NVIDIA GeForce GTX 660 vs NVIDIA GeForce GTX 960M Comparison

The NVIDIA GeForce GTX 960M and the NVIDIA GeForce GTX 660 represent two distinct approaches to GPU design from the same manufacturer, separated by roughly two and a half years of architectural evolution. The data shows a clear performance hierarchy, with the older desktop-oriented GTX 660 consistently outperforming the newer mobile-focused GTX 960M in the available benchmark tests. This analysis examines the head-to-head results, architectural differences, and the specific use cases where each card holds an advantage.

Head-to-Head Benchmarks

The benchmark results are unambiguous. In the Geekbench OpenCL test, the GTX 660 scores 11,347 while the GTX 960M scores 11,045. This gives the GTX 660 a 2.7% lead. While this is a relatively modest margin, it is consistent across the board. The gap widens dramatically in the Vulkan test, where the GTX 660 scores 11,415 against the GTX 960M’s 8,245. That is a 27.8% advantage for the older card, a substantial difference that highlights the GTX 660’s superior raw throughput in this API.

The win count reflects this dominance: the GTX 660 wins 2 out of 2 head-to-head benchmarks, while the GTX 960M wins none. The average benchmark scores further corroborate the trend, with the GTX 660 averaging 9,022 compared to the GTX 960M’s 9,645. Interestingly, despite losing the head-to-head, the GTX 960M has a higher average score because it only has two benchmark results, both of which are relatively strong. The GTX 660’s average is pulled down by its additional Metal benchmark score of 4,305.

Looking at the percentile ranks, the two cards are nearly identical in overall standing. The GTX 960M sits at the 46th percentile of all GPUs, while the GTX 660 sits at the 45th. This suggests that while the GTX 660 wins the direct comparisons, both cards occupy a similar tier in the broader GPU landscape. The nearest rivals for the GTX 960M include the NVIDIA Quadro K5000 (avg score 9,637, 0.1% behind) and the AMD Radeon Pro WX 2100 (avg score 9,653, 0.1% ahead), indicating it is tightly packed with professional-grade cards. The GTX 660’s rivals include the NVIDIA GeForce GTX 560 (avg score 9,058, 0.4% behind) and the AMD Radeon 550X (avg score 8,918, 1.2% ahead), showing it also competes in a similar bandwidth of performance.

Architecture Differences

The fundamental difference lies in the underlying chip design. The GTX 960M uses the GM107 chip based on the Maxwell architecture, while the GTX 660 uses the GK106 chip based on the Kepler architecture. Both are manufactured on a 28 nm process at TSMC, but the transistor counts and die sizes diverge significantly. The GTX 660 packs 2,540 million transistors into a 221 mm² die, while the GTX 960M fits 1,870 million transistors into a much smaller 148 mm² die. This results in a transistor density of 11.5M per mm² for the GTX 660 and 12.6M per mm² for the GTX 960M, showing the newer Maxwell design is more efficient at packing transistors.

The compute resources tell a clear story. The GTX 660 has 960 shading units, 80 texture mapping units (TMUs), and 24 raster operation units (ROPs). The GTX 960M has only 640 shading units, 40 TMUs, and 16 ROPs. This means the GTX 660 has 50% more shading units, double the TMUs, and 50% more ROPs. Consequently, the GTX 660 achieves a pixel rate of 20.64 GPixel/s and a texture rate of 82.56 GTexel/s, while the GTX 960M manages 18.82 GPixel/s and 47.04 GTexel/s. The FP32 throughput is also higher on the GTX 660 at 1.981 TFLOPS versus 1.505 TFLOPS for the GTX 960M.

Clock speeds are the one area where the newer card leads. The GTX 960M has a base clock of 1097 MHz and a boost clock of 1176 MHz. The GTX 660 has a lower base clock of 980 MHz and a boost clock of 1032 MHz. This clock advantage is not enough to overcome the GTX 660’s massive lead in core count and memory bandwidth. The GTX 660 offers 144.2 GB/s of bandwidth thanks to a 192-bit bus and 6 Gbps effective memory speed, while the GTX 960M is limited to 80.19 GB/s with a 128-bit bus and 5 Gbps effective speed. The GTX 660 also has a larger 2 GB memory allocation compared to the GTX 960M’s 4 GB, but the 4 GB is paired with a narrower bus.

The Verdict

The data is clear: the NVIDIA GeForce GTX 660 is the stronger performer in every direct benchmark comparison. Its 27.8% lead in Vulkan and 2.7% lead in OpenCL are decisive. The GTX 960M’s architectural advantages—higher clocks, a smaller die, and more efficient transistor density—do not translate into higher frame rates. The GTX 660’s 50% more shading units and double the TMUs are simply too large a deficit for the GTX 960M to overcome.

