NVIDIA GeForce GTX 660 Ti vs NVIDIA RTX PRO 6000 Blackwell Comparison
NVIDIA GeForce GTX 660 Ti
RTX PRO 6000 Blackwell
PERFORMANCE BENCHMARKS
Analysis: NVIDIA GeForce GTX 660 Ti vs NVIDIA RTX PRO 6000 Blackwell
Head-to-Head Benchmarks
The database shows no direct head-to-head benchmark results between the NVIDIA RTX PRO 6000 Blackwell and the NVIDIA GeForce GTX 660 Ti. Instead, the comparison relies on each card's average benchmark score across their respective test suites, and the picture is starkly one-sided.
The RTX PRO 6000 Blackwell records an average benchmark score of 16,408, based on its 3DMark Steel Nomad DX12 result. The GTX 660 Ti, in contrast, averages 15,063 across its Geekbench OpenCL and Vulkan tests. That is a raw difference of 1,345 points, placing the Blackwell card roughly 8.9% higher in aggregate. More revealing is the percentile placement: the RTX PRO 6000 sits in the 59th percentile of all GPUs, while the GTX 660 Ti lands in the 57th. Despite the massive generational gap, both cards hover in the same mid-range band of the database's overall distribution, which speaks to how benchmark methodology and workload mix can compress or inflate apparent performance gaps.
Looking at each card's nearest rivals clarifies the picture. The RTX PRO 6000 Blackwell scores within 0.6% of the NVIDIA GeForce RTX 5090 D V2 (16,504), and it is effectively tied with the AMD Radeon PRO W7500 (16,415, delta 0%) and AMD Radeon RX 5700 XT (16,361, delta 0.3%). The data shows the Blackwell card is not an outlier in its score class; it sits in a tight cluster where a 0.4% margin separates it from the AMD Radeon Pro 5600M (16,351). In other words, the RTX PRO 6000's benchmark result is competitive with a narrow band of contemporary and recent cards, but it does not dominate them by any wide margin.
The GTX 660 Ti's nearest rivals tell a similar story of tight clustering. It trails the AMD Radeon RX 7600 (15,171) by 0.7%, the NVIDIA GeForce RTX 3050 OEM (15,199) by 0.9%, and the AMD Radeon 680M (15,270) by 1.4%. Its closest competitor is the AMD Radeon Pro 560X (15,082), which sits just 0.1% behind. These deltas are all within a 1.5% spread, indicating that the GTX 660 Ti, despite its age, still performs in line with much newer entry-level and integrated parts in the database's measured workloads.
The biggest win for the RTX PRO 6000 Blackwell is not in any single benchmark, but in the sheer scale of its resources. The fp32 compute figure of 126.0 TFLOPS versus 2.634 TFLOPS for the GTX 660 Ti is a 47.8x gap. The pixel rate of 502.5 GPixel/s versus 27.44 GPixel/s is an 18.3x gap. The texture rate of 1,968.0 GTexel/s versus 109.8 GTexel/s is a 17.9x gap. These are not incremental improvements; they represent multiple orders of magnitude in raw throughput. The GTX 660 Ti's only comparative win is in efficiency, where its 150 W TDP is far lower than the 600 W TDP of the RTX PRO 6000, but that is a power draw figure, not a performance score.
FAQ
Q: Which GPU has a higher average benchmark score in the database?
A: The NVIDIA RTX PRO 6000 Blackwell records an average benchmark score of 16,408, compared to 15,063 for the NVIDIA GeForce GTX 660 Ti, a difference of 1,345 points.
Q: How does each card compare to its nearest rivals?
A: The RTX PRO 6000 Blackwell is within 0.6% of the RTX 5090 D V2 and effectively tied with the Radeon PRO W7500 (0% delta). The GTX 660 Ti is within 1.4% of the Radeon 680M and trails the RX 7600 by 0.7%.
Q: What is the memory capacity difference?
A: The RTX PRO 6000 Blackwell has 96 GB of GDDR7 memory on a 512-bit bus, while the GTX 660 Ti has 2 GB of GDDR5 memory on a 192-bit bus.
Q: Which card supports ray tracing and tensor cores?
A: Only the RTX PRO 6000 Blackwell includes dedicated hardware: 188 RT cores and 752 tensor cores. The GTX 660 Ti has no RT cores and no tensor cores listed in the database.
Q: What is the process node difference?
A: The RTX PRO 6000 Blackwell is built on a 5 nm process at TSMC, while the GTX 660 Ti uses a 28 nm process, also at TSMC.
Q: What are the API support differences?
A: The RTX PRO 6000 Blackwell supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4. The GTX 660 Ti supports DirectX 12 (11_0), OpenGL 4.6, and Vulkan 1.2.175.
Architecture Differences
The architecture gap between these two GPUs is the defining story. The RTX PRO 6000 Blackwell uses the GB202 chip with Blackwell 2.0 architecture, while the GTX 660 Ti uses the GK104 chip with Kepler architecture. Blackwell 2.0 is a modern, feature-complete design with dedicated ray tracing and tensor processing hardware. Kepler predates those technologies entirely.
