NVIDIA GeForce GTX 1660 vs NVIDIA GeForce GTX 960M Comparison

The NVIDIA GeForce GTX 1660 and the NVIDIA GeForce GTX 960M sit four years apart in the database, and the recorded data shows a clear generational gulf between a desktop Turing card and a mobile Maxwell part. The GTX 1660 is built on TSMC's 12 nm process with the TU116 chip, while the GTX 960M uses the 28 nm GM107 and ships as an MXM module rather than a desktop card. Across the two head-to-head benchmark results in the database, the GTX 1660 wins both, and by margins that make this one of the more lopsided matchups in the records.

Head-to-Head Benchmarks

Only two tests appear in both GPUs' result sets, but both are decisive. In Geekbench OpenCL, the GTX 1660 scores 47,850 against the GTX 960M's 11,045, a 333.2 percent advantage for the desktop card. Geekbench Vulkan widens that gap further: 50,137 versus 8,245, a 508.1 percent margin. The GTX 1660 takes two wins from two comparisons, and the GTX 960M takes none.

To put those margins in context, consider each card's position against its own nearest rivals. The GTX 960M's average benchmark score of 9,645 places it in the 46th percentile of all GPUs in the database, effectively neck-and-neck with the NVIDIA Quadro K5000 (0.1 percent apart), the AMD Radeon Pro WX 2100 (0.1 percent apart), the NVIDIA Quadro P4000 (0.2 percent apart), and the NVIDIA Tesla C2070 (0.7 percent apart). In other words, the GTX 960M performs like a professional workstation card of an earlier era. The GTX 1660's average score of 11,680 puts it in the 51st percentile, clustered with the AMD Radeon RX 7800 XT (0.5 percent), the AMD Radeon Pro 5500M (1.3 percent), the AMD Radeon RX 6500 XT (1.4 percent), and the NVIDIA Tesla K20c (1.8 percent). That is a higher tier of the database, though percentile-wise the two are separated by five points, which reflects how the averaging across mixed test suites compresses the raw gap seen in the direct comparisons.

The direct head-to-head numbers tell the real story. A 508.1 percent difference in Geekbench Vulkan means the GTX 1660 delivers more than six times the GTX 960M's score in that workload. Even in OpenCL, where the GTX 960M posts a comparatively stronger relative result, it still trails by a factor of more than four. The GTX 960M records no wins anywhere in the shared data.

The GTX 1660's broader result set reinforces its standing: a PassMark G3D score of 11,646, a PassMark GPU Compute score of 4,963, a 3DMark Steel Nomad DX12 score of 1,065, and PassMark DirectX results of 177 (DX9), 61 (DX10), 79 (DX11), and 49 (DX12), plus a PassMark G2D score of 776. None of these tests have recorded GTX 960M counterparts in the database, so they serve as context for the 1660's capabilities rather than direct comparisons.

FAQ

Q: How much faster is the GTX 1660 than the GTX 960M in shared benchmarks?

A: The GTX 1660 wins both recorded head-to-head tests: 333.2 percent faster in Geekbench OpenCL and 508.1 percent faster in Geekbench Vulkan.

Q: Which GPU has the higher average benchmark score?

A: The GTX 1660, at 11,680 versus 9,645 for the GTX 960M. The GTX 1660 also ranks in the 51st percentile of all GPUs in the database, against the 960M's 46th percentile.

Q: Which cards are the closest rivals to each GPU?

A: The GTX 1660's nearest rivals are the AMD Radeon RX 7800 XT, AMD Radeon Pro 5500M, AMD Radeon RX 6500 XT, and NVIDIA Tesla K20c. The GTX 960M's are the NVIDIA Quadro K5000, AMD Radeon Pro WX 2100, NVIDIA Quadro P4000, and NVIDIA Tesla C2070.

Q: Do either of these GPUs have ray tracing or AI acceleration hardware?

A: No. The database lists no RT cores and no tensor cores for either card, despite the GTX 1660 using the Turing architecture.

Q: How do their theoretical compute figures compare?

A: The GTX 1660 delivers 5.027 TFLOPS of FP32 performance against 1.505 TFLOPS for the GTX 960M. The GTX 1660 also records 10.05 TFLOPS of FP16 at a 2:1 ratio, while the 960M lists no FP16 figure.

Q: Are both GPUs still in production?

A: No, both are marked end-of-life in the database. The GTX 1660 launched on March 13, 2019, and the GTX 960M on March 12, 2015.

Architecture Differences

The architectural distance between these two is substantial. The GTX 1660 uses the TU116 chip on NVIDIA's Turing architecture, manufactured by TSMC on a 12 nm process, with 6,600 million transistors packed into a 284 mm² die for a density of 23.2M per mm². The GTX 960M uses the GM107 chip on the Maxwell architecture, also fabbed by TSMC but on a 28 nm process, with 1,870 million transistors on a 148 mm² die and a density of 12.6M per mm². The transistor count is roughly three and a half times higher on the Turing part, and the die is nearly twice the area, but density has also nearly doubled between generations.

