AMD Radeon 660M vs NVIDIA GeForce GTX 1080 Ti Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive NVIDIA GeForce GTX 1080 Ti victory in the two shared tests. In Geekbench OpenCL, the GTX 1080 Ti scores 67,929, while the AMD Radeon 660M manages only 12,876. That is a 427.6% lead for the NVIDIA card, meaning it delivers more than five times the compute throughput in this workload. The gap is similarly stark in Geekbench Vulkan, where the GTX 1080 Ti posts 40,511 against the Radeon 660M’s 14,748, a 174.7% advantage. The NVIDIA GPU wins both head-to-head tests, giving it a clean 2-0 record.

Context from the database’s average scores reinforces the scale of this mismatch. The GTX 1080 Ti carries an average benchmark score of 15,548, which places it at the 58th percentile of all GPUs. Its nearest rivals include the AMD Radeon R9 M380 (average 15,521, just 0.2% lower), the AMD Radeon Pro W5500 (15,679, 0.8% higher), and the NVIDIA GeForce RTX 2060 (15,290, 1.7% lower). The Radeon 660M, by contrast, sits at a 13,812 average score and the 55th percentile, with rivals like the NVIDIA RTX A2000 Mobile (13,821, 0.1% higher) and AMD Radeon RX 570X (13,871, 0.4% higher) clustered closely around it. The 1,736-point gap in average scores between the two cards is modest compared to the extreme deltas seen in the individual OpenCL and Vulkan tests, but it still represents a clear tier separation.

What is notable is that the GTX 1080 Ti’s best individual results are far above its average. Its Passmark G3D score of 18,600 and Geekbench OpenCL score of 67,929 are outliers that drag the average upward, while the Radeon 660M’s only two recorded benchmarks both sit well below the GTX 1080 Ti’s corresponding numbers. The data suggests the 660M is competitive with older discrete GPUs near the 55th percentile, but it is not in the same performance class as a high-end Pascal card from 2017.

FAQ

Q: Which GPU wins in Geekbench OpenCL?

A: The NVIDIA GeForce GTX 1080 Ti wins decisively, scoring 67,929 versus the AMD Radeon 660M’s 12,876. That is a 427.6% advantage for the NVIDIA card.

Q: How does the Radeon 660M compare in Vulkan performance?

A: The GTX 1080 Ti also wins in Geekbench Vulkan, with a score of 40,511 against the 660M’s 14,748, a 174.7% lead. The AMD GPU is not competitive in either shared test.

Q: What is the average benchmark score for each card?

A: The GTX 1080 Ti has an average benchmark score of 15,548, while the Radeon 660M averages 13,812. The NVIDIA card is 1,736 points higher, which is roughly 12.6% above the AMD part.

Q: Where does each GPU rank among all GPUs?

A: The GTX 1080 Ti sits at the 58th percentile of all GPUs, while the Radeon 660M is at the 55th percentile. Despite the large head-to-head deltas, both cards are near the middle of the overall distribution.

Q: Which rival is closest to the GTX 1080 Ti in average score?

A: The AMD Radeon R9 M380 is the nearest rival, with an average score of 15,521, just 0.2% below the GTX 1080 Ti. The NVIDIA GeForce RTX 2060 is also close, at 15,290, which is 1.7% lower.

Q: Does the Radeon 660M have any winning benchmark?

A: No. The head-to-head data shows the Radeon 660M wins zero tests. The GTX 1080 Ti wins both Geekbench OpenCL and Geekbench Vulkan.

The Verdict

The data is unambiguous: the NVIDIA GeForce GTX 1080 Ti is the far stronger performer for anyone prioritizing raw compute and graphics throughput. Its 427.6% OpenCL lead and 174.7% Vulkan lead are not marginal differences; they represent a completely different performance class. The GTX 1080 Ti also holds a higher average score (15,548 vs. 13,812) and a better percentile rank (58th vs. 55th). Anyone choosing between these two for gaming, rendering, or GPU compute should pick the GTX 1080 Ti without hesitation, provided the system can accommodate its dual-slot size and 250 W power draw.

