NVIDIA GeForce GTX 1660 Ti vs NVIDIA GeForce GTX 675MX Comparison

# The Verdict

The benchmark data presents a stark generational divide. The NVIDIA GeForce GTX 1660 Ti dominates the GeForce GTX 675MX in the sole shared benchmark, with a geekbench_opencl score of 51,029 versus 10,723 — a 79% lead for the newer card. This is not a close contest; it is a decisive victory for the Turing-based part. The data suggests that anyone considering these two GPUs for modern workloads should unequivocally choose the GTX 1660 Ti, provided their system can accommodate its dual-slot footprint and 1x 8-pin power connector. The GTX 675MX, while still functional, shows its age with an average benchmark score of 8,427, sitting at the 43rd percentile of all GPUs. The GTX 1660 Ti, despite its lower average score of 7,723 (41st percentile), achieves far higher peak performance in compute tasks, indicating that its average is dragged down by legacy DirectX 9 and 2D tests where older architectures can still compete. For gaming, content creation, or any compute-heavy application, the GTX 1660 Ti is the only rational choice from a pure performance standpoint.

# Architecture Differences

The architectural gulf between these two NVIDIA parts is vast, spanning two distinct design philosophies. The GTX 675MX is built on the GK104 chip using the Kepler architecture, manufactured on a 28 nm process at TSMC. It packs 3,540 million transistors into a 294 mm² die, yielding a transistor density of 12.0 million per square millimeter. Its memory subsystem relies on 2 GB of GDDR5 across a 256-bit bus, delivering 115.2 GB/s of bandwidth at an effective 3.6 Gbps. The shading unit count stands at 960, accompanied by 80 texture mapping units and 32 raster operation pipelines.

In contrast, the GTX 1660 Ti utilizes the TU116 chip under the Turing architecture, fabricated on a more advanced 12 nm process at the same foundry. This newer design crams 6,600 million transistors into a slightly smaller 284 mm² die, achieving a much higher density of 23.2 million per square millimeter. Memory is upgraded to 6 GB of GDDR6 on a narrower 192-bit bus, yet bandwidth jumps to 288.0 GB/s thanks to a 12 Gbps effective data rate. The compute core expands to 1,536 shading units, 96 TMUs, and 48 ROPs. Clock speeds also differ significantly: the GTX 1660 Ti has a base clock of 1500 MHz and a boost clock of 1770 MHz, while the GTX 675MX's base and boost clocks are not specified in the data — only its memory clock of 900 MHz (3.6 Gbps effective). The GTX 1660 Ti also supports a higher DirectX feature level (12_1 versus 11_0) and a newer Vulkan API version (1.4 versus 1.2.175). Neither card features ray tracing or tensor cores, so that aspect is equal.

# Head-to-Head Benchmarks

The only direct comparison available in the data is the geekbench_opencl test, and it is a landslide. The GTX 1660 Ti scores 51,029, while the GTX 675MX manages just 10,723. This represents a deltaPct of -79 from the older card's perspective, meaning the GTX 1660 Ti is roughly four times faster in this OpenCL compute workload. This massive gap can be attributed to several compounding factors: the GTX 1660 Ti's higher shading unit count (1,536 versus 960), its significantly faster memory (288 GB/s versus 115.2 GB/s), and its much higher pixel and texture rates. The GTX 1660 Ti delivers 84.96 GPixel/s and 169.9 GTexel/s, while the GTX 675MX trails at 13.08 GPixel/s and 52.32 GTexel/s. Floating-point performance tells a similar story: the GTX 1660 Ti reaches 5.437 TFLOPS FP32 (and 10.87 TFLOPS FP16 at 2:1 ratio), whereas the GTX 675MX peaks at 1,255.7 GFLOPS FP32 — a 4.3x disparity. In every measurable metric that feeds into compute performance, the Turing card wins by a wide margin.

While the GTX 1660 Ti has additional benchmark scores (PassMark tests, 3DMark, Vulkan), the GTX 675MX has no corresponding data for those tests, so cross-referencing is impossible. The wins tally confirms the direction: the GTX 1660 Ti claims 1 win, the GTX 675MX claims 0.

# FAQ

Q: Which card has a higher average benchmark score?

A: The GTX 675MX actually has a higher average benchmark score at 8,427, compared to the GTX 1660 Ti's 7,723. However, this average includes different test suites, and the sole shared test (geekbench_opencl) heavily favors the GTX 1660 Ti.

Q: How much faster is the GTX 1660 Ti in OpenCL compute?

A: In the geekbench_opencl test, the GTX 1660 Ti scores 51,029 versus 10,723 for the GTX 675MX, a deltaPct of -79 from the older card's perspective — meaning the GTX 1660 Ti is roughly 4.8 times faster.

