NVIDIA GeForce GTX 1660 vs NVIDIA GeForce GTX 480 Comparison

Head-to-Head Benchmarks

The only direct benchmark comparison recorded in the database is Geekbench OpenCL, and the result is decisive. The NVIDIA GeForce GTX 1660 scores 47,850, while the NVIDIA GeForce GTX 480 manages 13,300. That is a 72.2% advantage for the GTX 1660 in this compute workload. In other words, the GTX 480 delivers roughly a fraction over one-quarter of the GTX 1660's OpenCL throughput, a gap so large that no other metric in the database narrows meaningfully. The GTX 480 wins zero recorded head-to-head tests against the GTX 1660, while the GTX 1660 wins one. There is no recorded test in the database where the older Fermi card comes out ahead of the Turing part in a direct comparison. Every win in the head-to-head table belongs to the GTX 1660, and it holds that single victory by a margin that dwarfs any delta recorded for either card in their respective nearest rival lists.

Context from the database's nearest rival tables reinforces how different these two cards are in raw compute. The GTX 480's closest rivals in average benchmark score are the AMD Radeon Pro 555X at 13,321 (a 0.2% delta), the AMD FirePro M6100 at 13,354 (0.4%), the AMD Radeon RX 5500M at 13,356 (0.4%), and the AMD Radeon HD 8950M at 13,376 (0.6%). Those deltas are measured relative to the GTX 480's own average score of 13,300. The GTX 1660, by contrast, sits with an average benchmark score of 11,680, and its nearest rivals are the AMD Radeon RX 7800 XT at 11,627 (0.5%), the AMD Radeon Pro 5500M at 11,528 (1.3%), the AMD Radeon RX 6500 XT at 11,842 (a negative 1.4% delta), and the NVIDIA Tesla K20c at 11,479 (1.8%). The GTX 1660's average score is actually lower than its single OpenCL score of 47,850 because that average includes a wide spread of other benchmarks, including Passmark tests that range from 49 in DirectX 12 to 177 in DirectX 9. The GTX 480 has only one benchmark recorded, Geekbench OpenCL, so its average score equals that single result.

The percentile rankings tell a similar story from a different angle. The GTX 480 lands at the 53rd percentile among all GPUs in the database, while the GTX 1660 sits at the 51st percentile. Despite the GTX 1660's massive OpenCL lead, its aggregate percentile is slightly lower, which reflects the fact that its average score is dragged down by a broader set of legacy DirectX and 2D tests. The GTX 480's percentile is based on a single compute benchmark, so it does not carry the same multi-test baggage. A builder looking at raw compute should favor the GTX 1660 without hesitation. A builder looking at the database's overall percentile might see the two cards as near neighbors, but that view omits the 72.2% OpenCL gap.

Architecture Differences

The GTX 480 is built on the GF100 chip with the Fermi architecture, manufactured on a 40 nm process at TSMC. The GTX 1660 uses the TU116 chip with the Turing architecture, also from TSMC but on a 12 nm process. That process shrink is accompanied by a dramatic change in transistor count: the GTX 480 packs 3,100 million transistors on a 529 mm² die, while the GTX 1660 fits 6,600 million transistors on a 284 mm² die. Transistor density rises from 5.9 million per mm² on Fermi to 23.2 million per mm² on Turing. The GTX 480's die is nearly twice the physical size of the GTX 1660's die, yet it contains fewer than half the transistors. This is the core story of the architectural split: newer node technology allows far more logic in a smaller footprint, at the cost of a denser design.

Memory subsystems differ substantially as well. The GTX 480 carries 1536 MB of GDDR5 across a 384 bit bus, yielding a bandwidth of 177.4 GB/s. The GTX 480 memory clock is listed at 924 MHz with 3.7 Gbps effective. The GTX 1660 has 6 GB of GDDR5 on a 192 bit bus, with bandwidth of 192. The GTX 1660 memory clock runs at 2001 MHz with 8 Gbps effective. The GTX 1660's bandwidth edge is modest, 192.1 GB/s, but its memory size advantage is 6 GB versus 1536 MB, a 4x difference in capacity. The GTX 480 compensates with a wider bus width, 384 bit versus 192 bit, which historically helps with high-bandwidth scenarios. Pixel fill rates follow the same pattern. The GTX 480 reaches 21.03 GPixel/s pixel rate, while the GTX 1660 hits 85.68 GPixel/s, and texture rate jumps from 480's 42.06 GTexel/s to 1660's 157.1 GTexel/s. Shading unit counts move from 480 on Fermi to 1408 on Turing, and TMUs increase from 60 to 88. Texture units on the GTX 480: 60 TMUs on Fermi, 88 on Turing. ROPS remain equal at 48 on both cards, which is a rare point of parity in this comparison, as is the dual-slot width shared by both.

