times the ROP count. The RX 7800 XT also adds 60 dedicated ray tracing cores, a feature entirely absent from the Kepler architecture, which relied on compute shaders for any ray effects.

Clock speeds have scaled significantly. The GTX 680 runs at a 1006 MHz base and 1058 MHz boost, while the RX 7800 XT runs at 1295 MHz base and 2430 MHz boost, with a 2124 MHz game clock. Memory technology also differs: GDDR5 at 6 Gbps effective on the older card versus GDDR6 at 19.5 Gbps effective on the newer one. Both use a 256-bit bus, but bandwidth rises from 192.3 GB/s to 624.1 GB/s.

API support diverges sharply. The GTX 680 supports DirectX 12 (11_0) and Vulkan 1.2.175, while the RX 7800 XT supports DirectX 12 Ultimate (12_2) and Vulkan 1.4. The newer card also supports FP16 compute at a 1:1 ratio with FP32, whereas the GTX 680 has no FP16 path listed. Display outputs differ as well: dual DVI, HDMI 1.4a, and DisplayPort 1.2 on the GTX 680 versus HDMI 2.1a and three DisplayPort 2.1 ports on the RX 7800 XT. The bus interface advances from PCIe 3.0 x16 to PCIe 4.0 x16.

Head-to-Head Benchmarks

The database records two direct comparisons between these cards, and the results are lopsided. In Geekbench OpenCL, the RX 7800 XT scores 18290 against the GTX 680's 16906. That is a 7.6% lead for AMD, a modest but consistent margin that reflects the newer card's higher shading unit count and memory bandwidth working together in compute workloads.

The Geekbench Vulkan result is far more decisive. The RX 7800 XT scores 54228, while the GTX 680 scores 17646. The delta is 67.5% in favor of the RX 7800 XT, meaning the newer card is roughly three times faster in this API. Vulkan's lower overhead and explicit control benefit the RDNA 3.0 architecture substantially, while Kepler's older driver model and limited feature set hold it back. This is the single largest performance gap recorded between the two.

Outside the head-to-head, the database's average scores paint a more nuanced picture. The GTX 680's average of 14150 places it at the 55th percentile among all GPUs, with nearest rivals including the NVIDIA Tesla K10 (14029, 0.9% slower) and the GeForce GTX 1070 Ti (14277, 0.9% faster). The RX 7800 XT's average of 11627 lands at the 51st percentile, with nearest rivals like the GeForce GTX 1660 (11680, 0.5% faster) and the Radeon RX 6500 XT (11842, 1.8% faster). The lower average for the RX 7800 XT is driven by its inclusion of PassMark tests across multiple DirectX versions, where it scores 121 in DirectX 10, 249 in DirectX 11, and 93 in DirectX 12, alongside a strong 24176 in PassMark G3D and 13473 in GPU compute. The GTX 680's benchmark set lacks these PassMark entries entirely, so the averages are not directly comparable.

The Verdict

The data supports a clear split. For modern gaming and general-purpose DirectX 12 and Vulkan workloads, the RX 7800 XT is the superior choice: it wins both head-to-head tests, offers 16 GB of memory versus 2 GB, and delivers 37.32 TFLOPS against 3.250 TFLOPS. The 67.5% Vulkan advantage is decisive for anyone running current titles or Vulkan-based applications.

For legacy compatibility or specific compute benchmarks where the GTX 680's simpler driver stack and lower overhead historically performed well, the older card still holds a place. Its 55th percentile ranking versus the RX 7800 XT's 51st percentile suggests that in the database's measured set, the GTX 680 punches above its age. However, that ranking derives from a smaller, less demanding benchmark suite. The RX 7800 XT's active production status and DirectX 12 Ultimate support make it the forward-looking pick, while the GTX 680 is end-of-life with no successor within its own generation.

The RX 7800 XT also matches the GTX 680's launch MSRP of 499 USD, though the two launched over a decade apart. The newer card offers 8 times the memory, nearly 7 times the transistor density, and over 11 times the FP32 throughput for the same nominal price point.

Specification Differences

| Specification | GTX 680 | RX 7800 XT |

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

| Process node | 28 nm | 5 nm |

| Transistors | 3,540 million | 28,100 million |

| Die size | 294 mm² | 346 mm² |

| Transistor density | 12.0M / mm² | 81.2M / mm² |

| Base clock | 1006 MHz | 1295 MHz |

| Boost clock | 1058 MHz | 2430 MHz |

| Game clock | N/A | 2124 MHz |

| Memory size | 2 GB | 16 GB |

| Memory type | GDDR5 | GDDR6 |

| Memory clock | 1502 MHz, 6 Gbps effective | 2438 MHz, 19.5 Gbps effective |

| Memory bandwidth | 192.3 GB/s | 624.1 GB/s |

| Shading units | 1536 | 3840 |

| TMUs | 128 | 240 |

| ROPs | 32 | 96 |

| Ray tracing cores | 0 | 60 |

| FP32 | 3.250 TFLOPS | 37.32 TFLOPS |

| FP16 | N/A | 37.32 TFLOPS (1:1) |

| Pixel rate | 33.86 GPixel/s | 233.3 GPixel/s |

| Texture rate | 135.4 GTexel/s | 583.2 GTexel/s |

| TDP | 195 W | 263 W |

| Power connectors | 2x 6-pin | 2x 8-pin |

| Suggested PSU | 450 W | 600 W |

| Bus interface | PCIe 3.0 x16 | PCIe 4.0 x16 |

| DirectX | 12 (11_0) | 12 Ultimate (12_2) |

| Vulkan | 1.2.175 | 1.4 |

| Display outputs | 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2 | 1x HDMI 2.1a, 3x DisplayPort 2.1 |

| Release date | 2012-03-21 | 2023-09-05 |

| Production status | End-of-life | Active |

Where Each One Wins

The RX 7800 XT wins in every modern compute and graphics category measured. Its Geekbench Vulkan score of 54228 versus 17646 makes it the clear choice for Vulkan-native games and applications, where the 67.5% delta translates to roughly three times the frame throughput. Its OpenCL advantage of 18290 versus 16906, though smaller at 7.6%, still favors AMD in general-purpose compute. The 624.1 GB/s of bandwidth versus 192.3 GB/s means texture-heavy workloads and high-resolution assets will see outsized gains. The 16 GB memory capacity allows for datasets and game textures that would exceed the GTX 680's 2 GB limit entirely.

The GTX 680 wins in the database's percentile ranking, sitting at the 55th percentile versus the RX 7800 XT's 51st. This is a statistical artifact of different benchmark suites, but it does indicate that in the specific tests applied to it, the Kepler card remains competitive against its contemporary peers. Its 195 W TDP also makes it a lower-power card, and its dual-slot width with 38 mm thickness versus 50 mm means it fits in tighter chassis. The GTX 680's dual DVI outputs are a legacy advantage for older monitors, while the RX 7800 XT's DisplayPort 2.1 and HDMI 2.1a support future displays.

For users with legacy DirectX 9 or 10 workloads, the RX 7800 XT's PassMark scores of 304 and 121 in those categories suggest it handles them adequately, though without a direct GTX 680 comparison, the older card's simpler architecture may have an edge in driver overhead. The RX 7800 XT's ray tracing cores make it the only option for hardware-accelerated ray tracing, a feature the GTX 680 cannot offer regardless of driver maturity. The verdict from the data is unambiguous: the RX 7800 XT is the modern performer, while the GTX 680 remains a historical data point with a surprisingly resilient benchmark profile.