AMD Radeon RX 7900 XT vs NVIDIA GeForce GTX 660 Ti Comparison

# NVIDIA GeForce GTX 660 Ti vs AMD Radeon RX 7900 XT

The GeForce GTX 660 Ti and Radeon RX 7900 XT represent two distinct eras of GPU design, separated by a decade of architectural evolution. The GTX 660 Ti, a Kepler-based card from 2012, was built on TSMC's 28 nm process and aimed at the mainstream enthusiast segment. The RX 7900 XT, using RDNA 3.0 architecture on a 5 nm node, is a modern high-end part with vastly different compute capabilities. Their benchmark results tell a fascinating story of how different workloads age—and how raw compute power doesn't always translate directly to every test.

Head-to-Head Benchmarks

The two cards split their head-to-head benchmark results evenly, with each winning one test by a substantial margin. In Geekbench OpenCL, the GTX 660 Ti posts a score of 15113, while the RX 7900 XT manages only 12756. That gives the older NVIDIA card an 18.5% advantage in this specific compute workload. This result is counterintuitive given the massive specification gap, but it highlights how OpenCL performance can be heavily influenced by driver maturity and workload characteristics rather than raw hardware throughput.

The Vulkan benchmark tells the opposite story, and then some. The RX 7900 XT scores 71068 in Geekbench Vulkan, while the GTX 660 Ti trails far behind at 15012. The delta is a staggering -78.9% from the AMD card's perspective, meaning the RX 7900 XT is roughly 4.7 times faster in this API. This is the kind of generational leap one would expect from a card with nearly 4 times the shading units and a modern architecture designed for contemporary graphics APIs.

Looking at the broader benchmark landscape, the RX 7900 XT dominates in raw performance metrics. Its PassMark G3D score of 29009 dwarfs anything the GTX 660 Ti could produce, and its PassMark GPU Compute score of 16828 shows strong compute throughput. However, the GTX 660 Ti's average benchmark score of 15063 actually exceeds the RX 7900 XT's average of 13745. This inversion occurs because the RX 7900 XT's benchmark suite includes several low-scoring DX10, DX11, and DX12 PassMark tests (158, 340, and 110 respectively), which drag down its average.

The percentile rankings reflect this unusual situation. The GTX 660 Ti sits at the 57th percentile among all GPUs, while the RX 7900 XT ranks at the 55th percentile. Despite being vastly more powerful on paper, the RX 7900 XT's benchmark profile places it slightly lower in the overall distribution. The nearest rival data corroborates this: the RX 7900 XT's closest competitor is the AMD Radeon 660M with an average score of 13812 (just 0.5% higher), while the GTX 660 Ti's nearest rival is the AMD Radeon Pro 560X at 15082 (0.1% lower).

Where Each One Wins

The GTX 660 Ti wins in legacy compute workloads that favor its architecture and driver optimization. Its Geekbench OpenCL victory is significant, but it also maintains competitive standing in older API scenarios. The card supports DirectX 12 (11_0), which means it can run modern titles but without the full feature set of newer hardware. Its 2 GB of GDDR5 memory on a 192-bit bus provides 144.2 GB/s of bandwidth—adequate for its era but limiting for modern textures and resolutions.

The RX 7900 XT wins decisively in Vulkan-based workloads and modern gaming scenarios. Its 20 GB of GDDR6 memory on a 320-bit bus delivers 800.0 GB/s of bandwidth, a massive advantage for high-resolution texture streaming and compute-heavy effects. The card's 84 ray tracing cores enable hardware-accelerated ray tracing, a feature entirely absent from the GTX 660 Ti. Its DirectX 12 Ultimate (12_2) support includes features like mesh shaders and variable rate shading that the older card cannot utilize.

For gaming, the RX 7900 XT is the clear choice for new titles, especially those leveraging Vulkan or DX12 Ultimate features. The GTX 660 Ti, however, may still find a niche in older applications or specific OpenCL compute tasks where its driver stack excels. The RX 7900 XT's PassMark G2D score of 1232 also suggests stronger 2D performance, though this is rarely a deciding factor for gamers. In terms of raw compute, the RX 7900 XT's 51.48 TFLOPS FP32 performance is approximately 19.5 times the GTX 660 Ti's 2.634 TFLOPS, but this advantage doesn't manifest uniformly across all benchmark suites.

FAQ

Q: Which card has a higher average benchmark score?

A: The NVIDIA GeForce GTX 660 Ti has an average benchmark score of 15063, while the AMD Radeon RX 7900 XT averages 13745. However, this is skewed by the RX 7900 XT's inclusion of several low-scoring PassMark tests.

Q: How do the cards compare in Vulkan performance?

A: The RX 7900 XT scores 71068 in Geekbench Vulkan, which is 78.9% higher than the GTX 660 Ti's 15012 score. This represents the largest performance gap between the two cards.

Q: What is the memory capacity difference?

A: The RX 7900 XT has 20 GB of GDDR6 memory, while the GTX 660 Ti has 2 GB of GDDR5. The RX 7900 XT's memory bandwidth is 800.0 GB/s compared to 144.2 GB/s for the GTX 660 Ti.

Q: Which card supports ray tracing?

A: Only the AMD Radeon RX 7900 XT supports ray tracing, featuring 84 dedicated RT cores. The GTX 660 Ti has no ray tracing capabilities.