Compute capabilities diverge in FLOP counts as well. The GTX 480 delivers 1,345.0 GFLOPS of FP32 throughput, while the GTX 1660 delivers 5.027 TFLOPS of FP32. The GTX 1660 also lists FP16 performance at 10.05 TFLOPS (2:1), while the GTX 480 has no recorded FP16 figure. The GTX 1660's FP32 output is roughly 3.74x the GTX 480's FP32 output, a ratio that mirrors the OpenCL score gap. API support differs too. The GTX 480 supports DirectX 12 (11_0) on the GTX 480, while the GTX 1660 supports DirectX 12 (12_1) on Turing, and Vulkan support is absent on the GTX 480, while the GTX 1660 lists Vulkan 1.4. Both cards support OpenGL 4.6. Bus interface changes from PCIe 2.0 x16 on Fermi to PCIe 3.0 x16 on Turing, and board power characteristics shift from a 250 W TDP on the GTX 480 to 120 W TDP on the GTX 1660, with power connectors changing from 1x 6-pin + 1x 8-pin to 1x 8-pin on the GTX 1660. Suggested PSU recommendation falls from 600 W on the GTX 1660 to 300 W on the GTX 1660. The GTX 480 length is 267 mm (10.5 inches), while the GTX 1660 length is 229 mm (9 inches), with additional GTX 1660 dimensions of 111 mm height by 35 mm width, fields that are absent for the GTX 480. Display outputs on the GTX 480 are 2x DVI and 1x mini-HDMI 1.3a, against the GTX 1660's 1x DVI, 1x HDMI 2.0, and 1 DisplayPort 1.4a ports. Release timing differs by nearly a decade, with the GTX 480 launching 2010-03-25 and the GTX 1660 launching 2010-03-25, with the GTX 480 predecessor GeForce 200 series and successor GeForce 500 series for the GTX 1660 predecessor GeForce 10-series and successor GeForce 20-series.

The Verdict

The database points to the GTX 1660 for nearly every practical computing scenario. The GTX 1660 wins the only direct head-to-head benchmark by 72.2%, and its architecture supports a far larger feature set, more than double the transistor count on a smaller die, and a higher transistor density. The GTX 480's only advantages in the recorded data are its wider memory bus, its lower memory clock, and its lower TDP, its lower power connector count, and its lower pixel rate. The GTX 480 has a 384 bit memory bus exceeds the GTX 1660's 192 bit bus. The GTX 480's 48 TMUs equal the GTX 1660's 48 TMUs. The GTX 480 1,345.0 GFLOPS in FP32 is a fraction of the GTX 1660's 5.027 TFLOPS, and the GTX 480's 40 nm process and 3.1 billion transistor count is older and sparser than the GTX 1660's 12 nm and 6.6 billion transistors.

For a new build in the modern era, choose the GTX 1660. The GTX 1660's 12 nm Turing architecture with 23.2M transistors per mm² is a generational leap beyond Fermi's 40 nm Fermi GF100, and the 1660's 6 GB GDDR5 on a 192 bit bus with 192.1 GB/s bandwidth is enough for modern workloads. The GTX 480, with its 1536 MB GDDR5 and 384 bit bus, remains a legacy part for legacy use cases. The GTX 480's 250 W TDP and 1x 8-pin plus 1x 8-pin power connector configuration. The GTX 480, at 267 mm long, is a long card for a 229 mm GTX 1660. The GTX 480's 1,345.0 GFLOPS FP32 output is a quarter of the GTX 1660's 5.027 TFLOPS. The GTX 480's 21.03 GPixel/s pixel rate is lower than the GTX 1660's 85.03 GPixel/s. The GTX 480's 42.06 GTexel/s texture rate is lower than the 1660's 157.1 GTexel/s.

Pick the GTX 1660 for compute, for Vulkan, and for DirectX 12 (12_1) features. The GTX 480, with DirectX 12 (11_0), OpenGL 4.6. The GTX 480's 6,600 million transistors on a 284 mm² die, vs the GTX 480's 3,100 million transistors on a 529 mm² die. The GTX 1660's transistor density of 23.2M / mm². The GTX 480's 10.05 TFLOPS FP16 (2:1) capability. The GTX 1660's 5.027 TFLOPS FP32 output. The GTX 1660's 192.1 GB/s bandwidth. The GTX 480's 177.4 GB/s bandwidth. The GTX 480's 48 ROPs. The GTX 1660's 48 ROPs. The GTX 480's 60 TMUs. The GTX 1660's 88 TMUs. The GTX 480's 480 shading units. The GTX 1660's 1408 shading units. The GTX 480's 1,345.0 GFLOPS FP32. The GTX 1660's 5.027 TFLOPS FP32. The GTX 480's 1,345.0 GFLOPS FP32 output is 26.7% of the GTX 1660's 5.027 TFLOPS FP32 output, a difference of exactly 3.7 times.

FAQ

Q: Which GPU has a higher Geekbench OpenCL score?

A: The NVIDIA GeForce GTX 1660 scores 47,850, while the NVIDIA GeForce GTX 480 scores 13,300, giving the GTX 1660 a 72.2% lead in that test.

Q: How do their transistor counts compare?

A: The GTX 480 has 3,100 million transistors on a 529 mm² die, while the GTX 1660 has 6,600 million transistors on a 284 mm² die. The GTX 1660 has more than double the transistor count in roughly half the die area.

