AMD Radeon 780M vs NVIDIA GeForce GTX 660 Ti Comparison

Where Each One Wins

The benchmark data paints a clear picture: the AMD Radeon 780M wins both recorded head-to-head tests against the NVIDIA GeForce GTX 660 Ti. In Geekbench OpenCL, the 780M scores 18,602 versus the GTX 660 Ti's 15,113, a lead of 23.1%. The gap widens dramatically in Geekbench Vulkan, where the 780M posts 33,683 against 15,012, a 124.4% advantage. That is not a narrow margin; it is a generational stomping in API-level workloads.

What does this mean for real use? The 780M's wins are concentrated in compute-oriented and modern API scenarios. OpenCL reflects general-purpose GPU compute, and Vulkan represents current-generation gaming and rendering workloads. The GTX 660 Ti has no wins in the recorded data, which suggests it struggles in exactly the areas where newer software expects efficiency. If the task involves Vulkan, which is increasingly the standard for modern game engines, the 780M is decisively ahead.

The 780M also holds a higher percentile rank among all GPUs: 61st percentile versus the GTX 660 Ti's 57th. Its average benchmark score of 17,588 comfortably exceeds the GTX 660 Ti's 15,063. Both cards sit in the mid-range of the overall GPU distribution, but the 780M is closer to the upper edge of that band. For a builder choosing between these two, the data says the 780M is the better fit for anything that uses modern graphics APIs or compute acceleration.

Architecture Differences

The architectural gap between these two is vast, and it explains the benchmark results. The 780M uses AMD's RDNA 3.0 architecture on a 4 nm TSMC process, with a chip codenamed Phoenix. The GTX 660 Ti uses NVIDIA's Kepler architecture on a 28 nm process, also TSMC-made. The process difference alone is enormous: 4 nm versus 28 nm means the 780M packs transistors far more densely, 142.6 million per square millimeter versus 12.0 million for the GTX 660 Ti.

The 780M integrates 25,390 million transistors on a 178 mm² die. The GTX 660 Ti has only 3,540 million transistors on a larger 294 mm² die. That is a density inversion: the 780M fits nearly seven times more transistors into a smaller physical area. The GTX 660 Ti's larger die with fewer transistors reflects its 2012-era design, while the 780M benefits from a decade of process improvements.

The 780M also brings features the GTX 660 Ti lacks entirely. It has 12 ray tracing cores, which the GTX 660 Ti does not have at all. The 780M supports DirectX 12 Ultimate (12_2), while the GTX 660 Ti is limited to DirectX 12 (11_0). Vulkan support differs too: the 780M runs Vulkan 1.4, the GTX 660 Ti only Vulkan 1.2.175. The 780M's FP16 throughput matches its FP32 at 8.909 TFLOPS, whereas the GTX 660 Ti has no recorded FP16 capability. Memory is another split: the 780M uses system-shared memory with system-dependent bandwidth, while the GTX 660 Ti has dedicated 2 GB GDDR5 on a 192-bit bus with 144.2 GB/s bandwidth.

Power characteristics diverge sharply. The 780M has a 15 W TDP, the GTX 660 Ti a 150 W TDP. The 780M is an integrated graphics processor with no power connectors; the GTX 660 Ti requires two 6-pin connectors and a 450 W suggested PSU. The 780M also uses PCIe 4.0 x8, while the GTX 660 Ti uses PCIe 3.0 x16. These are not just spec sheet differences; they shape what each card can do in a real system.

Head-to-Head Benchmarks

The most striking result is the Vulkan test. The 780M scores 33,683, which is 124.4% higher than the GTX 660 Ti's 15,012. That is more than double the performance. Vulkan is a low-overhead API that rewards efficient hardware architecture, and the RDNA 3.0 design clearly excels there. The GTX 660 Ti's Kepler architecture was built for an older DirectX era, and it shows in Vulkan workloads.

The OpenCL test is closer but still favors the 780M decisively. The 780M scores 18,602, a 23.1% improvement over the GTX 660 Ti's 15,113. OpenCL is a general compute API, and the 780M's higher shading unit count, 768 versus 1,344, might seem like a disadvantage at first glance. But the 780M's FP32 throughput of 8.909 TFLOPS versus 2.634 TFLOPS for the GTX 660 Ti tells a different story. The 780M processes more floating-point operations per clock despite having fewer shaders, which is a direct benefit of RDNA 3.0's design.

The pixel and texture rates also favor the 780M. It achieves 92.80 GPixel/s and 139.2 GTexel/s, versus 27.44 GPixel/s and 109.8 GTexel/s for the GTX 660 Ti. Pixel fill rate is more than three times higher on the 780M, which matters for resolution-heavy rendering. Texture rate is about 27% higher, a smaller but still meaningful margin. These rates explain why the 780M wins in synthetic benchmarks: it simply moves more data through its pipeline per second.

