NVIDIA GeForce GTX 660 Ti vs NVIDIA GeForce RTX 2060 Comparison

The NVIDIA GeForce RTX 2060 and the NVIDIA GeForce GTX 660 Ti represent two distinct eras of GPU design, separated by a generational gap in both architecture and process technology. While the GTX 660 Ti was a capable performer in its day, the benchmark data provided shows a decisive and overwhelming performance advantage for the RTX 2060, making the comparison less about competition and more about illustrating technological progress.

FAQ

Q: How much faster is the RTX 2060 than the GTX 660 Ti in the available benchmarks?

A: The performance gap is enormous. In the Geekbench OpenCL test, the RTX 2060 scores 65,014 compared to the GTX 660 Ti's 15,113, which is a 330.2% difference. In Geekbench Vulkan, the RTX 2060's 65,846 score versus 15,012 represents a 338.6% lead.

Q: What is the average benchmark score for each card, and how do they rank globally?

A: The RTX 2060 has an average benchmark score of 15,290, placing it in the 58th percentile of all GPUs. The GTX 660 Ti has an average score of 15,063, putting it in the 57th percentile. Despite the massive difference in individual benchmark tests, their average scores are relatively close.

Q: Do the cards use different memory technologies?

A: Yes. The RTX 2060 uses 6 GB of GDDR6 memory, while the GTX 660 Ti uses 2 GB of GDDR5. The RTX 2060's memory bandwidth is 336.0 GB/s, more than double the GTX 660 Ti's 144.2 GB/s.

Q: Are there any architectural features present in the RTX 2060 that are not in the GTX 660 Ti?

A: Yes. The RTX 2060, based on the Turing architecture, includes 30 dedicated RT cores and 240 tensor cores. The GTX 660 Ti, based on the older Kepler architecture, has no RT cores or tensor cores listed.

Q: What is the difference in their process nodes and transistor counts?

A: The RTX 2060 is built on a 12 nm process at TSMC and contains 10,800 million transistors. The GTX 660 Ti uses a much older 28 nm process and has 3,540 million transistors. The newer node allows for a much higher transistor density of 24.3M / mm² versus 12.0M / mm².

Q: Which card has a higher launch MSRP?

A: The RTX 2060 had a launch MSRP of 349 USD, while the GTX 660 Ti had a launch MSRP of 299 USD.

Architecture Differences

The architectural divide between these two GPUs is stark. The RTX 2060 is built on the Turing architecture using a 12 nm process at TSMC, a significant leap from the 28 nm process used for the GTX 660 Ti's Kepler architecture. This process shrink allows the RTX 2060 to pack 10,800 million transistors onto a 445 mm² die, compared to the GTX 660 Ti's 3,540 million transistors on a 294 mm² die.

The most significant functional differences are the inclusion of 30 RT cores and 240 tensor cores in the RTX 2060, which are entirely absent from the GTX 660 Ti. These dedicated hardware units enable real-time ray tracing and AI-accelerated features that the older Kepler card cannot perform. The RTX 2060 also features a more modern feature set, supporting DirectX 12 Ultimate (12_2) and Vulkan 1.4, whereas the GTX 660 Ti is limited to DirectX 12 (11_0) and Vulkan 1.2.175. Furthermore, the RTX 2060's memory subsystem is fundamentally newer, utilizing 6 GB of GDDR6 with a 336.0 GB/s bandwidth, compared to the GTX 660 Ti's 2 GB of GDDR5 with 144.2 GB/s. The shading unit count also increased from 1,344 to 1,920, and the texture mapping units went from 112 to 120, though the RTX 2060's 48 ROPs double the GTX 660 Ti's 24.

The Verdict

The data is unambiguous. The RTX 2060 is the superior product by every measurable benchmark metric included in the fact pack. In the two head-to-head tests available, it wins decisively with a 330.2% lead in Geekbench OpenCL and a 338.6% lead in Geekbench Vulkan. For any user looking for modern gaming performance, hardware-accelerated ray tracing, or AI features, the RTX 2060 is the only sensible choice.