Q: What is the difference in memory bandwidth?

A: The GTX 1660 Ti offers 288.0 GB/s of bandwidth from its 6 GB GDDR6 memory on a 192-bit bus, while the GTX 675MX provides 115.2 GB/s from 2 GB GDDR5 on a 256-bit bus. The newer card has 2.5x the bandwidth.

Q: Do both cards support the same DirectX version?

A: No. The GTX 1660 Ti supports DirectX 12 (12_1), while the GTX 675MX supports DirectX 12 (11_0) — a lower feature level. Both support OpenGL 4.6, but Vulkan support differs (1.4 versus 1.2.175).

Q: What are the power and physical requirements?

A: The GTX 1660 Ti has a TDP of 120 W, requires a 300 W suggested PSU, uses a dual-slot design with a 1x 8-pin power connector, and measures 229 mm in length. The GTX 675MX has a 100 W TDP, uses an MXM module form factor with no additional power connectors, and its physical dimensions are portable-device dependent.

Q: Which card was released later?

A: The GTX 1660 Ti was released on 2019-02-21, while the GTX 675MX dates back to 2012-09-30 — a gap of over six years.

# Where Each One Wins

Based strictly on the data, the GTX 1660 Ti wins in every category that matters for modern workloads. It dominates OpenCL compute, has more than double the memory capacity (6 GB versus 2 GB), and offers significantly higher pixel rate (84.96 GPixel/s versus 13.08 GPixel/s), texture rate (169.9 GTexel/s versus 52.32 GTexel/s), and FP32 throughput (5.437 TFLOPS versus 1,255.7 GFLOPS). It also supports a newer DirectX feature level and Vulkan version, making it better suited for current and future titles that leverage these APIs.

The GTX 675MX, however, has a few niche advantages. Its lower TDP (100 W versus 120 W) and lack of external power connectors make it more suitable for compact or legacy systems where power delivery is constrained. Its MXM-B (3.0) form factor is designed for laptop upgrades, whereas the GTX 1660 Ti's dual-slot PCIe 3.0 x16 design targets desktop towers. The older card's higher average benchmark score (8,427 versus 7,723) and higher percentile ranking (43rd versus 41st) suggest it may perform relatively better in specific legacy or 2D workloads, though the lack of overlapping tests makes this speculative. For anyone building a new system or upgrading a desktop, the GTX 1660 Ti is the clear winner. For those stuck with an MXM-based laptop, the GTX 675MX might be the only compatible option.

# Specification Differences

| Specification | GTX 675MX | GTX 1660 Ti |

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

| Chip | GK104 | TU116 |

| Architecture | Kepler | Turing |

| Generation | GeForce 600M | GeForce 16 |

| Process Node | 28 nm | 12 nm |

| Transistors | 3,540 million | 6,600 million |

| Die Size | 294 mm² | 284 mm² |

| Transistor Density | 12.0M / mm² | 23.2M / mm² |

| Base Clock | Not specified | 1500 MHz |

| Boost Clock | Not specified | 1770 MHz |

| Memory Clock | 900 MHz (3.6 Gbps effective) | 1500 MHz (12 Gbps effective) |

| Memory Size | 2 GB | 6 GB |

| Memory Type | GDDR5 | GDDR6 |

| Memory Bus Width | 256 bit | 192 bit |

| Memory Bandwidth | 115.2 GB/s | 288.0 GB/s |

| Shading Units | 960 | 1536 |

| TMUs | 80 | 96 |

| ROPs | 32 | 48 |

| Pixel Rate | 13.08 GPixel/s | 84.96 GPixel/s |

| Texture Rate | 52.32 GTexel/s | 169.9 GTexel/s |

| FP32 Performance | 1,255.7 GFLOPS | 5.437 TFLOPS |

| FP16 Performance | Not specified | 10.87 TFLOPS (2:1) |

| TDP | 100 W | 120 W |

| Slot Width | MXM Module | Dual-slot |

| Power Connectors | None | 1x 8-pin |

| Suggested PSU | Not specified | 300 W |

| Bus Interface | MXM-B (3.0) | PCIe 3.0 x16 |

| Display Outputs | Portable Device Dependent | 1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a |

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

| OpenGL | 4.6 | 4.6 |

| Vulkan | 1.2.175 | 1.4 |

| Release Date | 2012-09-30 | 2019-02-21 |

| Predecessor | GeForce 500M | GeForce 10 |

| Successor | GeForce 700M | GeForce 20 |

| Launch MSRP | Not specified | 279 USD (stated once) |