NVIDIA GeForce GTX 660 Ti vs NVIDIA Quadro RTX 4000 Comparison

FAQ

Q: How do the two GPUs compare in average benchmark score?

A: The NVIDIA Quadro RTX 4000 has an average benchmark score of 17,789, while the NVIDIA GeForce GTX 660 Ti scores 15,063. The Quadro RTX 4000 outperforms the GTX 660 Ti by roughly 18% on average across all recorded tests.

Q: Which card wins in Geekbench OpenCL performance?

A: The Quadro RTX 4000 scores 74,540 in Geekbench OpenCL, versus 15,113 for the GTX 660 Ti. That represents a 393.2% advantage for the Quadro, meaning it delivers nearly five times the compute throughput in this workload.

Q: What about Vulkan performance? Is the gap similar?

A: The gap widens slightly in Geekbench Vulkan. The Quadro RTX 4000 records 78,844 points, while the GTX 660 Ti manages 15,012. The delta is 425.2%, making the Quadro roughly 5.3 times faster in this graphics API test.

Q: Which card has more shading units and texture mapping units?

A: The Quadro RTX 4000 contains 2,304 shading units and 144 TMUs. The GTX 660 Ti has 1,344 shading units and 112 TMUs. The Quadro also packs 64 ROPs compared to 24 on the GTX 660 Ti.

Q: Do these cards support ray tracing or tensor cores?

A: Yes for the Quadro RTX 4000, which includes 36 RT cores and 288 tensor cores. The GTX 660 Ti has neither, as its Kepler architecture predates those hardware features entirely.

Q: What are the launch MSRP values for each card?

A: The Quadro RTX 4000 had a launch MSRP of 899 USD. The GTX 660 Ti had a launch MSRP of 299 USD. Both products are now end-of-life.

Where Each One Wins

The recorded data splits cleanly between two very different eras of GPU design. The Quadro RTX 4000 wins both head-to-head benchmark tests, and by enormous margins. In Geekbench OpenCL, the Quadro delivers 393.2% higher scores, and in Geekbench Vulkan it leads by 425.2%. This is not a close contest; it is a generational chasm.

However, the GTX 660 Ti still holds relevance in one specific context: its percentile standing. The GTX 660 Ti sits at the 57th percentile of all GPUs in the database, while the Quadro RTX 4000 sits at the 61st percentile. That 4-point gap is far narrower than the raw benchmark deltas suggest. The explanation lies in the benchmark distribution: the GTX 660 Ti's nearest rivals, such as the AMD Radeon Pro 560X (15,082 average score, 0.1% behind) and the AMD Radeon RX 7600 (15,171 average score, 0.7% ahead), cluster tightly around its score. The Quadro RTX 4000's rivals, including the AMD Radeon HD 7790 (17,666 average score, 0.7% behind) and the NVIDIA GeForce RTX 4060 (17,639 average score, 0.9% behind), also cluster tightly, but at a much higher performance tier.

For users prioritizing raw compute throughput, particularly in OpenCL or Vulkan workloads, the Quadro RTX 4000 is the clear winner. For users with legacy DirectX 9 or DirectX 10 applications, the GTX 660 Ti is the only one of the two with recorded scores in those categories, though the database does not include comparable DirectX 9 or DirectX 10 tests for the Quadro RTX 4000. The Quadro's recorded DirectX 11 score of 128 and DirectX 12 score of 52 (Passmark) suggest it handles newer APIs, while the GTX 660 Ti's absence from those specific tests limits direct comparison.

Architecture Differences

The Quadro RTX 4000 uses the TU104 chip built on Turing architecture, fabricated on a 12 nm process at TSMC. It packs 13,600 million transistors into a 545 mm² die, yielding a transistor density of 25.0 million per square millimeter. The GTX 660 Ti uses the GK104 chip on Kepler architecture, built on a 28 nm process, also at TSMC. It contains 3,540 million transistors across a 294 mm² die, for a density of 12.0 million per square millimeter.

This architectural gap explains nearly everything about the performance difference. Turing introduced dedicated RT cores for real-time ray tracing and tensor cores for AI acceleration; the Quadro RTX 4000 has 36 of the former and 288 of the latter. Kepler predates both features entirely. The Quadro also supports DirectX 12 Ultimate (12_2), while the GTX 660 Ti tops out at DirectX 12 (11_0). Vulkan support differs too: the Quadro supports Vulkan 1.4, the GTX 660 Ti supports Vulkan 1.2.175. OpenGL support is identical at 4.6.

Memory technology also diverges sharply. The Quadro uses GDDR6 with 8 GB capacity on a 256-bit bus, achieving 416.0 GB/s bandwidth. The GTX 660 Ti uses GDDR5 with 2 GB capacity on a 192-bit bus, achieving 144.2 GB/s. That is a 2.9x difference in raw memory bandwidth.

