NVIDIA GeForce GTX 480 vs NVIDIA GeForce GTX 680 Comparison

# The Verdict

The data presents a clear generational leap. The NVIDIA GeForce GTX 680 holds a decisive advantage over the GTX 480, winning the sole head-to-head benchmark and posting a significantly higher average score. The GTX 680 achieves an average benchmark score of 14,150, while the GTX 480 trails at 13,300. This represents a 27.1% lead for the GTX 680 in the Geekbench OpenCL test, the only direct comparison available.

Looking at the percentile rankings, the GTX 680 sits in the 55th percentile of all GPUs, while the GTX 480 sits in the 53rd. While that two-percentile gap might seem modest, the actual score difference is substantial. The GTX 680's nearest rivals include the NVIDIA Tesla K10 (0.9% ahead), the GeForce GTX 1070 Ti (0.9% behind), and the Intel Iris Xe MAX Graphics (1.2% behind). The GTX 480's closest competitors are all AMD mobile or workstation parts, with the Radeon Pro 555X just 0.2% behind, indicating the older card is now competing in a lower performance tier.

For buyers, the choice is straightforward. The GTX 680 is the superior card in every measured category. It offers higher raw performance, better efficiency metrics, and newer architectural features. The GTX 480, despite its larger die and higher power draw, cannot match the Kepler-based card. The GTX 680 is the pick for anyone seeking better performance and modern feature support. The GTX 480 should only be considered by those with legacy system requirements, such as older PCIe 2.0 motherboards or specific software compatibility needs that predate Kepler.

# Where Each One Wins

The GTX 680 wins the only directly comparable workload — Geekbench OpenCL — with a score of 16,906 versus 13,300 for the GTX 480. That 27.1% margin is substantial and indicates that in compute-oriented tasks, the newer architecture delivers markedly better throughput. The GTX 680 also holds an advantage in API support, offering Vulkan 1.2.175, which the GTX 480 lacks entirely. This makes the GTX 680 the better choice for any modern application leveraging Vulkan for rendering or compute.

The GTX 480 does not win any direct benchmark comparison. Its lone score of 13,300 in Geekbench OpenCL is its only data point, and it trails every competitor in its nearestRivals list by a small margin (0.2% to 0.6% behind). However, there are qualitative areas where the older card retains relevance. The GTX 480 offers a wider 384-bit memory bus compared to the GTX 680's 256-bit bus, which could theoretically benefit certain bandwidth-sensitive workloads, though its actual bandwidth of 177.4 GB/s is lower than the GTX 680's 192.3 GB/s. The GTX 480 also has more ROPs (48 versus 32), which historically helps with fill-rate-bound scenarios, but its pixel rate of 21.03 GPixel/s falls well short of the GTX 680's 33.86 GPixel/s.

For display connectivity, the GTX 680 offers a more modern output set with 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The GTX 480 provides 2x DVI and 1x mini-HDMI 1.3a. The GTX 680's DisplayPort support makes it more suitable for modern multi-monitor setups or high-refresh-rate displays that utilize DisplayPort signaling.

# Architecture Differences

The two cards represent fundamentally different design philosophies from NVIDIA. The GTX 680 is built on the GK104 chip using the Kepler architecture, manufactured on a 28 nm process at TSMC. The GTX 480 uses the GF100 chip with the Fermi architecture, fabricated on a 40 nm process. This process shrink is significant: the GTX 680 packs 3,540 million transistors into a 294 mm² die, achieving a transistor density of 12.0 million per mm². The GTX 480 contains 3,100 million transistors spread across a much larger 529 mm² die, yielding a density of only 5.9 million per mm². The Kepler design is notably more efficient in transistor packing.

Clock speeds further separate the two. The GTX 680 features a base clock of 1006 MHz with a boost clock of 1058 MHz, while the GTX 480 has no listed base or boost clocks, only a memory clock of 924 MHz (3.7 Gbps effective). The GTX 680's memory runs at 1502 MHz (6 Gbps effective), giving it higher bandwidth despite the narrower bus. The GTX 680's shading unit count is dramatically higher at 1,536 versus 480 for the GTX 480, and its texture mapping units total 128 versus 60. The GTX 480 does have more ROPs at 48 versus 32, but the Kepler card's higher clocks compensate.

Compute throughput tells the story of architectural efficiency. The GTX 680 delivers 3.250 TFLOPS of FP32 performance, while the GTX 480 manages 1,345.0 GFLOPS. The GTX 680's texture rate of 135.4 GTexel/s dwarfs the GTX 480's 42.06 GTexel/s, and the pixel rate advantage is similar at 33.86 GPixel/s versus 21.03 GPixel/s. The power envelopes reflect the efficiency gains: the GTX 680 draws 195 W with a suggested 450 W PSU, while the GTX 480 consumes 250 W and requires a 600 W PSU. The GTX 680 uses two 6-pin power connectors; the GTX 480 needs one 6-pin and one 8-pin.