The GTX 660 Ti is an end-of-life product from a previous generation. Its 2 GB of VRAM and older architecture are bottlenecks for modern workloads. Its only conceivable advantage would be in a legacy system where its specific power connector setup (2x 6-pin) or physical dimensions are required, but even then its performance is drastically lower. The benchmark data shows that while both cards sit in a similar percentile of all GPUs (58th and 57th), this is likely due to the average score formula and the limited number of tests for the GTX 660 Ti, masking the true extent of the RTX 2060's dominance in modern compute tasks.

Specification Differences

Here is a direct comparison of the key specifications where the two cards differ:

• Process Node: 12 nm (RTX 2060) vs. 28 nm (GTX 660 Ti)
• Transistors: 10,800 million vs. 3,540 million
• Die Size: 445 mm² vs. 294 mm²
• Base Clock: 1365 MHz vs. 915 MHz
• Boost Clock: 1680 MHz vs. 980 MHz
• Memory Size: 6 GB vs. 2 GB
• Memory Type: GDDR6 vs. GDDR5
• Memory Bandwidth: 336.0 GB/s vs. 144.2 GB/s
• Shading Units: 1920 vs. 1344
• TMUs: 120 vs. 112
• ROPs: 48 vs. 24
• RT Cores: 30 vs. None
• Tensor Cores: 240 vs. None
• Pixel Rate: 80.64 GPixel/s vs. 27.44 GPixel/s
• Texture Rate: 201.6 GTexel/s vs. 109.8 GTexel/s
• FP32 Performance: 6.451 TFLOPS vs. 2.634 TFLOPS
• TDP: 160 W vs. 150 W
• Power Connectors: 1x 8-pin vs. 2x 6-pin
• Display Outputs: 1x DVI, 1x HDMI 2.0, 2x DisplayPort 1.4a, 1x USB Type-C vs. 2x DVI, 1x HDMI 1.4a, 1x DisplayPort 1.2
• DirectX Support: 12 Ultimate (12_2) vs. 12 (11_0)
• Vulkan Support: 1.4 vs. 1.2.175
• Dimensions (Length): 229 mm vs. 241 mm

Head-to-Head Benchmarks

The benchmark comparison is a one-sided affair. The two tests that pit these cards directly against each other show a massive generational leap in compute performance.

In Geekbench OpenCL, the RTX 2060 delivers a score of 65,014, while the GTX 660 Ti manages only 15,113. This results in a delta of 330.2%, with the RTX 2060 being more than four times faster in this compute workload. This test highlights the combined benefits of the newer architecture, higher shader count, and faster memory.

The Geekbench Vulkan test shows an even larger relative gap. Here, the RTX 2060 scores 65,846, compared to the GTX 660 Ti's 15,012, a delta of 338.6%. This reinforces that the RTX 2060's advantage is not limited to a single API but extends to modern graphics interfaces. The GTX 660 Ti has no wins in the head-to-head data; the RTX 2060 wins both tests, giving it a 2-0 record.

Where Each One Wins

Based on the data, the use-case split is clear-cut.

NVIDIA GeForce RTX 2060: This card wins in every single benchmark category where data exists. It is the definitive choice for any modern application that leverages OpenCL or Vulkan, which includes most contemporary games and compute-heavy creative software. Its inclusion of RT cores and tensor cores also makes it the only card of the two capable of hardware-accelerated ray tracing and DLSS-style AI upscaling, though no specific benchmark data is provided for those features. Its 6 GB of GDDR6 memory and 336.0 GB/s bandwidth provide a substantial advantage in texture-heavy and high-resolution workloads.

NVIDIA GeForce GTX 660 Ti: The data shows no scenario where the GTX 660 Ti is the preferred choice. It loses in both available benchmarks by a wide margin. Its only potential niche would be in a very old system where its 2x 6-pin power connectors are the only available option, or where its shorter length is a requirement, but even then the performance penalty is severe. It is simply outclassed in every performance metric provided.