AMD Radeon RX 580 vs NVIDIA GeForce GTX 680 Comparison

The recorded data shows a decisive generational shift between these two end-of-life graphics cards. In every shared benchmark, the AMD Radeon RX 580 outperforms the NVIDIA GeForce GTX 680 by a wide margin, reflecting a five-year gap in architecture and process technology. The GTX 680, a Kepler-era flagship from 2012, retains only its place in history; the RX 580, a Polaris part from 2017, is the clear performance winner across the board.

Head-to-Head Benchmarks

The head-to-head results are unambiguous. The AMD Radeon RX 580 wins all three recorded comparisons, with no benchmark favoring the NVIDIA GeForce GTX 680.

In Geekbench Metal, the RX 580 scores 45,235 against the GTX 680's 7,897. That is a delta of -82.5% for the GTX 680, meaning the RX 580 delivers roughly 5.7 times the Metal performance. This is the largest gap in the dataset, and it highlights how far compute-oriented API performance has advanced.

Geekbench OpenCL tells a similar story. The RX 580 posts 37,453, while the GTX 680 manages 16,906. The delta is -54.9%, indicating the RX 580 is about 2.2 times faster in OpenCL workloads. This aligns with the raw FP32 throughput figures, where the RX 580's 6.175 TFLOPS nearly doubles the GTX 680's 3.250 TFLOPS.

Geekbench Vulkan narrows the relative gap slightly but still favors AMD heavily. The RX 580 records 45,173, the GTX 680 records 17,646, and the delta stands at -60.9%. In percentage terms, the RX 580 is roughly 2.6 times faster. Across all three tests, the AMD card's advantage ranges from 2.2x to 5.7x, making any direct competition moot.

Looking at the wider database context, the GTX 680's average benchmark score is 14,150, placing it at the 55th percentile of all GPUs. Its nearest rivals include the NVIDIA Tesla K10 at 14,029 (0.9% slower), the NVIDIA GeForce GTX 1070 Ti at 14,277 (0.9% faster), and the Intel Iris Xe MAX Graphics at 14,315 (1.2% faster). The RX 580, by contrast, has an average score of 12,928, which sits at the 53rd percentile. Its nearest rivals are the NVIDIA GeForce RTX 3050 Ti Mobile at 12,940 (0.1% faster), the NVIDIA GeForce GTX 1660 SUPER at 12,986 (0.4% faster), and the AMD Radeon 740M at 12,870 (0.5% slower). Interestingly, despite the RX 580 winning every head-to-head test, its overall average score is lower than the GTX 680's. This is because the RX 580's benchmark suite includes additional tests like Passmark DirectX 9, 10, 11, and 12, which pull its average down, whereas the GTX 680 only has Geekbench entries. The data does not allow a direct apples-to-apples average comparison across all tests.

Architecture Differences

The two cards come from different process nodes and design philosophies. The GTX 680 uses NVIDIA's Kepler architecture on a 28 nm process from TSMC, with a die size of 294 mm² and 3,540 million transistors. The RX 580 uses AMD's GCN 4.0 architecture on a 14 nm process from GlobalFoundries, with a die size of 232 mm² and 5,700 million transistors.

The transistor density tells the story of manufacturing progress. The GTX 680 packs 12.0 million transistors per mm², while the RX 580 achieves 24.6 million per mm², more than double the density. This allows AMD to fit far more logic into a smaller physical package.

The compute configurations differ significantly. The GTX 680 has 1,536 shading units, 128 texture mapping units, and 32 ROPs. The RX 580 has 2,304 shading units, 144 texture units, and the same 32 ROPs. The RX 580's higher shading unit count directly explains its FP32 advantage: 6.175 TFLOPS versus 3.250 TFLOPS. The RX 580 also supports FP16 at a 1:1 ratio (6.175 TFLOPS), while the GTX 680 has no recorded FP16 capability.

Memory is another major divergence. The GTX 680 comes with 2 GB of GDDR5 on a 256-bit bus, yielding 192.3 GB/s of bandwidth. The RX 580 ships with 8 GB of GDDR5 on the same 256-bit bus, but at a higher effective speed of 8 Gbps, producing 256.0 GB/s. That is a 33% bandwidth increase, which matters for texture-heavy and high-resolution workloads.

Pixel and texture rates follow the same pattern. The GTX 680 achieves 33.86 GPixel/s and 135.4 GTexel/s. The RX 580 reaches 42.88 GPixel/s and 193.0 GTexel/s. The RX 580 leads by 26.6% in pixel fill and 42.6% in texture fill.

API support also differs. The GTX 680 lists DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175. The RX 580 lists DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3. The RX 580's full DirectX 12_0 feature level and newer Vulkan revision indicate better support for modern rendering features.

Power and physical design show trade-offs. The GTX 680 has a 195 W TDP and requires two 6-pin power connectors. The RX 580 has a lower 185 W TDP and needs a single 8-pin connector. Both recommend a 450 W power supply. The GTX 680 is slightly longer at 256 mm versus the RX 580's 241 mm, but both are dual-slot designs.

The Verdict

The data is clear: the AMD Radeon RX 580 is the superior card for any modern workload. It wins every head-to-head benchmark, offers more memory, higher bandwidth, more shading units, and better API support.