Specification Differences

The two cards differ across nearly every measurable specification. The Quadro RTX 4000 has a base clock of 1005 MHz and a boost clock of 1545 MHz. The GTX 660 Ti runs at 915 MHz base and 980 MHz boost. Memory clocks also differ: the Quadro's memory runs at 1625 MHz with 13 Gbps effective data rate, while the GTX 660 Ti's memory runs at 1502 MHz with 6 Gbps effective.

Shading units: 2,304 versus 1,344. TMUs: 144 versus 112. ROPs: 64 versus 24. The Quadro's pixel rate is 98.88 GPixel/s versus 27.44 GPixel/s on the GTX 660 Ti. Texture rate: 222.5 GTexel/s versus 109.8 GTexel/s. FP32 throughput: 7.119 TFLOPS versus 2.634 TFLOPS. The Quadro also supports FP16 at 14.24 TFLOPS (2:1 ratio); the GTX 660 Ti has no recorded FP16 capability.

Power and physical design differ as well. The Quadro RTX 4000 has a TDP of 160 W, is single-slot, and uses one 8-pin power connector. The GTX 660 Ti has a TDP of 150 W, is dual-slot, and uses two 6-pin connectors. Both recommend a 450 W power supply and use PCIe 3.0 x16. Both cards measure 241 mm in length (9.5 inches). Display outputs differ: the Quadro offers 3x DisplayPort 1.4a plus 1x USB Type-C, while the GTX 660 Ti offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2. The Quadro's height is 111 mm (4.4 inches); the GTX 660 Ti's height is not recorded.

Head-to-Head Benchmarks

The database includes two direct head-to-head tests between these cards. In Geekbench OpenCL, the Quadro RTX 4000 scores 74,540 against 15,113 for the GTX 660 Ti. The delta is 393.2%. This test measures general-purpose compute performance across a wide range of workloads, and the Turing card's advantage comes from its combination of higher shader count, faster clocks, and much larger memory bandwidth.

In Geekbench Vulkan, the Quadro RTX 4000 scores 78,844 against 15,012 for the GTX 660 Ti. The delta is 425.2%. Vulkan is a low-overhead graphics API that scales well with raw hardware throughput. The Quadro's lead here is even larger than in OpenCL, suggesting that its architecture handles modern graphics workloads with particular efficiency relative to Kepler.

The GTX 660 Ti does not win either recorded test. The Quadro RTX 4000 secures 2 wins, the GTX 660 Ti secures 0. But the broader database context matters. The GTX 660 Ti's average score of 15,063 places it at the 57th percentile, with its nearest rival (the AMD Radeon Pro 560X) just 0.1% behind. The Quadro RTX 4000's average of 17,789 places it at the 61st percentile, with its nearest rival (the AMD Radeon HD 7790) just 0.7% behind. Both cards sit in competitive neighborhoods for their respective eras, but the Quadro's neighborhood is simply faster.

The Verdict

The data points to a decisive outcome. The NVIDIA Quadro RTX 4000 outperforms the NVIDIA GeForce GTX 660 Ti in every recorded benchmark, with advantages ranging from 393.2% in OpenCL to 425.2% in Vulkan. Its architecture, built on Turing with 36 RT cores, 288 tensor cores, and 8 GB of GDDR6 memory, represents a fundamentally newer design philosophy. The GTX 660 Ti, built on Kepler with 2 GB of GDDR5, belongs to a previous generation that lacks dedicated ray tracing, tensor acceleration, and modern API support.

For users working with OpenCL or Vulkan compute workloads, the Quadro RTX 4000 is the only rational choice from this pairing. Its average benchmark score of 17,789 versus 15,063 also indicates that across the full database of tests, the Quadro maintains a consistent edge. The GTX 660 Ti, however, remains a capable card for its era. Its 57th percentile ranking shows it still outperforms a meaningful share of all GPUs in the database, and its nearest rivals, the AMD Radeon Pro 560X and AMD Radeon RX 7600, sit within 1% of its average score.

The verdict depends on workload context. For modern applications, particularly those leveraging DirectX 12 Ultimate, Vulkan 1.4, or FP16 compute, the Quadro RTX 4000 is overwhelmingly superior. For legacy DirectX 9 or DirectX 10 applications, the GTX 660 Ti has recorded scores (205 and 108 respectively in Passmark tests) that the Quadro cannot directly match, simply because no comparable tests exist in the database for the newer card. Users running such legacy software may find the GTX 660 Ti sufficient. Everyone else should look to the Quadro RTX 4000.