NVIDIA GeForce GTX 480 vs NVIDIA GeForce GTX 670 Comparison

The NVIDIA GeForce GTX 480 and GTX 670 represent two distinct generations of NVIDIA’s high-end graphics lineup, and the benchmark data shows a clear generational shift in performance and efficiency. In the only head-to-head benchmark available, the GTX 670 decisively outperforms the GTX 480, scoring 15,341 in Geekbench OpenCL compared to the GTX 480’s 13,300. This translates to a 13.3% advantage for the GTX 670, which is a significant margin given that both cards target similar performance tiers. The GTX 480, however, holds its own in terms of raw compute density, though the newer architecture ultimately wins the day.

Head-to-Head Benchmarks

The sole direct comparison in the data is the Geekbench OpenCL test, and it paints a one-sided picture. The GTX 670 scores 15,341, while the GTX 480 scores 13,300, giving the GTX 670 a 13.3% lead. This is not a marginal victory; it is a substantial gap that reflects the architectural improvements between the Fermi and Kepler designs. For context, the GTX 480’s score of 13,300 places it in the 53rd percentile of all GPUs, while the GTX 670’s score of 15,341 places it in the 52nd percentile. Interestingly, the GTX 670’s average benchmark score across all tests is 12,773, which is lower than its OpenCL score, while the GTX 480’s average is exactly 13,300. This suggests the GTX 670’s OpenCL performance is a standout result for that card, whereas the GTX 480 is more consistent across its limited test suite.

The win count is equally lopsided: the GTX 670 wins 1 benchmark, while the GTX 480 wins 0. This is not a close contest. The 13.3% delta is the only measurable difference, and it favors the newer card in every way measured. When looking at the nearest rivals, the GTX 480 sits within 0.2% to 0.6% of cards like the AMD Radeon Pro 555X, FirePro M6100, RX 5500M, and HD 8950M. The GTX 670, meanwhile, trades blows with the GTX 590 and Radeon Pro 455 (both 0.4% slower), the RX 7600M XT (0.5% faster), and the FirePro W5100 (0.6% slower). This shows that while the GTX 670 is clearly faster than the GTX 480, both cards are competitive with their respective peers from the same era.

FAQ

Q: Which card is faster in Geekbench OpenCL?

A: The GTX 670 is faster, scoring 15,341 versus the GTX 480’s 13,300, a 13.3% advantage. This is the only head-to-head benchmark in the data, and the GTX 670 wins it outright.

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

A: The GTX 480 is nearly identical to the AMD Radeon Pro 555X, which scores 13,321 (a 0.2% difference), and the FirePro M6100 at 13,354 (0.4% difference). The GTX 670 is similarly close to the GTX 590, which scores 12,830 (0.4% slower), and the Radeon Pro 455 at 12,831 (0.4% slower).

Q: What are the average benchmark scores for each card?

A: The GTX 480 has an average benchmark score of 13,300, which matches its single OpenCL result. The GTX 670 has an average score of 12,773, which is lower than its OpenCL score of 15,341 because it also runs Metal and Vulkan tests that score lower.

Q: Do these cards support Vulkan?

A: The GTX 670 supports Vulkan 1.2.175, while the GTX 480 has no Vulkan support listed. The GTX 670 also has a Geekbench Vulkan score of 15,553, which is its highest benchmark result.

Q: What is the memory configuration difference?

A: The GTX 480 has 1536 MB of GDDR5 on a 384-bit bus, yielding 177.4 GB/s bandwidth. The GTX 670 has 2 GB of GDDR5 on a 256-bit bus, yielding 192.3 GB/s bandwidth. Despite the narrower bus, the GTX 670 has higher total bandwidth.

Q: Which card has a higher transistor density?

A: The GTX 670 has a much higher transistor density at 12.0M / mm², compared to the GTX 480’s 5.9M / mm². This is due to the GTX 670’s smaller 28 nm process node versus the GTX 480’s 40 nm node.

Architecture Differences

The architectural gap between these two cards is stark, and it explains the performance delta. The GTX 480 uses the GF100 chip built on the Fermi architecture, a 40 nm design from TSMC. The GTX 670 uses the GK104 chip on the Kepler architecture, also from TSMC but on a much more advanced 28 nm process. This process shrink allows the GTX 670 to pack 3,540 million transistors into a 294 mm² die, while the GTX 480 packs 3,100 million transistors into a much larger 529 mm² die. The transistor density tells the story: the GTX 670 achieves 12.0M / mm² versus the GTX 480’s 5.9M / mm².

The shading unit count is a major differentiator. The GTX 480 has 480 shading units, 60 texture mapping units (TMUs), and 48 render output units (ROPs). The GTX 670, by contrast, has 1,344 shading units, 112 TMUs, and only 32 ROPs. This is a massive increase in shading and texturing capability, but a reduction in ROPs. The pixel rate reflects this: the GTX 670 pushes 27.44 GPixel/s versus the GTX 480’s 21.03 GPixel/s, but the texture rate is where the GTX 670 truly shines, at 109.8 GTexel/s versus 42.06 GTexel/s. The FP32 compute performance also jumps dramatically, from 1,345.0 GFLOPS on the GTX 480 to 2.634 TFLOPS on the GTX 670, nearly double.