The GTX 680's only redeeming quality is its higher percentile rank (55th versus 53rd) and higher average score (14,150 versus 12,928), but that average is skewed by the different test suites. In direct comparison, the RX 580 is between 2.2 and 5.7 times faster depending on the API.

For users with legacy software that relies on older DirectX versions, the RX 580 also holds an edge. Passmark results show the RX 580 scoring 124 in DirectX 9, 46 in DirectX 10, 60 in DirectX 11, and 43 in DirectX 12. The GTX 680 has no recorded Passmark scores, so no direct comparison is possible for those specific tests.

The RX 580's 8 GB memory capacity makes it suitable for modern game textures and compute buffers, while the GTX 680's 2 GB is a hard limitation. The RX 580 also consumes 10 W less power despite delivering nearly double the FP32 throughput, evidence of the 14 nm process advantage.

Specification Differences

The two cards differ on nearly every specification that matters.

Process node: the GTX 680 uses 28 nm TSMC, the RX 580 uses 14 nm GlobalFoundries. Transistor count: 3,540 million versus 5,700 million. Die size: 294 mm² versus 232 mm². Transistor density: 12.0M per mm² versus 24.6M per mm².

Clock speeds: the GTX 680 runs at 1006 MHz base and 1058 MHz boost, with memory at 1502 MHz (6 Gbps effective). The RX 580 runs at 1257 MHz base and 1340 MHz boost, with memory at 2000 MHz (8 Gbps effective).

Memory: 2 GB versus 8 GB, both GDDR5, both 256-bit, but bandwidth differs at 192.3 GB/s versus 256.0 GB/s.

Compute units: 1,536 shading units versus 2,304, 128 TMUs versus 144, and 32 ROPs on both. FP32: 3.250 TFLOPS versus 6.175 TFLOPS. FP16: not listed versus 6.175 TFLOPS (1:1).

Power: 195 W TDP versus 185 W TDP. Power connectors: 2x 6-pin versus 1x 8-pin. Suggested PSU is 450 W for both.

Display outputs: the GTX 680 has 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The RX 580 has 1x HDMI 2.0b and 3x DisplayPort 1.4a.

APIs: DirectX 12 (11_0) versus DirectX 12 (12_0), OpenGL 4.6 on both, Vulkan 1.2.175 versus 1.3.

Dimensions: the GTX 680 is 256 mm long, 111 mm high, 38 mm wide. The RX 580 is 241 mm long, with height and width not recorded. Both are dual-slot.

Release dates: the GTX 680 launched on 2012-03-21, the RX 580 on 2017-04-17. The GTX 680's launch MSRP was 499 USD; the RX 580's was 229 USD.

FAQ

Q: Which card wins in Geekbench Metal?

A: The AMD Radeon RX 580 wins with a score of 45,235, compared to the NVIDIA GeForce GTX 680's 7,897, a delta of -82.5% for the GTX 680.

Q: How much faster is the RX 580 in OpenCL?

A: The RX 580 scores 37,453 versus the GTX 680's 16,906, a delta of -54.9%, meaning the RX 580 is roughly 2.2 times faster.

Q: What is the memory capacity difference?

A: The GTX 680 has 2 GB of GDDR5, while the RX 580 has 8 GB of GDDR5. Both use a 256-bit bus.

Q: Does the RX 580 support FP16?

A: Yes, the RX 580 lists FP16 at 6.175 TFLOPS with a 1:1 ratio. The GTX 680 has no recorded FP16 capability.

Q: Which card has a higher TDP?

A: The GTX 680 has a 195 W TDP, while the RX 580 has a 185 W TDP. The RX 580 draws less power despite higher performance.

Q: What is the DirectX support difference?

A: The GTX 680 supports DirectX 12 (11_0), while the RX 580 supports DirectX 12 (12_0), which includes the full feature level.

Where Each One Wins

The AMD Radeon RX 580 wins in every recorded benchmark category. It dominates Geekbench Metal, OpenCL, and Vulkan. Its 8 GB memory and 256.0 GB/s bandwidth make it the choice for modern games, compute workloads, and any application that benefits from large texture pools or FP16 processing.

The NVIDIA GeForce GTX 680 wins only in the abstract sense of its higher overall percentile rank (55th versus 53rd) and higher average benchmark score (14,150 versus 12,928). That advantage exists only because the GTX 680 was tested with a smaller, less demanding benchmark suite. In any direct head-to-head test, the RX 580 is faster.

For users constrained by power connectors, the RX 580 requires a single 8-pin, while the GTX 680 needs two 6-pin connectors. Both fit in dual-slot layouts, but the RX 580 is shorter at 241 mm versus 256 mm, which may improve case compatibility.

For display connectivity, the RX 580 offers HDMI 2.0b and three DisplayPort 1.4a outputs, supporting higher refresh rates and modern monitors. The GTX 680's HDMI 1.4a and DisplayPort 1.2 are outdated for current displays.

The verdict is simple: the RX 580 is the better card for every use case represented in the data. The GTX 680 is a historical artifact, outclassed in performance, memory, and efficiency.