The GTX 680 supports PCIe 3.0 x16, doubling the interface bandwidth of the GTX 480's PCIe 2.0 x16. Both cards offer DirectX 12 (11_0) and OpenGL 4.6, but only the GTX 680 supports Vulkan 1.2.175. The GTX 480 has no Vulkan support listed. The GTX 680 is also slightly shorter at 256 mm versus 267 mm for the GTX 480.

# FAQ

Q: Which card is faster in compute workloads?

A: The GTX 680 is significantly faster. In the Geekbench OpenCL test, it scores 16,906 versus 13,300 for the GTX 480, a 27.1% advantage. Its FP32 throughput of 3.250 TFLOPS also far exceeds the GTX 480's 1,345.0 GFLOPS.

Q: Does the GTX 480 offer any advantages in memory or bandwidth?

A: The GTX 480 has a wider 384-bit memory bus and more ROPs (48 versus 32), but its actual memory bandwidth of 177.4 GB/s is lower than the GTX 680's 192.3 GB/s. The GTX 680's higher memory clock of 6 Gbps effective compensates for the narrower 256-bit bus.

Q: Which card is more power-efficient?

A: The GTX 680 is substantially more efficient. It has a TDP of 195 W and requires a 450 W suggested PSU, while the GTX 480 has a TDP of 250 W and needs a 600 W PSU. The GTX 680 also uses simpler power connectors (2x 6-pin versus 1x 6-pin + 1x 8-pin).

Q: Do both cards support modern graphics APIs?

A: Both support DirectX 12 (11_0) and OpenGL 4.6. However, only the GTX 680 supports Vulkan 1.2.175. The GTX 480 has no Vulkan support listed in its specifications.

Q: How do these cards compare to their nearest rivals?

A: The GTX 680's average score of 14,150 places it within 1.4% of several competitors, including the NVIDIA Tesla K10, GeForce GTX 1070 Ti, Intel Iris Xe MAX Graphics, and AMD Radeon Vega 11. The GTX 480's average of 13,300 sits just behind AMD Radeon Pro 555X, FirePro M6100, Radeon RX 5500M, and Radeon HD 8950M by margins of 0.2% to 0.6%.

Q: Which card has better display output capabilities?

A: The GTX 680 offers a more modern set of outputs: 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The GTX 480 provides 2x DVI and 1x mini-HDMI 1.3a, lacking DisplayPort connectivity.

# Head-to-Head Benchmarks

The only direct benchmark comparison available is the Geekbench OpenCL test, and the result is emphatic. The GTX 680 scores 16,906, while the GTX 480 manages 13,300. That 27.1% delta is the single largest margin in the entire dataset for these two cards. To put it in context, the GTX 680's nearest rivals are all within 1.4% of its average score, meaning this OpenCL lead over the GTX 480 is roughly twenty times larger than the gap between the GTX 680 and its closest competitors.

The GTX 680's win in this test aligns with its architectural advantages. Its FP32 compute of 3.250 TFLOPS is more than double the GTX 480's 1,345.0 GFLOPS. The shading unit count of 1,536 versus 480 gives the Kepler card a 3.2x theoretical advantage in shader throughput. The texture rate tells a similar story: 135.4 GTexel/s versus 42.06 GTexel/s is a 3.2x difference. Even the pixel rate, where the GTX 480's higher ROP count might have helped, favors the GTX 680 at 33.86 GPixel/s versus 21.03 GPixel/s.

The memory subsystem also favors the newer card. The GTX 680's 192.3 GB/s bandwidth exceeds the GTX 480's 177.4 GB/s despite the latter's wider 384-bit bus. The GTX 680's memory runs at 6 Gbps effective versus 3.7 Gbps for the GTX 480, more than compensating for the narrower interface. The GTX 680 also has 2 GB of memory versus 1536 MB on the GTX 480, providing additional headroom for larger textures or datasets.

In summary, the GTX 680 wins the only direct benchmark by a wide margin, and its architectural specifications reinforce that result across every measurable compute category. The GTX 480, while historically significant as a Fermi-generation part, is outclassed in every metric where direct comparison is possible. The 27.1% OpenCL delta is consistent with the theoretical compute advantage of 3.250 TFLOPS versus 1,345.0 GFLOPS, suggesting the benchmark accurately reflects the underlying hardware capabilities. There are no benchmark categories where the GTX 480 emerges victorious.