The AMD Radeon 660M, however, is not without a role. It is an integrated GPU (IGP) with a 40 W TDP and no external power connectors, which makes it suitable for thin-and-light laptops where the GTX 1080 Ti is physically impossible to install. The 660M’s average score of 13,812 places it within 1.5% of the NVIDIA RTX A2000 Mobile and the AMD Radeon RX 570X, meaning it performs in line with those discrete mobile and desktop parts. But the Radeon 660M is not a substitute for the GTX 1080 Ti; it is a power-efficient option for portable systems, not a high-end desktop replacement. The GTX 1080 Ti is the clear choice for desktop users, while the 660M only makes sense where the GTX 1080 Ti cannot physically fit.

Specification Differences

The two GPUs differ in nearly every measurable specification. The GTX 1080 Ti uses 3,584 shading units, 224 texture mapping units, and 88 render output units, while the Radeon 660M has 384 shading units, 24 TMUs, and 16 ROPs. The NVIDIA card’s memory configuration is 11 GB of GDDR5X on a 352-bit bus with 484.4 GB/s bandwidth, whereas the 660M relies on system-shared memory with a system-dependent bandwidth. Clock speeds also differ: the GTX 1080 Ti runs at a base of 1481 MHz and a boost of 1582 MHz, while the 660M has a 1500 MHz base and a 1900 MHz boost. The Radeon 660M has a higher boost clock but far fewer compute units.

The GTX 1080 Ti’s pixel rate is 139.2 GPixel/s and texture rate is 354.4 GTexel/s, compared to the 660M’s 30.40 GPixel/s and 45.60 GTexel/s. FP32 throughput is 11.34 TFLOPS for the NVIDIA card versus 1,459.2 GFLOPS for the AMD part. FP16 performance is 177.2 GFLOPS (1:64 ratio) on the GTX 1080 Ti and 2.918 TFLOPS (2:1 ratio) on the 660M. The TDP gap is enormous: 250 W for the GTX 1080 Ti versus 40 W for the 660M. The NVIDIA card requires a 600 W suggested PSU and dual-slot cooling, while the AMD part is an IGP with no power connectors and no dedicated dimensions. Bus interfaces are PCIe 3.0 x16 for the GTX 1080 Ti and PCIe 4.0 x8 for the 660M. Display outputs are 1x HDMI 2.0 and 3x DisplayPort 1.4a on the NVIDIA card, whereas the 660M is portable-device dependent. The GTX 1080 Ti has a 267 mm length, 112 mm height, and 40 mm width; the 660M has no listed dimensions.

Architecture Differences

The architectural split is fundamental. The NVIDIA GeForce GTX 1080 Ti is built on the Pascal architecture, using the GP102 chip manufactured on TSMC’s 16 nm process. It contains 11,800 million transistors on a 471 mm² die, giving a transistor density of 25.1 million per mm². The AMD Radeon 660M uses the RDNA 2.0 architecture on the Rembrandt chip, fabricated on TSMC’s 6 nm node. It packs 13,100 million transistors into a 208 mm² die, achieving 63.0 million transistors per mm². The 660M’s process node is more advanced and denser, but the GTX 1080 Ti’s larger die and dedicated memory bus give it a bandwidth advantage that the 660M cannot overcome with system-shared memory.

The Radeon 660M includes 6 ray tracing cores, a feature the GTX 1080 Ti lacks entirely. Neither GPU has tensor cores. The DirectX support differs: the 660M supports DirectX 12 Ultimate (12_2), while the GTX 1080 Ti is limited to DirectX 12 (12_1). Both support OpenGL 4.6 and Vulkan 1.4. The Radeon 660M is part of the Navi II IGP (Rembrandt Mobile) generation, succeeding Vega II IGP and preceding Navi III IGP, while the GTX 1080 Ti belongs to the GeForce 10-series, succeeding GeForce 900 and preceding GeForce 20. The release dates are also far apart: the GTX 1080 Ti launched on March 9, 2017, while the 660M arrived on January 3, 2022. Both are now marked as end-of-life in the database.

The core count difference is the most impactful architectural factor. The GTX 1080 Ti has 9.3 times the shading units of the 660M (3584 vs. 384), which directly translates into its massive FP32 and texture rate leads. The 660M’s RDNA 2.0 architecture brings modern features like ray tracing and a higher boost clock, but these do not compensate for the sheer lack of compute hardware. The 660M’s 2:1 FP16 ratio also means it can accelerate half-precision workloads far better than the GTX 1080 Ti’s 1:64 ratio, but in the recorded benchmarks, that advantage does not materialize into a win. The GTX 1080 Ti’s older Pascal architecture, with its dedicated 352-bit GDDR5X interface, remains far more capable in the tests captured by the database.