NVIDIA GeForce GTX 660 Ti vs NVIDIA GeForce GTX 965M Comparison

The data shows a close contest between two end-of-life NVIDIA mobile and desktop parts. The NVIDIA GeForce GTX 660 Ti, a 2012 desktop card, consistently outperforms the NVIDIA GeForce GTX 965M, a 2015 laptop chip, in the available benchmarks. The GTX 660 Ti wins both head-to-head tests, but the margins are slim, making the 965M a respectable performer given its mobile constraints.

The Verdict

From the benchmark results, the NVIDIA GeForce GTX 660 Ti is the clear winner, taking both head-to-head tests. However, the practical choice depends entirely on your use case. The GTX 660 Ti is a desktop card with a dual-slot footprint, requiring a 450 W power supply and two 6-pin connectors. It is not a drop-in replacement for a laptop. The GTX 965M, as an MXM module with no power connectors, is designed for portable systems.

If you are building or upgrading a desktop with a PCIe 3.0 x16 slot, the GTX 660 Ti offers a 4.2% lead in OpenCL and a 5% lead in Vulkan over the 965M. That advantage, while modest, comes with a much larger texture rate (109.8 GTexel/s vs 60.80 GTexel/s) and nearly double the FP32 throughput (2.634 TFLOPS vs 1.946 TFLOPS). The GTX 660 Ti also has a 192-bit memory bus, providing 144.2 GB/s of bandwidth compared to the 965M’s 128-bit bus and 80.19 GB/s.

For a laptop user, the GTX 965M’s performance is not far behind. Its average benchmark score of 14404 places it at the 56th percentile of all GPUs, just one point below the GTX 660 Ti’s 57th percentile. The 965M also supports DirectX 12 (12_1), while the 660 Ti only reaches DirectX 12 (11_0). The GTX 660 Ti is end-of-life, as is the 965M, but the 965M is the newer part by release date.

Strictly from the data, the GTX 660 Ti is the faster card. Choose it for a fixed desktop build where power and space are not concerns. Choose the GTX 965M if you need a mobile solution and can accept a small performance deficit.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The NVIDIA GeForce GTX 660 Ti has an average benchmark score of 15063, which is higher than the GTX 965M’s average of 14404.

Q: What is the performance difference in the Vulkan test?

A: The GTX 660 Ti scores 15012 in Geekbench Vulkan, beating the GTX 965M’s score of 14299 by a delta of 5%.

Q: Does the GTX 965M support a newer DirectX version?

A: Yes. The GTX 965M supports DirectX 12 (12_1), while the GTX 660 Ti supports DirectX 12 (11_0).

Q: What are the memory bandwidth figures for each card?

A: The GTX 660 Ti has a memory bandwidth of 144.2 GB/s, and the GTX 965M has a memory bandwidth of 80.19 GB/s.

Q: Which card has a higher transistor count?

A: The GTX 965M has 5,200 million transistors on a 398 mm² die, while the GTX 660 Ti has 3,540 million transistors on a 294 mm² die.

Q: How does the GTX 660 Ti compare to its nearest rival, the AMD Radeon Pro 560X?

A: The GTX 660 Ti’s average score of 15063 is 0.1% lower than the AMD Radeon Pro 560X’s average score of 15082.

Architecture Differences

The two GPUs come from different NVIDIA microarchitectures. The GTX 660 Ti is based on the Kepler architecture, using the GK104 chip. The GTX 965M is based on Maxwell 2.0, using the GM204 chip. Both are manufactured by TSMC on a 28 nm process node, but the transistor density differs slightly. The GTX 965M’s GM204 packs 13.1M transistors per mm², while the GTX 660 Ti’s GK104 has 12.0M per mm².

The GTX 965M’s chip is physically larger and more complex. It has 5,200 million transistors on a 398 mm² die, compared to the GTX 660 Ti’s 3,540 million transistors on a 294 mm² die. Despite having more transistors, the GTX 965M’s compute configuration is smaller: it has 1024 shading units and 64 texture mapping units (TMUs), versus the GTX 660 Ti’s 1344 shading units and 112 TMUs. However, the GTX 965M has more render output units (ROPs)—32 versus 24 on the GTX 660 Ti.

The memory subsystems also differ. The GTX 660 Ti uses a 192-bit bus, while the GTX 965M uses a 128-bit bus. Even though both have 2 GB of GDDR5 memory, the GTX 660 Ti’s wider bus gives it significantly higher bandwidth. The GTX 965M compensates with a higher pixel rate of 30.40 GPixel/s, beating the GTX 660 Ti’s 27.44 GPixel/s.

