AMD Radeon RX 7900 XT vs NVIDIA GeForce GTX 680 Comparison

The GeForce GTX 680 and the Radeon RX 7900 XT are separated by a decade of GPU architecture, and the benchmark data reflects a fundamental split: the older NVIDIA card wins the Geekbench OpenCL compute test, while the AMD card dominates the Geekbench Vulkan test by a massive margin. The GTX 680 delivers a 32.5% higher OpenCL score (16906 vs 12756), but the RX 7900 XT crushes it in Vulkan with a 71068 score versus 17646, a 75.2% difference. Both cards hold the same 55th percentile ranking among all GPUs, yet their average benchmark scores are close—14150 for the GTX 680 and 13745 for the RX 7900 XT—which places them within 1% of each other on average, despite the stark differences in their underlying specifications.

The Verdict

The data points to two distinct buyers. The GTX 680 is the choice for anyone prioritizing OpenCL compute workloads, as it outperforms the RX 7900 XT by 32.5% in that specific test. Its 14150 average benchmark score also edges out the RX 7900 XT's 13745, and it sits 0.9% above the NVIDIA Tesla K10 while being only 0.9% below the GTX 1070 Ti in the nearest rival rankings. The RX 7900 XT is the definitive winner for Vulkan-based gaming and modern API workloads, delivering a 71068 Vulkan score that is over four times higher than the GTX 680's 17646. The RX 7900 XT also shows its strength in DirectX 12 with a 5623 score in the 3DMark Steel Nomad test, a benchmark the GTX 680 does not even have a recorded result for.

For a user with a modern workload that leverages DirectX 12 Ultimate or Vulkan, the RX 7900 XT is the only rational choice from the data. The GTX 680's lead in OpenCL is notable, but that API is less common in contemporary gaming and more relevant to specific compute tasks. The RX 7900 XT's 55th percentile ranking, despite its massive Vulkan advantage, is pulled down by its lower OpenCL score and a 1.5% deficit to the RX 9070 XT, while the GTX 680's same percentile is supported by consistent performance across its available tests. The verdict is clear: the GTX 680 wins a narrow compute niche, but the RX 7900 XT wins the overall modern performance picture by a landslide in the API that matters most for current software.

Where Each One Wins

The GTX 680 wins decisively in OpenCL compute. Its 16906 score against the RX 7900 XT's 12756 represents a 32.5% lead, making it the superior option for any application that relies on this cross-platform API for general-purpose GPU computing. This is the only head-to-head benchmark where the GTX 680 comes out ahead, and it is a significant one. The card's 1536 shading units and 128 texture mapping units, while far fewer than the AMD card's 5376 and 336 respectively, are evidently organized in a way that excels at this specific workload pattern.

The RX 7900 XT wins everywhere else. Its Vulkan score of 71068 is not just a win—it is a rout, outperforming the GTX 680 by 75.2%. This indicates a fundamental advantage in modern low-level API performance, which is critical for current game engines and VR applications. Beyond Vulkan, the RX 7900 XT has a 5623 score in the 3DMark Steel Nomad DX12 test, which is a modern benchmark that the GTX 680 cannot even run, as evidenced by its absence from the data. The RX 7900 XT also has a comprehensive set of Passmark scores (158 in DirectX 10, 340 in DirectX 11, 110 in DirectX 12, 322 in DirectX 9, 1232 in G2D, 29009 in G3D, and 16828 in GPU compute), all of which are absent for the GTX 680, showing it is a more complete and modern performer across the board.

Architecture Differences

The architectural gap between these two cards is enormous. The GTX 680 uses the GK104 chip on the Kepler architecture, built on a 28 nm process at TSMC with 3,540 million transistors on a 294 mm² die, giving a transistor density of 12.0 million per mm². The RX 7900 XT uses the Navi 31 chip on the RDNA 3.0 architecture, also fabricated by TSMC but on a much more advanced 5 nm process, with 57,700 million transistors packed into a 529 mm² die, achieving a density of 109.1 million per mm²—a 9.1x increase in density.

The memory subsystems are equally divergent. The GTX 680 has 2 GB of GDDR5 on a 256-bit bus, yielding 192.3 GB/s of bandwidth. The RX 7900 XT has 20 GB of GDDR6 on a 320-bit bus, yielding 800.0 GB/s of bandwidth—a 4.2x increase in capacity and a 4.2x increase in bandwidth. Clock speeds tell a similar story: the GTX 680 runs at a 1006 MHz base and 1058 MHz boost, while the RX 7900 XT runs at a 1387 MHz base and 2394 MHz boost, with a 2025 MHz game clock. The RX 7900 XT also introduces hardware features the GTX 680 lacks entirely: 84 ray tracing cores and support for DirectX 12 Ultimate (12_2), whereas the GTX 680 is limited to DirectX 12 (11_0). The RX 7900 XT's FP32 throughput is 51.48 TFLOPS versus the GTX 680's 3.250 TFLOPS, and it supports FP16 at 103.0 TFLOPS (2:1), a feature the GTX 680 does not have. The RX 7900 XT also uses a PCIe 4.0 x16 interface and a 2x 8-pin power connector, compared to the GTX 680's PCIe 3.0 x16 and 2x 6-pin connectors.