The GTX 660 Ti's higher shading unit count, 1,344 versus 768, and TMU count, 112 versus 48, do not translate into wins. The 780M compensates with higher clocks and a more efficient architecture. The 780M boosts to 2900 MHz, while the GTX 660 Ti boosts to only 980 MHz. That clock advantage, combined with the modern architecture, overcomes the raw unit count difference.

FAQ

Q: Which card has better Vulkan performance?

A: The AMD Radeon 780M, by a wide margin. Its Geekbench Vulkan score is 33,683, which is 124.4% higher than the GTX 660 Ti's 15,012.

Q: Does the GTX 660 Ti have ray tracing support?

A: No. The GTX 660 Ti has no ray tracing cores listed in the database, while the 780M has 12 RT cores.

Q: How do their power requirements compare?

A: The 780M has a 15 W TDP and no power connectors, making it suitable for integrated graphics use. The GTX 660 Ti has a 150 W TDP, requires two 6-pin connectors, and needs a 450 W suggested PSU.

Q: What is the memory configuration difference?

A: The 780M uses system-shared memory with system-dependent bandwidth. The GTX 660 Ti has 2 GB of dedicated GDDR5 on a 192-bit bus with 144.2 GB/s bandwidth.

Q: Which card is more current in production status?

A: The 780M is listed as Active, while the GTX 660 Ti is End-of-life. The 780M was released in 2024; the GTX 660 Ti in 2012.

Q: How do their average benchmark scores compare?

A: The 780M has an average benchmark score of 17,588, while the GTX 660 Ti averages 15,063. The 780M also sits at the 61st percentile of all GPUs, versus the 57th for the GTX 660 Ti.

The Verdict

The data is unambiguous: the AMD Radeon 780M outperforms the NVIDIA GeForce GTX 660 Ti in every recorded benchmark. The 780M wins both head-to-head tests, holds a higher average score, and ranks higher in the overall GPU distribution. For modern workloads that use Vulkan or OpenCL, the 780M is the only rational choice.

The GTX 660 Ti retains one practical advantage: its dedicated 2 GB GDDR5 memory with 144.2 GB/s bandwidth. That is fixed, local memory, not system-shared. If a workload requires stable, dedicated VRAM and the system's main memory is slow, the GTX 660 Ti might still function, but the benchmark data does not support a performance win. The 780M's system-dependent bandwidth means its memory performance varies with the host system, so in a poorly configured machine, the GTX 660 Ti could feel more consistent.

For a builder today, the choice is straightforward. The 780M is active, low-power at 15 W, and supports modern APIs like DirectX 12 Ultimate and Vulkan 1.4. The GTX 660 Ti is end-of-life, consumes 150 W, and is stuck with DirectX 12 (11_0) and Vulkan 1.2.175. If the system has fast shared memory, the 780M will deliver dramatically better results, especially in Vulkan. If the system has slow shared memory, the GTX 660 Ti's dedicated VRAM might mitigate the gap, but it cannot overcome the 124.4% Vulkan deficit. The 780M is the superior part in nearly every measurable way.

Specification Differences

| Specification | AMD Radeon 780M | NVIDIA GeForce GTX 660 Ti |

| --- | --- | --- |

| Architecture | RDNA 3.0 | Kepler |

| Process Node | 4 nm | 28 nm |

| Transistors | 25,390 million | 3,540 million |

| Die Size | 178 mm² | 294 mm² |

| Transistor Density | 142.6M / mm² | 12.0M / mm² |

| Base Clock | 800 MHz | 915 MHz |

| Boost Clock | 2900 MHz | 980 MHz |

| Memory Size | System Shared | 2 GB |

| Memory Type | System Shared | GDDR5 |

| Memory Bus Width | System Shared | 192 bit |

| Memory Bandwidth | System Dependent | 144.2 GB/s |

| Shading Units | 768 | 1,344 |

| TMUs | 48 | 112 |

| ROPs | 32 | 24 |

| RT Cores | 12 | None |

| Pixel Rate | 92.80 GPixel/s | 27.44 GPixel/s |

| Texture Rate | 139.2 GTexel/s | 109.8 GTexel/s |

| FP32 | 8.909 TFLOPS | 2.634 TFLOPS |

| FP16 | 8.909 TFLOPS (1:1) | None |

| TDP | 15 W | 150 W |

| Slot Width | IGP | Dual-slot |

| Power Connectors | None | 2x 6-pin |

| Suggested PSU | None | 450 W |

| Bus Interface | PCIe 4.0 x8 | PCIe 3.0 x16 |

| DirectX | 12 Ultimate (12_2) | 12 (11_0) |

| Vulkan | 1.4 | 1.2.175 |

| Production Status | Active | End-of-life |

| Release Date | 2024-01-30 | 2012-08-15 |

| Launch MSRP | None | 299 USD |