Q: What are the memory capacities and bus widths?

A: The GTX 480 uses 1536 MB of GDDR5 on a 384 bit bus, while the GTX 1660 uses 6 GB of GDDR5 on a 192 bit bus. The GTX 480 has the wider bus, but the GTX 1660 has four times the capacity.

Q: Do both cards support the same APIs?

A: No. The GTX 480 supports DirectX 12 (11_0) and OpenGL 4.6, with no Vulkan support listed. The GTX 1660 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4.

Q: What are the power requirements?

A: The GTX 480 has a 250 W TDP and requires a 600 W suggested PSU with 1x 6-pin and 1x 8-pin connectors. The GTX 1660 has a 120 W TDP and requires a 300 W suggested PSU with a single 8-pin connector.

Q: Which GPU has the higher pixel fill rate?

A: The GTX 1660 reaches 85.68 GPixel/s, while the GTX 480 reaches 21.03 GPixel/s, making the GTX 1660 more than four times faster in this metric.

Where Each One Wins

The GTX 1660 wins in every recorded head-to-head benchmark, with a single Geekbench OpenCL score of 47,850 versus 13,300. It also wins on transistor count (6,600 million versus 3,100 million), transistor density (23.2M per mm² versus 5.9M per mm²), memory capacity (6 GB versus 1536 MB), memory bandwidth (192.1 GB/s versus 177.4 GB/s), shading units (1408 versus 480), TMUs (88 versus 60), pixel rate (85.68 GPixel/s versus 21.03 GPixel/s), texture rate (157.1 GTexel/s versus 42.06 GTexel/s), FP32 compute (5.027 TFLOPS versus 1,345.0 GFLOPS), FP16 capability (10.05 TFLOPS (2:1) versus no figure), API support (DirectX 12 (12_1) and Vulkan 1.4 versus DirectX 12 (11_0) and no Vulkan), PCIe interface (3.0 versus 2.0), and display outputs (HDMI 2.0 and DisplayPort 1.4a versus mini-HDMI 1.3a). The GTX 1660 also has the lower TDP (120 W versus 250 W), lower PSU recommendation (300 W versus 600 W), and a shorter board length (229 mm versus 267 mm). The GTX 1660 has a higher average benchmark score among its nearest rivals context? No, the GTX 480's average benchmark score is 13,300, while the GTX 1660's is 11,680, meaning the GTX 480 has the higher average score in the database, despite losing the head-to-head OpenCL test by a wide margin. That is a nuance: the GTX 480's single benchmark average sits above the GTX 1660's multi-test average, and its percentile ranking is also slightly higher (53rd versus 51st). So the GTX 480 wins on average benchmark score and percentile, plus it has a wider memory bus (384 bit versus 192 bit), the same ROP count (48), and a larger die area (529 mm² versus 284 mm²). Its launch MSRP was 499 USD once, while the GTX 1660's launch MSRP was 219 USD, but pricing is not a factor in this analysis beyond noting the MSRP figures.

Specification Differences

The two cards differ in nearly every recorded field. Process node: 40 nm on the GTX 480, 12 nm on the GTX 1660. Foundry is TSMC for both. Transistors: 3,100 million versus 6,600 million. Die size: 529 mm² versus 284 mm². Transistor density: 5.9M per mm² versus 23.2M per mm². Memory size: 1536 MB versus 6 GB. Memory type: GDDR5 on both. Memory bus: 384 bit versus 192 bit. Memory bandwidth: 177.4 GB/s versus 192.1 GB/s. Shading units: 480 versus 1408. TMUs: 60 versus 88. ROPs: 48 on both. Pixel rate: 21.03 GPixel/s versus 85.68 GPixel/s. Texture rate: 42.06 GTexel/s versus 157.1 GTexel/s. FP32: 1,345.0 GFLOPS versus 5.027 TFLOPS. FP16: not listed versus 10.05 TFLOPS (2:1). TDP: 250 W versus 120 W. Power connectors: 1x 6-pin + 1x 8-pin versus 1x 8-pin. Suggested PSU: 600 W versus 300 W. Bus interface: PCIe 2.0 x16 versus PCIe 3.0 x16. Display outputs: 2x DVI and 1x mini-HDMI 1.3a versus 1x DVI, 1x HDMI 2.0, and 1x DisplayPort 1.4a. DirectX: 12 (11_0) versus 12 (12_1). OpenGL: 4.6 on both. Vulkan: not listed versus 1.4. Dimensions: 267 mm length versus 229 mm length, with height and width listed only for the GTX 1660 (111 mm height, 35 mm width). Architecture: Fermi (GF100) versus Turing (TU116). Generation: GeForce 400 versus GeForce 16. Release date: 2010-03-25 versus 2019-03-13. Predecessor: GeForce 200 versus GeForce 10. Successor: GeForce 500 versus GeForce 20. Production status: end-of-life for both. Percentile: 53rd versus 51st. Average benchmark score: 13,300 versus 11,680. Launch MSRP: 499 USD versus 219 USD.