NVIDIA GeForce GTX 1660 vs NVIDIA Tesla K20c Comparison

The NVIDIA GeForce GTX 1660 and NVIDIA Tesla K20c are both end-of-life GPUs, but they serve fundamentally different purposes and eras. The data shows a decisive victory for the GTX 1660 in the available head-to-head benchmark, yet the Tesla K20c’s architecture offers distinct compute-oriented traits that matter in specific workloads. The GTX 1660 wins the only shared test by a massive margin, but the Tesla K20c counters with a wider memory bus and higher raw shader count, making it a specialized tool rather than a general-purpose performer.

Where Each One Wins

The GTX 1660 dominates in every measurable benchmark category where both cards have data. In the sole head-to-head test, Geekbench OpenCL, the GTX 1660 scores 47,850 against the Tesla K20c’s 11,479, a 316.8% advantage. This is not a close contest; it is a generational gap expressed in raw compute throughput. The GTX 1660’s average benchmark score of 11,680 also edges out the Tesla K20c’s 11,479, reinforcing its overall superiority in general-purpose compute tasks.

The Tesla K20c’s only wins are architectural, not benchmark-based. It packs 2,496 shading units versus the GTX 1660’s 1,408, and its 208 TMUs dwarf the GTX 1660’s 88. This suggests the Tesla K20c could excel in workloads that scale with shader count and texture operations, such as certain scientific simulations or rendering tasks that are not captured in the available benchmark suite. However, the data does not include any test where the Tesla K20c actually wins, so these advantages remain theoretical rather than proven.

For gaming and consumer-facing DirectX workloads, the GTX 1660 is the clear choice. Its PassMark scores—177 in DirectX 9, 79 in DirectX 11, 49 in DirectX 12, and 61 in DirectX 10—indicate functional driver support across legacy and modern APIs. The Tesla K20c has no display outputs, making it unsuitable for any interactive graphics work. The GTX 1660 also wins on efficiency metrics, with a 120 W TDP versus the Tesla K20c’s 225 W, and a smaller 300 W suggested PSU versus 550 W.

Architecture Differences

The two cards represent opposite ends of NVIDIA’s architectural evolution. The GTX 1660 uses the TU116 chip on the Turing architecture, built on a 12 nm process at TSMC. It packs 6,600 million transistors into a 284 mm² die, achieving a transistor density of 23.2M per mm². The Tesla K20c uses the GK110 chip on the older Kepler architecture, fabricated on a 28 nm process, also at TSMC. It holds 7,080 million transistors on a much larger 561 mm² die, with a lower density of 12.6M per mm². This means the Tesla K20c has more transistors overall, but the GTX 1660 packs them more efficiently.

Memory configurations diverge sharply. The GTX 1660 has 6 GB of GDDR5 on a 192-bit bus, delivering 192.1 GB/s of bandwidth. The Tesla K20c has 5 GB of GDDR5 on a wider 320-bit bus, achieving 208.0 GB/s. Despite having less memory, the Tesla K20c’s wider bus gives it higher bandwidth, which can benefit memory-intensive compute tasks. Clock speeds also differ: the GTX 1660 has a base of 1530 MHz and a boost of 1785 MHz, while the Tesla K20c lists no base or boost clocks, only a memory clock of 1300 MHz (5.2 Gbps effective).

Compute capabilities highlight the generational shift. The GTX 1660 delivers 5.027 TFLOPS of FP32 performance and 10.05 TFLOPS of FP16 (2:1), while the Tesla K20c manages only 3.524 TFLOPS of FP32 and has no FP16 support. The GTX 1660 also has higher pixel and texture rates: 85.68 GPixel/s and 157.1 GTexel/s, versus the Tesla K20c’s 36.71 GPixel/s and 146.8 GTexel/s. The Tesla K20c counters with more shading units (2,496 vs 1,408), more TMUs (208 vs 88), and more ROPs in the GTX 1660’s favor (48 vs 40).

Head-to-Head Benchmarks

The only direct comparison in the data is the Geekbench OpenCL test, and it is a landslide. The GTX 1660 scores 47,850, while the Tesla K20c scores 11,479. This yields a delta of 316.8% in favor of the GTX 1660. To put that in context, the GTX 1660’s nearest rival in its own list, the AMD Radeon RX 7800 XT, is only 0.5% ahead, and the Tesla K20c is 1.8% behind. But in the head-to-head, the GTX 1660 is over four times faster. This suggests the Geekbench OpenCL workload heavily favors the Turing architecture’s newer instruction set and higher clock speeds.

Looking at the broader benchmark picture, the GTX 1660 has 10 recorded benchmarks, including PassMark G3D at 11,646 and PassMark GPU Compute at 4,963. The Tesla K20c has only one recorded benchmark, the Geekbench OpenCL score. This disparity in data availability itself tells a story: the Tesla K20c is a niche compute card with limited public benchmarking, while the GTX 1660 was widely tested across consumer workloads. The GTX 1660’s avgBenchmarkScore of 11,680 places it at the 51st percentile of all GPUs, exactly matching the Tesla K20c’s percentile, but the GTX 1660 achieves this with a broader and higher-scoring test suite.

