NVIDIA GeForce GTX 1660 Ti vs NVIDIA GeForce GTX 670M Comparison

The GeForce GTX 1660 Ti and the GeForce GTX 670M are separated by seven years of GPU architecture, and the benchmark data reflects a generational chasm. The GTX 1660 Ti is a desktop Turing-based card, while the GTX 670M is a mobile Fermi-era part. The database shows a single shared benchmark result, and it is not close.

Head-to-Head Benchmarks

The only overlapping test in the database is Geekbench OpenCL, and the results are lopsided. The GeForce GTX 1660 Ti scores 51029, while the GeForce GTX 670M scores 6513. This gives the GTX 1660 Ti a 683.5% advantage, a massive lead that dwarfs any other comparison in the database. For context, the GTX 670M's score of 6513 places it just 0.3% ahead of the NVIDIA GeForce GT 555M and 0.5% ahead of the NVIDIA Quadro M5000M, while being 0.6% behind the Intel UHD Graphics P750. The GTX 1660 Ti, by contrast, sits at the 41st percentile of all GPUs, while the GTX 670M sits at the 38th. The raw score difference is not a marginal edge; it is a complete overhaul in compute capability.

Architecture Differences

The two cards come from entirely different eras of NVIDIA's design. The GTX 1660 Ti uses the TU116 chip on a 12 nm TSMC process, packing 6,600 million transistors onto a 284 mm² die. This yields a transistor density of 23.2 million per square millimeter. The GTX 670M uses the GF114 chip on a 40 nm TSMC process, with 1,950 million transistors spread across a larger 332 mm² die, resulting in a density of just 5.9 million per square millimeter. The 1660 Ti is built on the Turing architecture, while the 670M is based on Fermi 2.0. The 1660 Ti has no ray tracing cores or tensor cores, same as the 670M, so the architectural leap is in raw throughput and efficiency rather than new fixed-function hardware.

The compute resources are starkly different. The GTX 1660 Ti has 1536 shading units, 96 texture mapping units, and 48 ROPs. The GTX 670M has 336 shading units, 56 TMUs, and 24 ROPs. The 1660 Ti's pixel rate is 84.96 GPixel/s versus 8.372 GPixel/s for the 670M, and its texture rate is 169.9 GTexel/s versus 33.49 GTexel/s. FP32 performance is 5.437 TFLOPS for the 1660 Ti versus 803.7 GFLOPS for the 670M. The 1660 Ti also supports FP16 at 10.87 TFLOPS (2:1), a feature the 670M lacks entirely. The 670M has no base or boost clock listed in the database, while the 1660 Ti runs at 1500 MHz base and 1770 MHz boost. Memory clocks differ as well: the 1660 Ti uses 1500 MHz (12 Gbps effective) GDDR6, while the 670M uses 750 MHz (3 Gbps effective) GDDR5.

FAQ

Q: Which GPU has higher memory bandwidth?

A: The GTX 1660 Ti has 288.0 GB/s bandwidth, which is four times the 72.00 GB/s of the GTX 670M, despite both using a 192-bit bus.

Q: Can the GTX 670M run the same APIs as the GTX 1660 Ti?

A: Both support DirectX 12 and OpenGL 4.6, but the 1660 Ti supports DirectX 12_1, while the 670M supports 12 (11_0). The 1660 Ti also supports Vulkan 1.4, while the 670M has no Vulkan support listed.

Q: What is the power draw difference?

A: The GTX 1660 Ti is rated at 120 W TDP, while the GTX 670M is rated at 75 W TDP. The 1660 Ti requires a 300 W suggested PSU and a single 8-pin connector, while the 670M has no power connectors listed.

Q: How do their average benchmark scores compare?

A: The GTX 1660 Ti has an average benchmark score of 7723, while the GTX 670M has an average score of 6513. The 1660 Ti is 0.7% ahead of the AMD Radeon 540 in the database's nearest rival list, while the 670M is 0.3% ahead of the GT 555M.

Q: Which card has more memory?

A: The GTX 1660 Ti has 6 GB of GDDR6, while the GTX 670M has 1536 MB (1.5 GB) of GDDR5.

Q: Are both cards still in production?

A: No, both are marked as end-of-life in the database. The 1660 Ti was released on 2019-02-21, and the 670M was released on 2012-03-21.

The Verdict

The data is unambiguous. The GTX 1660 Ti outperforms the GTX 670M by 683.5% in the one shared benchmark, and its architecture, memory subsystem, and compute resources are all in a different class. The 670M is a legacy mobile part from 2012, and its 38th percentile ranking places it near the bottom of modern GPU performance. The 1660 Ti, despite being end-of-life, sits at the 41st percentile and has enough compute headroom for modern workloads, as shown by its 12871 Passmark G3D score and 5226 GPU compute score. The 670M's only listed benchmark is Geekbench OpenCL at 6513, which is lower than the 1660 Ti's OpenCL score of 51029 by a factor of nearly eight.

For a builder, the choice is obvious if both cards are on the table: the 1660 Ti is the only viable option for any current gaming or compute task. The 670M is not competitive in any database-recorded metric. The 1660 Ti also has a launch MSRP of 279 USD, which the 670M does not have listed, but the performance gap justifies the difference regardless of price.

Specification Differences

The two cards differ in nearly every specification field. The 1660 Ti uses a 12 nm process, the 670M uses 40 nm. Transistor count is 6,600 million versus 1,950 million. Die size is 284 mm² versus 332 mm². The 1660 Ti has 1536 shading units, 96 TMUs, and 48 ROPs; the 670M has 336 shading units, 56 TMUs, and 24 ROPs. Memory is 6 GB GDDR6 versus 1536 MB GDDR5, with bandwidth of 288.0 GB/s versus 72.00 GB/s. The 1660 Ti has a base clock of 1500 MHz and boost of 1770 MHz; the 670M has no clock listed. Pixel rate is 84.96 GPixel/s versus 8.372 GPixel/s. Texture rate is 169.9 GTexel/s versus 33.49 GTexel/s. FP32 is 5.437 TFLOPS versus 803.7 GFLOPS. FP16 is 10.87 TFLOPS (2:1) on the 1660 Ti, absent on the 670M. TDP is 120 W versus 75 W. The 1660 Ti is dual-slot with a 1x 8-pin connector and 300 W suggested PSU; the 670M is an MXM Module with no connectors. The bus interface is PCIe 3.0 x16 versus MXM-B (3.0). Display outputs are 1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a on the 1660 Ti, versus portable-device-dependent on the 670M. DirectX support is 12 (12_1) versus 12 (11_0), and Vulkan is 1.4 versus none.

Where Each One Wins

The GTX 1660 Ti wins every recorded benchmark where both have data. It wins in raw compute, memory bandwidth, pixel throughput, and texture throughput. It is the clear choice for desktop gaming, rendering, or any GPU-accelerated workload that leverages OpenCL or modern APIs. Its 12871 Passmark G3D score and 804 G2D score indicate strong performance across both 3D and 2D tasks. The GTX 670M has no wins in the head-to-head data. Its only advantage is a lower TDP of 75 W, which matters only for legacy laptop upgrades where power and cooling are constrained, but even then its performance is severely limited by 72 GB/s bandwidth and 336 shading units. The database shows no scenario where the 670M leads, so any use case that favors it would be based on hardware compatibility with older MXM slots, not performance. For anyone with a modern PCIe slot and a 300 W PSU, the 1660 Ti is the only rational pick from this pair.