API support diverges, too. The GTX 965M supports Vulkan 1.4 and DirectX 12 (12_1). The GTX 660 Ti supports Vulkan 1.2.175 and DirectX 12 (11_0). Both support OpenGL 4.6. These differences reflect the newer architecture of the GTX 965M, despite its lower raw compute throughput.

Specification Differences

The most obvious difference is the form factor. The GTX 660 Ti is a dual-slot desktop card, 241 mm long, requiring two 6-pin power connectors and a 450 W power supply. The GTX 965M is an MXM module with no power connectors and no suggested PSU, designed for portable devices. The display outputs also differ: the GTX 660 Ti has 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the GTX 965M’s outputs are listed as "Portable Device Dependent."

Clock speeds are similar. The GTX 660 Ti has a base clock of 915 MHz and a boost clock of 980 MHz. The GTX 965M has a base clock of 924 MHz and a boost clock of 950 MHz. The memory clocks differ more: the GTX 660 Ti runs at 1502 MHz (6 Gbps effective), while the GTX 965M runs at 1253 MHz (5 Gbps effective).

The GTX 660 Ti has 1344 shading units, 112 TMUs, and 24 ROPs. The GTX 965M has 1024 shading units, 64 TMUs, and 32 ROPs. The result is that the GTX 660 Ti has a higher texture rate (109.8 GTexel/s vs 60.80 GTexel/s) and higher FP32 performance (2.634 TFLOPS vs 1.946 TFLOPS). The GTX 965M has a higher pixel rate (30.40 GPixel/s vs 27.44 GPixel/s). The GTX 660 Ti has a higher TDP of 150 W, while the GTX 965M does not list a TDP.

The GTX 660 Ti has a launch MSRP of 299 USD. The GTX 965M has no launch MSRP listed. The GTX 660 Ti was released on 2012-08-15, and the GTX 965M on 2015-01-08. Both are end-of-life.

Head-to-Head Benchmarks

The head-to-head data shows a narrow but consistent win for the GTX 660 Ti. In the Geekbench OpenCL test, the GTX 660 Ti scores 15113 against the GTX 965M’s 14509, a delta of 4.2%. In the Geekbench Vulkan test, the GTX 660 Ti scores 15012 against 14299, a delta of 5%.

These margins are small in absolute terms. The GTX 965M is not far behind, which is notable given that it is a mobile part. The GTX 660 Ti’s advantage is likely tied to its higher memory bandwidth (144.2 GB/s vs 80.19 GB/s) and greater shading unit count (1344 vs 1024). The GTX 965M’s higher pixel rate (30.40 GPixel/s vs 27.44 GPixel/s) does not translate into a benchmark win in these tests.

Looking at the nearest rivals provides context. The GTX 660 Ti’s average score of 15063 is just 0.1% below the AMD Radeon Pro 560X (15082) and 0.7% below the AMD Radeon RX 7600 (15171). The GTX 965M’s average score of 14404 is 0.1% above the AMD Radeon RX Vega 11 (14385) and 0.2% above the NVIDIA GeForce GTX TITAN (14373). This places both cards in a similar performance tier, with the GTX 660 Ti sitting slightly higher.

The GTX 660 Ti wins 2 head-to-head tests, and the GTX 965M wins 0. The data is unambiguous: the GTX 660 Ti is the faster card in both OpenCL and Vulkan workloads.

Where Each One Wins

NVIDIA GeForce GTX 660 Ti: This card wins in raw compute and memory throughput. It is ahead of the GTX 965M by 4.2% in OpenCL and 5% in Vulkan. It has a higher texture rate (109.8 GTexel/s vs 60.80 GTexel/s) and higher FP32 performance (2.634 TFLOPS vs 1.946 TFLOPS). Its memory bandwidth of 144.2 GB/s is nearly double the 965M’s 80.19 GB/s. For desktop users with a PCIe 3.0 x16 slot and a 450 W PSU, the GTX 660 Ti is the better performer. It is also a dual-slot card with dedicated display outputs (2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2), making it straightforward to connect to monitors.

NVIDIA GeForce GTX 965M: This card wins in portability and modern API support. It is an MXM module with no power connectors, suitable for laptops. It supports DirectX 12 (12_1) and Vulkan 1.4, while the GTX 660 Ti only supports DirectX 12 (11_0) and Vulkan 1.2.175. The GTX 965M also has a higher pixel rate (30.40 GPixel/s vs 27.44 GPixel/s) and more ROPs (32 vs 24), which can benefit certain rasterization workloads. Its average benchmark score of 14404 is within 4.6% of the GTX 660 Ti’s 15063, which is a small gap for a mobile part. If you need a GPU for a portable system and can accept a modest performance deficit, the GTX 965M is the only realistic choice between the two.