Q: What are the power consumption differences?

A: The GTX 660 Ti has a TDP of 150 W and requires a 450 W power supply, while the RX 7900 XT has a TDP of 300 W and requires a 700 W power supply. The RX 7900 XT uses two 8-pin power connectors compared to two 6-pin on the GTX 660 Ti.

Q: How do their DirectX support levels differ?

A: The GTX 660 Ti supports DirectX 12 (11_0), while the RX 7900 XT supports DirectX 12 Ultimate (12_2). The RX 7900 XT also supports newer Vulkan version 1.4 compared to 1.2.175 on the GTX 660 Ti.

Specification Differences

The two cards differ dramatically in nearly every specification. The GTX 660 Ti uses a GK104 chip with 3,540 million transistors on a 294 mm² die, while the RX 7900 XT uses a Navi 31 chip with 57,700 million transistors on a 529 mm² die. This represents a transistor density of 12.0M per mm² for the NVIDIA card versus 109.1M per mm² for the AMD card, reflecting the transition from 28 nm to 5 nm process technology.

Clock speeds differ substantially. The GTX 660 Ti has a base clock of 915 MHz and a boost clock of 980 MHz, while the RX 7900 XT has a base clock of 1387 MHz and a boost clock of 2394 MHz, with a game clock of 2025 MHz. Memory clocks also differ: the GTX 660 Ti operates at 1502 MHz with 6 Gbps effective speed, while the RX 7900 XT runs at 2500 MHz with 20 Gbps effective.

Compute unit counts show a massive gap. The GTX 660 Ti has 1344 shading units, 112 texture mapping units (TMUs), and 24 raster output units (ROPs). The RX 7900 XT features 5376 shading units, 336 TMUs, and 192 ROPs. Pixel and texture rates follow suit: the GTX 660 Ti achieves 27.44 GPixel/s and 109.8 GTexel/s, while the RX 7900 XT reaches 459.6 GPixel/s and 804.4 GTexel/s. FP32 performance is 2.634 TFLOPS versus 51.48 TFLOPS, with the RX 7900 XT also offering 103.0 TFLOPS FP16 via 2:1 ratio.

Physical and interface specifications also differ. The GTX 660 Ti is 241 mm (9.5 inches) long, while the RX 7900 XT is 276 mm (10.9 inches) with a height of 110 mm and width of 51 mm. Both are dual-slot cards. The GTX 660 Ti uses PCIe 3.0 x16, while the RX 7900 XT uses PCIe 4.0 x16. Display outputs vary: the GTX 660 Ti offers 2x DVI, 1x HDMI 1.4a, and 1x DisplayPort 1.2, while the RX 7900 XT provides 1x HDMI 2.1a, 2x DisplayPort 2.1, and 1x USB Type-C.

Architecture Differences

The architectural divide between these GPUs spans a decade of GPU evolution. The GTX 660 Ti is built on NVIDIA's Kepler architecture, released in 2012 as part of the GeForce 600 series. Kepler was designed for efficiency and scalability, with a focus on improving performance per watt over its predecessor. The RX 7900 XT uses AMD's RDNA 3.0 architecture, part of the Radeon RX 7000 series with the codename "Plum Bonito" and chip designation Navi 31. RDNA 3.0 represents a major redesign with a chiplet approach and enhanced compute units.

The process node difference is stark: 28 nm for Kepler versus 5 nm for RDNA 3.0, both fabricated by TSMC. This explains the transistor density jump from 12.0M/mm² to 109.1M/mm². The RX 7900 XT's 84 ray tracing cores are a fundamental architectural addition, enabling hardware-accelerated ray tracing that simply didn't exist in consumer GPUs when Kepler launched. The GTX 660 Ti relies entirely on traditional rasterization.

Memory architecture reflects different design philosophies. The GTX 660 Ti's 192-bit bus with 2 GB GDDR5 is a classic mid-range configuration, while the RX 7900 XT's 320-bit bus with 20 GB GDDR6 targets high-resolution gaming and content creation workloads. The RX 7900 XT's dual 8-pin power connectors and 300 W TDP indicate a power-hungry performance part, whereas the GTX 660 Ti's 150 W TDP and dual 6-pin connectors were considered modest for its time.

API support shows the generational gap. The GTX 660 Ti caps at DirectX 12 (11_0), meaning it can run DX12 titles but only with feature level 11_0, missing many modern effects. The RX 7900 XT supports DirectX 12 Ultimate (12_2), the full-featured modern API. Vulkan support also differs: version 1.2.175 on the GTX 660 Ti versus version 1.4 on the RX 7900 XT. Both support OpenGL 4.6, but the RX 7900 XT's newer driver stack and hardware provide significantly better performance in modern OpenGL applications.

The production status for both cards is end-of-life, with the GTX 660 Ti releasing in August 2012 and the RX 7900 XT in November 2022. Their launch MSRPs were 299 USD and 899 USD, respectively. The GTX 660 Ti's predecessor was the GeForce 500 series, with the GeForce 700 series as its successor. The RX 7900 XT's predecessor was Navi II, with Navi IV following it. These architectural differences—from ray tracing support to memory capacity to API capabilities—explain why the newer card excels in modern workloads while the older card still holds its own in specific legacy benchmarks.