Memory architecture also differs significantly. The GTX 480 uses a 384-bit bus with 1536 MB of GDDR5, while the GTX 670 uses a 256-bit bus with 2 GB of GDDR5. Despite the narrower bus, the GTX 670’s memory clock is higher, resulting in 192.3 GB/s of bandwidth versus 177.4 GB/s. The GTX 670 also supports newer APIs, including Vulkan 1.2.175, while the GTX 480 has no Vulkan support listed. Both cards support DirectX 12 (11_0) and OpenGL 4.6, but the GTX 670’s feature set is more modern overall.

Specification Differences

The specification sheets reveal several key differences beyond raw performance. The process node is a clear generational jump: the GTX 480 is on 40 nm, while the GTX 670 is on 28 nm. This leads to the transistor and die size differences already noted. The GTX 670 has a base clock of 915 MHz and a boost clock of 980 MHz, while the GTX 480 has no base or boost clock listed, only a memory clock of 924 MHz (3.7 Gbps effective). The GTX 670’s memory clock is 1502 MHz (6 Gbps effective), which is significantly faster.

Power consumption is another major divergence. The GTX 480 has a TDP of 250 W and requires a 600 W suggested PSU, along with a 1x 6-pin + 1x 8-pin power connector setup. The GTX 670 is much more efficient, with a TDP of 170 W and a 450 W suggested PSU, using only 2x 6-pin connectors. This is a 32% reduction in TDP, which is remarkable given the GTX 670’s higher performance. The physical dimensions also differ: the GTX 480 is 267 mm (10.5 inches) long, while the GTX 670 is 241 mm (9.5 inches) long, with listed height of 111 mm and width of 38 mm. Both are dual-slot cards.

The bus interface differs as well. The GTX 480 uses PCIe 2.0 x16, while the GTX 670 uses PCIe 3.0 x16. Display outputs are also different: the GTX 480 has 2x DVI and 1x mini-HDMI 1.3a, while the GTX 670 has 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The GTX 670 offers more modern connectivity options. The launch MSRP is 499 USD for the GTX 480 and 399 USD for the GTX 670, reflecting the GTX 670’s position as a more affordable high-end card at launch.

The Verdict

The data is unambiguous: the GTX 670 is the superior card. It wins the only benchmark, with a 13.3% lead in Geekbench OpenCL, and it does so while consuming significantly less power and offering a more modern feature set. The GTX 480’s only claim to fame is its older architecture, which cannot match the Kepler design’s efficiency or compute throughput. For anyone choosing between these two, the GTX 670 is the clear winner on performance, power draw, and connectivity. The GTX 480’s higher TDP and larger physical footprint are not offset by any measurable performance advantage in the data provided. The GTX 670’s average benchmark score of 12,773 is lower than its OpenCL score, but it still exceeds the GTX 480’s average of 13,300 when considering the full test suite, though that suite includes Metal and Vulkan tests the GTX 480 cannot run. The GTX 670 is the more capable, more efficient, and more future-proof card.

Where Each One Wins

The GTX 670 wins in every measurable category. It has a 13.3% lead in OpenCL performance, which is the only head-to-head result. It also wins on compute power, with 2.634 TFLOPS FP32 versus 1,345.0 GFLOPS, and on texture throughput, with 109.8 GTexel/s versus 42.06 GTexel/s. The GTX 670 also wins on memory bandwidth, at 192.3 GB/s versus 177.4 GB/s, and on pixel rate, at 27.44 GPixel/s versus 21.03 GPixel/s. It wins on efficiency, with a 170 W TDP versus 250 W, and on connectivity, with support for Vulkan and a DisplayPort 1.2 output. The GTX 670’s boost clock of 980 MHz gives it a dynamic performance advantage that the GTX 480 lacks entirely.

The GTX 480’s wins are limited to a few hardware specifications. It has more ROPs, at 48 versus 32, which could theoretically benefit certain pixel-heavy workloads, but the GTX 670’s higher pixel rate suggests this does not translate to real-world gains. The GTX 480 also has a wider memory bus, at 384-bit versus 256-bit, but the GTX 670’s higher memory clock compensates for this. In terms of raw die size, the GTX 480 is larger, but that is a disadvantage, not a win. The GTX 480’s only true advantage is that it was released earlier, on 2010-03-25, versus the GTX 670’s 2012-05-09, making it a more mature product in terms of driver updates at the time. However, the data shows no scenario where the GTX 480 outperforms the GTX 670 in any benchmark or efficiency metric. The GTX 670 is the definitive choice for anyone valuing performance, power draw, or modern features.