For users prioritizing raw performance in compute and API-specific workloads, the GTX 660 is the obvious choice. Its higher pixel rate, texture rate, and FP32 throughput make it better suited for tasks that leverage these capabilities. The GTX 960M, however, is not without merit. Its 4 GB memory buffer is double that of the GTX 660, which can be beneficial for workloads that are memory-capacity-bound rather than bandwidth-bound. Additionally, its dramatically lower TDP of 75 watts versus 140 watts makes it a far more power-efficient option.

The GTX 960M also boasts a more modern feature set in terms of API support. It supports Vulkan 1.4, while the GTX 660 is limited to Vulkan 1.2.175. Both support DirectX 12 (11_0) and OpenGL 4.6. This means the GTX 960M has better future-proofing for newer Vulkan titles, even if it currently loses in that benchmark. The choice ultimately depends on whether the user needs maximum performance today (GTX 660) or lower power consumption and a larger memory pool (GTX 960M).

Specification Differences

The two cards differ in nearly every specification. The GTX 660 uses the GK106 chip with 2,540 million transistors on a 221 mm² die, while the GTX 960M uses the GM107 chip with 1,870 million transistors on a 148 mm² die. The GTX 660 has 960 shading units, 80 TMUs, and 24 ROPs, while the GTX 960M has 640 shading units, 40 TMUs, and 16 ROPs. Clock speeds favor the GTX 960M with a base of 1097 MHz and boost of 1176 MHz, versus 980 MHz and 1032 MHz for the GTX 660.

Memory configurations are also starkly different. The GTX 960M has 4 GB of GDDR5 on a 128-bit bus with 80.19 GB/s bandwidth and 5 Gbps effective speed. The GTX 660 has 2 GB of GDDR5 on a 192-bit bus with 144.2 GB/s bandwidth and 6 Gbps effective speed. The GTX 660 achieves higher pixel and texture rates at 20.64 GPixel/s and 82.56 GTexel/s, respectively, versus 18.82 GPixel/s and 47.04 GTexel/s for the GTX 960M. FP32 performance is 1.981 TFLOPS for the GTX 660 and 1.505 TFLOPS for the GTX 960M.

Power and physical specifications are equally divergent. The GTX 660 has a TDP of 140 W and requires a 6-pin power connector, while the GTX 960M has a TDP of 75 W and requires no power connectors. The GTX 660 is a dual-slot card measuring 241 mm (9.5 inches) in length and uses a PCIe 3.0 x16 interface. The GTX 960M is an MXM module using an MXM-B (3.0) interface. Display outputs also differ: the GTX 660 offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the GTX 960M’s outputs are listed as portable device dependent. The GTX 660 has a launch MSRP of 229 USD.

FAQ

Q: Which GPU wins in the Geekbench Vulkan test?

A: The NVIDIA GeForce GTX 660 wins with a score of 11,415, compared to the GTX 960M’s 8,245. This represents a 27.8% advantage for the GTX 660.

Q: How much faster is the GTX 660 in OpenCL?

A: The GTX 660 scores 11,347 in Geekbench OpenCL, while the GTX 960M scores 11,045. The GTX 660 is 2.7% faster in this test.

Q: Does the GTX 960M have more memory than the GTX 660?

A: Yes, the GTX 960M has 4 GB of GDDR5 memory, while the GTX 660 has 2 GB. However, the GTX 660 has a wider 192-bit memory bus and higher bandwidth of 144.2 GB/s versus 80.19 GB/s for the GTX 960M.

Q: What is the TDP difference between the two cards?

A: The GTX 960M has a TDP of 75 W, while the GTX 660 has a TDP of 140 W. The GTX 960M is significantly more power-efficient.

Q: Which card has higher FP32 performance?

A: The GTX 660 has higher FP32 performance at 1.981 TFLOPS, compared to the GTX 960M’s 1.505 TFLOPS.

Q: What Vulkan versions do the cards support?

A: The GTX 960M supports Vulkan 1.4, while the GTX 660 supports Vulkan 1.2.175. Both support DirectX 12 (11_0) and OpenGL 4.6.

Where Each One Wins

The NVIDIA GeForce GTX 660 wins in scenarios demanding raw compute performance. Its 50% more shading units, double the TMUs, and higher texture rate make it the better choice for applications that are heavily shader-bound or texture-bound. The 2.7% OpenCL lead and 27.8% Vulkan lead indicate it excels in general-purpose GPU computing and modern low-level graphics APIs. Its higher pixel rate also suggests it handles fill-rate-intensive tasks better. The GTX 660 is the pick for users who need maximum throughput and are willing to accept a 140 W TDP and a 6-pin power requirement.

The NVIDIA GeForce GTX 960M wins in power-sensitive and capacity-oriented environments. Its 75 W TDP, which is nearly half that of the GTX 660, makes it suitable for compact or mobile systems where thermal and power budgets are tight. The 4 GB memory buffer, double that of the GTX 660, provides more headroom for workloads that store large datasets in VRAM, even if the bandwidth is lower. Its support for Vulkan 1.4 also gives it a feature advantage for future software. The GTX 960M is the logical choice for users who prioritize efficiency and memory capacity over peak compute performance.