AMD Radeon 680M vs NVIDIA GeForce GTX 660 Ti Comparison

The AMD Radeon 680M and the NVIDIA GeForce GTX 660 Ti occupy opposite ends of the hardware spectrum, yet benchmark data places them in a statistical tie on average. The Radeon 680M, an integrated processor from 2023, achieves an average benchmark score of 15270, while the dedicated GTX 660 Ti from 2012 scores 15063. The data shows a near-identical percentile ranking at 57 for both parts, but the way they achieve those scores differs dramatically, driven by a decade of architectural evolution.

Where Each One Wins

The benchmark results are unequivocal in their split. The AMD Radeon 680M wins both head-to-head comparisons, with the GeForce GTX 660 Ti failing to secure a single victory in the available tests. The largest margin comes in the Geekbench OpenCL compute test, where the Radeon 680M scores 23468 against the GTX 660 Ti’s 15113, a decisive 55.3% advantage. This is a pure compute workload that favors the newer architecture’s raw throughput and modern instruction set.

The second head-to-head test, Geekbench Vulkan, shows a similar trend. The Radeon 680M posts 21965 points, while the GTX 660 Ti manages 15012, resulting in a 46.3% delta in favor of the AMD part. The GTX 660 Ti’s Vulkan support is limited to version 1.2.175, whereas the Radeon 680M supports Vulkan 1.4, and this generational gap in API capability likely contributes to the substantial performance difference. The Radeon 680M’s wins are not marginal; they represent a significant performance tier separation in these specific workloads.

Architecture Differences

The fundamental difference lies in manufacturing process and design philosophy. The Radeon 680M is built on a 6 nm TSMC process, packing 13,100 million transistors into a 208 mm² die, yielding a transistor density of 63.0M per mm². In contrast, the GTX 660 Ti uses a 28 nm TSMC process with 3,540 million transistors on a larger 294 mm² die, resulting in a density of just 12.0M per mm². This is a five-fold difference in density, enabling the Radeon 680M to integrate far more complex logic in a smaller area.

The Radeon 680M is an integrated graphics processor (IGP) based on RDNA 2.0 architecture, part of the Rembrandt+ chip. It operates with a base clock of 2000 MHz and a boost clock of 2200 MHz. Its memory is system-shared, with bandwidth described as system-dependent. This is a fundamental departure from the GTX 660 Ti, which is a discrete card with 2 GB of dedicated GDDR5 memory on a 192-bit bus, delivering 144.2 GB/s of bandwidth. The Radeon 680M has no dedicated memory, relying entirely on shared system resources.

Core configurations also differ sharply. The Radeon 680M has 768 shading units, 48 texture mapping units (TMUs), and 32 raster operations pipelines (ROPs). It also includes 12 ray tracing cores, a feature entirely absent from the Kepler-based GTX 660 Ti. The GTX 660 Ti has 1344 shading units, 112 TMUs, and 24 ROPs. Despite having fewer shading units, the Radeon 680M achieves higher pixel rate (70.40 GPixel/s vs 27.44 GPixel/s) and a comparable texture rate (105.6 GTexel/s vs 109.8 GTexel/s). Its FP32 throughput is 3.379 TFLOPS versus 2.634 TFLOPS for the GTX 660 Ti, a 28% advantage in raw floating-point performance.

The Radeon 680M also supports FP16 at 6.758 TFLOPS (2:1 ratio), a capability the GTX 660 Ti lacks entirely. Power consumption tells the integration story: the Radeon 680M is rated at 50 W TDP with no power connectors, while the GTX 660 Ti draws 150 W and requires two 6-pin connectors. The Radeon 680M is an IGP with a PCIe 4.0 x8 interface, whereas the GTX 660 Ti is a dual-slot card using PCIe 3.0 x16 and measuring 241 mm in length. The architectural gulf is clear: one is a modern, power-efficient integrated solution, the other a legacy discrete GPU with a higher power footprint.

The Verdict

The data directs a clear choice based on workload and system context. For any application leveraging compute APIs like OpenCL or Vulkan, the AMD Radeon 680M is the superior performer, with benchmark deltas of 55.3% and 46.3% respectively. The Radeon 680M also offers modern feature support including DirectX 12 Ultimate (12_2), Vulkan 1.4, and ray tracing cores, making it the only viable option for contemporary graphics workloads.