The core configuration scales accordingly. The GTX 1660 carries 1,408 shading units, 88 texture mapping units, and 48 render output units, against 640 shading units, 40 TMUs, and 16 ROPs on the GTX 960M. Clocks also favor the newer card: 1530 MHz base and 1785 MHz boost for the 1660, versus 1097 MHz base and 1176 MHz boost for the 960M. The resulting throughput figures are dramatic. Pixel fill rate is 85.68 GPixel/s against 18.82 GPixel/s, texture rate is 157.1 GTexel/s against 47.04 GTexel/s, and FP32 compute is 5.027 TFLOPS against 1.505 TFLOPS, a gap of more than three times that closely mirrors the OpenCL benchmark result.

Memory is another dividing line. Both use GDDR5, but the GTX 1660 has 6 GB on a 192-bit bus running at an effective 8 Gbps, yielding 192.1 GB/s of bandwidth. The GTX 960M has 4 GB on a 128-bit bus at an effective 5 Gbps, yielding 80.19 GB/s. The bandwidth advantage, like the compute advantage, sits at well over double for the desktop card.

Neither GPU has dedicated RT cores or tensor cores, so hardware ray tracing and AI acceleration are absent from both feature sets. On the API side they are surprisingly close on paper: both support OpenGL 4.6 and Vulkan 1.4, and both report DirectX 12, though the 1660's feature level is 12_1 versus 11_0 for the 960M.

Form factor is the final architectural distinction. The GTX 1660 is a dual-slot desktop card with a PCIe 3.0 x16 interface, an 8-pin power connector, a 120 W TDP, and a suggested PSU of 300 W. The GTX 960M is an MXM module with an MXM-B (3.0) bus, no external power connectors, and a 75 W TDP, drawing everything it needs from the host laptop. Its display outputs are portable-device dependent, whereas the 1660 provides one DVI, one HDMI 2.0, and one DisplayPort 1.4a.

The Verdict

The data supports exactly one conclusion: the GTX 1660 is the stronger GPU in every recorded comparison. It wins both shared benchmarks, by 333.2 percent in Geekbench OpenCL and 508.1 percent in Geekbench Vulkan, holds the higher average score, the higher percentile ranking, and the stronger rival cluster, and it delivers more than triple the theoretical FP32 compute and more than double the memory bandwidth. Users choosing purely on performance have no reason to prefer the GTX 960M.

The 960M's remaining relevance is qualitative rather than numerical. It is an MXM module designed for laptops, with no power connectors and a lower TDP, so it exists in a category the desktop-oriented 1660 cannot occupy. Both are end-of-life parts. The GTX 1660's launch MSRP was 219 USD. The GTX 960M's value proposition rests entirely on platform compatibility: it slots into machines that accept MXM-B (3.0) modules, and its nearest-rival cohort (Quadro K5000, Radeon Pro WX 2100, Quadro P4000, Tesla C2070) indicates it performs at the level of older professional hardware. For anyone comparing raw graphics capability, the database records no scenario in which the GTX 960M comes out ahead.

Specification Differences

• Architecture: Turing (GTX 1660) vs Maxwell (GTX 960M)
• Chip: TU116 vs GM107
• Process node: 12 nm vs 28 nm (both TSMC)
• Transistors: 6,600 million vs 1,870 million
• Die size: 284 mm² vs 148 mm²
• Transistor density: 23.2M / mm² vs 12.6M / mm²
• Shading units: 1,408 vs 640
• TMUs: 88 vs 40
• ROPs: 48 vs 16
• Base clock: 1530 MHz vs 1097 MHz
• Boost clock: 1785 MHz vs 1176 MHz
• Memory size: 6 GB vs 4 GB (both GDDR5)
• Bus width: 192 bit vs 128 bit
• Memory speed: 8 Gbps effective vs 5 Gbps effective
• Bandwidth: 192.1 GB/s vs 80.19 GB/s
• FP32 compute: 5.027 TFLOPS vs 1.505 TFLOPS
• FP16 compute: 10.05 TFLOPS (2:1) vs not listed
• Pixel rate: 85.68 GPixel/s vs 18.82 GPixel/s
• Texture rate: 157.1 GTexel/s vs 47.04 GTexel/s
• TDP: 120 W vs 75 W
• Form factor: Dual-slot desktop card vs MXM module
• Bus interface: PCIe 3.0 x16 vs MXM-B (3.0)
• Power connectors: 1x 8-pin vs none
• Suggested PSU: 300 W vs not specified
• Display outputs: 1x DVI, 1x HDMI 2.0, 1x DisplayPort 1.4a vs portable device dependent
• DirectX feature level: 12_1 vs 11_0
• Dimensions: 229 mm x 111 mm x 35 mm vs not applicable
• Release date: March 13, 2019 vs March 12, 2015
• Average benchmark score: 11,680 vs 9,645
• Percentile vs all GPUs: 51st vs 46th
• Launch MSRP: 219 USD vs not listed