The transistor count difference is enormous. The GB202 chip packs 92,200 million transistors on a 750 mm² die, yielding a transistor density of 122.9 million per square millimeter. The GK104 chip has 3,540 million transistors on a 294 mm² die, with a density of 12.0 million per square millimeter. That is a 26x increase in transistor count and a 10.2x increase in density, enabled by the move from 28 nm to 5 nm fabrication at the same foundry.
The Blackwell architecture introduces dedicated RT cores (188 of them) and tensor cores (752 of them), which are entirely absent from the Kepler design. These specialized units enable hardware-accelerated ray tracing and AI workloads, capabilities the GTX 660 Ti simply does not possess. The shading unit count also diverges massively: 24,064 shading units on the Blackwell card versus 1,344 on the Kepler card. Texture mapping units number 752 versus 112, and ROPs number 192 versus 24.
Memory architecture is another clear divider. The RTX PRO 6000 Blackwell uses GDDR7 memory, a current-generation standard, while the GTX 660 Ti uses GDDR5. The Blackwell card's memory bus is 512 bits wide, compared to 192 bits on the GTX 660 Ti. The Blackwell card also supports PCIe 5.0 x16, while the GTX 660 Ti uses PCIe 3.0 x16. Display outputs differ as well: the RTX PRO 6000 has 4x DisplayPort 2.1b, while the GTX 660 Ti has 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2.
Specification Differences
The recorded specifications show differences across nearly every measurable field. Clock speeds: the RTX PRO 6000 Blackwell has a base clock of 1590 MHz and a boost clock of 2617 MHz. The GTX 660 Ti has a base clock of 915 MHz and a boost clock of 980 MHz. Memory clock is listed as 1750 MHz (28 Gbps effective) for the Blackwell card versus 1502 MHz (6 Gbps effective) for the GTX 660 Ti.
Memory bandwidth: 1.79 TB/s for the RTX PRO 6000 versus 144.2 GB/s for the GTX 660 Ti. That is a 12.4x difference. The GTX 660 Ti's fp16 performance is not listed (null), while the RTX PRO 6000 shows 126.0 TFLOPS fp16 at a 1:1 ratio with fp32. The Blackwell card also lists 126.0 TFLOPS fp32, while the GTX 660 Ti lists 2.634 TFLOPS.
Power and physical specifications differ significantly. The RTX PRO 6000 has a 600 W TDP with a single 16-pin power connector and a suggested PSU of 1000 W. The GTX 660 Ti has a 150 W TDP with two 6-pin connectors and a suggested PSU of 450 W. Both are dual-slot cards, but the RTX PRO 6000 is longer at 304 mm (12 inches) versus 241 mm (9.5 inches) for the GTX 660 Ti. The Blackwell card also has listed height (137 mm, 5.4 inches) and width (40 mm, 1.6 inches), while those fields are null for the GTX 660 Ti.
Production status and release dates diverge completely. The RTX PRO 6000 Blackwell was released on 2025-03-17 and is marked as Active. The GTX 660 Ti was released on 2012-08-15 and is marked as End-of-life. The GTX 660 Ti's predecessor is the GeForce 500 series and its successor is the GeForce 700 series, while the RTX PRO 6000's predecessor is Workstation Ada and it has no successor listed.
The Verdict
The data points to a straightforward conclusion for different user profiles. The RTX PRO 6000 Blackwell is the clear choice for anyone needing current-generation features, massive memory capacity, and dedicated compute hardware. Its 96 GB GDDR7 memory, 188 RT cores, and 752 tensor cores make it suitable for workloads that the GTX 660 Ti cannot handle at all. The benchmark scores show it competing with modern cards like the RTX 5090 D V2 and the Radeon PRO W7500, which places it in a contemporary performance class.
The GTX 660 Ti, based on the recorded data, remains relevant only in legacy contexts. Its average benchmark score of 15,063 keeps it within 1.4% of modern integrated graphics like the AMD Radeon 680M, which suggests it can still run basic or older workloads. Its 150 W TDP is a quarter of the RTX PRO 6000's power draw, which could matter in systems with limited power delivery. But its 2 GB memory capacity, lack of RT and tensor cores, and end-of-life production status make it a poor fit for any modern compute task.
The percentile data is the most telling statistic. Both cards sit in the 57th to 59th percentile of all GPUs, meaning the database's overall distribution places them in the same broad performance tier despite the 13-year and multiple-generation gap. That said, the RTX PRO 6000 achieves that percentile with 126.0 TFLOPS of fp32 compute and 1.79 TB/s of bandwidth, while the GTX 660 Ti achieves it with 2.634 TFLOPS and 144.2 GB/s. The workloads that produced these scores are different, and the benchmark suites are not directly comparable.
For a user choosing between these two, the decision hinges on workload. The RTX PRO 6000 Blackwell is the only option with modern API support (DirectX 12 Ultimate, Vulkan 1.4), ray tracing hardware, and tensor cores. The GTX 660 Ti offers a lower power draw and a far smaller physical footprint, but its DirectX 12 (11_0) support and lack of dedicated compute units place it firmly in the past. The database shows the RTX PRO 6000 as the active, forward-looking product, while the GTX 660 Ti is end-of-life with a successor already listed. For any new deployment, the RTX PRO 6000 Blackwell is the data-backed recommendation. For legacy system maintenance or low-power secondary use, the GTX 660 Ti retains a narrow niche.