The GTX 1660’s nearest rivals include the AMD Radeon Pro 5500M (1.3% ahead) and the AMD Radeon RX 6500 XT (1.4% behind), showing it sits in a competitive mid-range consumer segment. The Tesla K20c’s nearest rivals include the NVIDIA GeForce GTX 780M (1.9% ahead) and the AMD Radeon RX 7800 XT (1.3% behind), indicating it is bracketed by older and newer parts alike. Neither card is a performance leader, but the GTX 1660’s benchmark presence is far more robust.

FAQ

Q: Which card is faster in the only shared benchmark?

A: The GTX 1660 is 316.8% faster in Geekbench OpenCL, scoring 47,850 versus the Tesla K20c’s 11,479.

Q: Does the Tesla K20c have any benchmark wins?

A: No. The data records 0 wins for the Tesla K20c and 1 win for the GTX 1660 in head-to-head tests.

Q: Which card has more shading units?

A: The Tesla K20c has 2,496 shading units, while the GTX 1660 has 1,408. However, the GTX 1660 still delivers higher FP32 performance at 5.027 TFLOPS versus 3.524 TFLOPS.

Q: What are the memory bandwidth differences?

A: The Tesla K20c has a 320-bit bus and 208.0 GB/s bandwidth, while the GTX 1660 has a 192-bit bus and 192.1 GB/s bandwidth. The Tesla K20c is faster here despite having less memory (5 GB vs 6 GB).

Q: Can the Tesla K20c output video?

A: No. The Tesla K20c has no display outputs, while the GTX 1660 has 1x DVI, 1x HDMI 2.0, and 1x DisplayPort 1.4a.

Q: Which card has better API support?

A: The GTX 1660 supports DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.4. The Tesla K20c supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.

The Verdict

The GTX 1660 is the clear winner for almost any use case. It is 316.8% faster in the head-to-head benchmark, has a higher average benchmark score (11,680 vs 11,479), and offers modern API support including DirectX 12 (12_1) and Vulkan 1.4. It also has display outputs, making it usable for gaming or workstation graphics, and it consumes less power (120 W vs 225 W). The GTX 1660’s launch MSRP was 219 USD, and it was released in 2019, while the Tesla K20c launched in 2012 with a launch MSRP of 3,199 USD.

The Tesla K20c is a specialized compute card with no display outputs, so it is only relevant for headless compute clusters or scientific workloads that leverage its 2,496 shading units and 208 TMUs. Its wider 320-bit memory bus provides higher bandwidth (208.0 GB/s) than the GTX 1660, which could benefit memory-bound tasks. However, the lack of FP16 support and lower FP32 throughput make it less versatile. The Tesla K20c’s only advantage is architectural, not demonstrated in any benchmark win.

For a consumer or general-purpose workstation, the data unequivocally recommends the GTX 1660. For a niche compute deployment where raw shader count and memory bandwidth matter more than clock speed or power efficiency, the Tesla K20c might still have a role, but the benchmark evidence does not support that choice. The GTX 1660 wins on every metric with actual performance data.

Specification Differences

| Field | NVIDIA GeForce GTX 1660 | NVIDIA Tesla K20c |

|-------|--------------------------|-------------------|

| Architecture | Turing | Kepler |

| Chip | TU116 | GK110 |

| Process Node | 12 nm | 28 nm |

| Transistors | 6,600 million | 7,080 million |

| Die Size | 284 mm² | 561 mm² |

| Transistor Density | 23.2M / mm² | 12.6M / mm² |

| Base Clock | 1530 MHz | null |

| Boost Clock | 1785 MHz | null |

| Memory Clock | 2001 MHz (8 Gbps effective) | 1300 MHz (5.2 Gbps effective) |

| Memory Size | 6 GB | 5 GB |

| Memory Bus Width | 192 bit | 320 bit |

| Memory Bandwidth | 192.1 GB/s | 208.0 GB/s |

| Shading Units | 1408 | 2496 |

| TMUs | 88 | 208 |

| ROPs | 48 | 40 |

| Pixel Rate | 85.68 GPixel/s | 36.71 GPixel/s |

| Texture Rate | 157.1 GTexel/s | 146.8 GTexel/s |

| FP32 Performance | 5.027 TFLOPS | 3.524 TFLOPS |

| FP16 Performance | 10.05 TFLOPS (2:1) | null |

| TDP | 120 W | 225 W |

| Power Connectors | 1x 8-pin | 1x 6-pin + 1x 8-pin |

| Suggested PSU | 300 W | 550 W |

| Bus Interface | PCIe 3.0 x16 | PCIe 2.0 x16 |

| Display Outputs | 1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a | No outputs |

| DirectX Support | 12 (12_1) | 12 (11_0) |

| Vulkan Support | 1.4 | 1.2.175 |

| Slot Width | Dual-slot | Dual-slot |

| Length | 229 mm (9 inches) | 267 mm (10.5 inches) |

| Height | 111 mm (4.4 inches) | null |

| Width | 35 mm (1.4 inches) | null |

| Release Date | 2019-03-13 | 2012-11-11 |

| Launch MSRP | 219 USD | 3,199 USD |