The GTX 660 Ti’s sole advantage lies in its discrete memory configuration. With 2 GB of dedicated GDDR5 and fixed 144.2 GB/s bandwidth, it does not compete with system memory for capacity. However, the benchmark data does not show this translating into wins in the tested workloads. The GTX 660 Ti is end-of-life, while the Radeon 680M is active production. For a new system build, the Radeon 680M is the rational choice, delivering higher compute performance at a fraction of the power draw. The GTX 660 Ti remains a historical artifact; its performance tier is matched or exceeded by an integrated solution from a decade later.

FAQ

Q: Which GPU has a higher average benchmark score?

A: The AMD Radeon 680M has an average benchmark score of 15270, compared to the NVIDIA GeForce GTX 660 Ti’s 15063. The Radeon 680M’s nearest rival is the NVIDIA GeForce GTX 580 at 15283, with a delta of -0.1%.

Q: How does the Geekbench OpenCL performance compare?

A: The Radeon 680M scores 23468 in Geekbench OpenCL, while the GTX 660 Ti scores 15113. This gives the Radeon 680M a 55.3% performance advantage in this test.

Q: Does the GTX 660 Ti support ray tracing?

A: No. The GTX 660 Ti has no ray tracing cores listed. The Radeon 680M includes 12 ray tracing cores.

Q: What are the memory specifications for each product?

A: The Radeon 680M uses system-shared memory with system-dependent bandwidth. The GTX 660 Ti has 2 GB of GDDR5 memory on a 192-bit bus with 144.2 GB/s bandwidth.

Q: What is the difference in power consumption?

A: The Radeon 680M has a TDP of 50 W and requires no power connectors. The GTX 660 Ti has a TDP of 150 W and requires two 6-pin power connectors.

Q: Which product supports a newer DirectX version?

A: The Radeon 680M supports DirectX 12 Ultimate (12_2). The GTX 660 Ti supports DirectX 12 (11_0).

Head-to-Head Benchmarks

The Geekbench OpenCL result is the most lopsided comparison in the dataset. The Radeon 680M’s score of 23468 dwarfs the GTX 660 Ti’s 15113, a delta of 55.3%. This workload stresses raw compute throughput, where the Radeon 680M’s 3.379 TFLOPS FP32 performance and modern RDNA 2.0 architecture provide a substantial edge over the GTX 660 Ti’s 2.634 TFLOPS. The GTX 660 Ti’s higher shading unit count (1344 vs 768) does not compensate for the architectural efficiency and clock speed advantages of the Radeon 680M, which runs at 2200 MHz boost versus 980 MHz boost.

The Vulkan benchmark reinforces the same conclusion, though with a slightly narrower margin. The Radeon 680M scores 21965, and the GTX 660 Ti scores 15012, resulting in a 46.3% delta. Vulkan is a lower-level API that benefits from modern driver optimization and hardware design. The Radeon 680M supports Vulkan 1.4, while the GTX 660 Ti is limited to Vulkan 1.2.175. This version gap, combined with the Radeon 680M’s superior pixel rate (70.40 GPixel/s vs 27.44 GPixel/s), explains the consistent victory.

It is notably the GTX 660 Ti has no result for the 3DMark Steel Nomad DX12 test, while the Radeon 680M scores 378 in that benchmark. The absence of this data point for the GTX 660 Ti likely reflects its lack of full DirectX 12 Ultimate support, as it only reaches DirectX 12 (11_0). In the two tests where both products have scores, the Radeon 680M wins decisively. The average benchmark scores place the Radeon 680M at 15270, which is 0.5% ahead of the NVIDIA GeForce RTX 3050 OEM and 0.7% ahead of the AMD Radeon RX 7600. The GTX 660 Ti’s average of 15063 places it 1.4% behind the Radeon 680M, with its nearest rival being the AMD Radeon Pro 560X at 15082, a delta of -0.1%. The performance relationship is clear: the integrated Radeon 680M outperforms the dedicated GTX 660 Ti in every measurable benchmark category available.