FAQ

Q: Which card has the higher average benchmark score?

A: The GeForce GTX 680 has a higher average benchmark score of 14150, compared to the Radeon RX 7900 XT's 13745.

Q: Is the GTX 680 better at any modern API benchmark?

A: Yes, the GTX 680 is 32.5% faster than the RX 7900 XT in the Geekbench OpenCL test, scoring 16906 versus 12756.

Q: How large is the RX 7900 XT's Vulkan advantage?

A: The RX 7900 XT scores 71068 in Geekbench Vulkan, which is 75.2% higher than the GTX 680's 17646 score.

Q: Do both cards occupy the same performance percentile?

A: Yes, both the GTX 680 and the RX 7900 XT hold a 55th percentile ranking among all GPUs in the database.

Q: What are the closest rivals to each card based on average score?

A: The GTX 680 is 0.9% above the NVIDIA Tesla K10 (14029) and 0.9% below the GTX 1070 Ti (14277). The RX 7900 XT is 0.5% below the AMD Radeon 660M (13812) and 1.5% above the RX 9070 XT (13543).

Q: Which card supports ray tracing hardware?

A: Only the Radeon RX 7900 XT has ray tracing cores, with 84 of them; the GeForce GTX 680 has none.

Head-to-Head Benchmarks

The only two directly comparable benchmarks in the data are Geekbench OpenCL and Geekbench Vulkan, and each card wins one. The GTX 680's OpenCL victory is its single most impressive result. Scoring 16906 against the RX 7900 XT's 12756, it achieves a 32.5% lead. This is a substantial margin that suggests the Kepler architecture's compute scheduling is highly efficient for this API, despite having roughly one-third the shading units and one-quarter the texture units of the RDNA 3.0 card. In the context of its nearest rivals, this score helps the GTX 680 achieve an average of 14150, which is 0.9% above the Tesla K10 and only 0.9% below the GTX 1070 Ti.

The RX 7900 XT's Vulkan performance is the most dominant result in the entire comparison. Its 71068 score versus 17646 is a 75.2% blowout, meaning the RX 7900 XT is roughly four times faster in this API. This is the kind of delta that indicates a completely different class of hardware. The RX 7900 XT's 84 ray tracing cores and DirectX 12 Ultimate support are part of this, but the raw compute power is also evident: its 51.48 TFLOPS FP32 throughput is over 15 times that of the GTX 680's 3.250 TFLOPS. This single benchmark explains why the RX 7900 XT, despite its lower OpenCL score and lower average, is the more capable modern card. Its Vulkan score alone is over five times the GTX 680's average benchmark score of 14150.

Specification Differences

The specification sheet shows a generational leap in nearly every measurable category. Process node and transistor count: The RX 7900 XT uses a 5 nm process versus the GTX 680's 28 nm, and packs 57,700 million transistors versus 3,540 million. Memory: The RX 7900 XT has 20 GB of GDDR6 on a 320-bit bus with 800.0 GB/s bandwidth, versus 2 GB of GDDR5 on a 256-bit bus with 192.3 GB/s. Core configuration: The RX 7900 XT has 5376 shading units, 336 TMUs, and 192 ROPs, versus the GTX 680's 1536 shading units, 128 TMUs, and 32 ROPs. Clocks: The RX 7900 XT has a 1387 MHz base and 2394 MHz boost, versus the GTX 680's 1006 MHz base and 1058 MHz boost. Compute rates: The RX 7900 XT delivers 51.48 TFLOPS FP32, 459.6 GPixel/s pixel rate, and 804.4 GTexel/s texture rate, versus the GTX 680's 3.250 TFLOPS, 33.86 GPixel/s, and 135.4 GTexel/s. Power and interface: The RX 7900 XT has a 300 W TDP with 2x 8-pin connectors and a 700 W suggested PSU, versus the GTX 680's 195 W TDP with 2x 6-pin and a 450 W suggested PSU. It also uses PCIe 4.0 x16 versus PCIe 3.0 x16. API support: The RX 7900 XT supports DirectX 12 Ultimate (12_2), Vulkan 1.4, and has 84 ray tracing cores; the GTX 680 is limited to DirectX 12 (11_0), Vulkan 1.2.175, and has no ray tracing cores. Physical dimensions: The RX 7900 XT is longer (276 mm vs 256 mm) and wider (51 mm vs 38 mm), though nearly the same height (110 mm vs 111 mm).