AMD Radeon 660M vs NVIDIA GeForce GTX 1080 Comparison

The AMD Radeon 660M and NVIDIA GeForce GTX 1080 occupy opposite ends of the performance spectrum, and the recorded data leaves no ambiguity about which one delivers higher raw compute. Across the two shared benchmark tests, the GeForce GTX 1080 wins both outright, with the Radeon 660M trailing by 74.9% in OpenCL and 51.5% in Vulkan. The GTX 1080 also holds a significantly higher average benchmark score, 11,960 versus 13,812 for the 660M, though the percentile rankings tell a more nuanced story: the 660M sits at the 55th percentile while the GTX 1080 sits at the 51st. This apparent contradiction stems from the fact that the GTX 1080's average is dragged down by its inclusion of older DirectX 9 and compute tests, while the 660M only has two modern API benchmarks in its record. The verdict is clear for anyone needing sustained high-end performance: the GTX 1080 is the dominant part. The 660M, however, is a modern integrated solution with architectural advantages that matter in specific contexts, particularly power efficiency and API support.

FAQ

Q: Which GPU wins in the head-to-head benchmark tests?

A: The NVIDIA GeForce GTX 1080 wins both recorded tests. In geekbench_opencl, it scores 51,204 against the AMD Radeon 660M's 12,876, a delta of 74.9% in favor of NVIDIA. In geekbench_vulkan, the GTX 1080 scores 30,398 versus 14,748, a 51.5% advantage.

Q: How do the two compare in terms of average benchmark scores?

A: The AMD Radeon 660M has a higher average benchmark score of 13,812, while the NVIDIA GeForce GTX 1080 averages 11,960. The 660M's average is based on only two tests, both of which are modern API workloads, whereas the GTX 1080's average includes legacy DirectX 9 and compute tests that lower its mean.

Q: What are the percentile rankings for each GPU?

A: The AMD Radeon 660M lands at the 55th percentile among all GPUs in the database. The NVIDIA GeForce GTX 1080 sits at the 51st percentile. Despite the GTX 1080's higher raw scores in individual tests, the 660M ranks slightly higher overall due to the composition of its benchmark set.

Q: What is the process node difference between the two?

A: The AMD Radeon 660M is built on a 6 nm process at TSMC, while the NVIDIA GeForce GTX 1080 uses a 16 nm process, also at TSMC. The 660M's smaller node allows for a much higher transistor density, 63.0 million per square millimeter compared to 22.9 million for the GTX 1080.

Q: Does the Radeon 660M support hardware ray tracing?

A: Yes. The AMD Radeon 660M features 6 ray tracing cores, while the NVIDIA GeForce GTX 1080 has no ray tracing cores listed in its specifications. The 660M also supports DirectX 12 Ultimate (12_2), whereas the GTX 1080 is limited to DirectX 12 (12_1).

Q: What is the memory configuration for each GPU?

A: The AMD Radeon 660M uses system shared memory, with its size, type, bus width, and bandwidth all dependent on the host system. The NVIDIA GeForce GTX 1080 has 8 GB of dedicated GDDR5X memory on a 256-bit bus, delivering 320.3 GB/s of bandwidth.

Architecture Differences

The architectural gap between these two GPUs is generational and foundational. The AMD Radeon 660M is built on the RDNA 2.0 architecture, using the Rembrandt chip, and is classified as a Navi II IGP for mobile platforms. It is manufactured on a 6 nm process at TSMC, packing 13,100 million transistors into a 208 mm² die. The NVIDIA GeForce GTX 1080, by contrast, uses the Pascal architecture with the GP104 chip, fabricated on a 16 nm process at the same foundry, with 7,200 million transistors on a larger 314 mm² die. The 660M's transistor density is 63.0 million per square millimeter versus 22.9 million for the GTX 1080, a direct consequence of the process node difference.

The compute core layouts diverge sharply. The Radeon 660M has 384 shading units, 24 texture mapping units, and 16 raster output pipelines. The GTX 1080 has 2,560 shading units, 160 TMUs, and 64 ROPs. This explains the massive throughput differences: the GTX 1080 delivers 8.873 TFLOPS of FP32 compute versus 1,459.2 GFLOPS for the 660M, and its texture rate is 277.3 GTexel/s compared to 45.60 GTexel/s. Pixel rates follow the same pattern, with the GTX 1080 hitting 110.9 GPixel/s and the 660M managing 30.40 GPixel/s.

Ray tracing is a key differentiator. The 660M includes 6 dedicated ray tracing cores, a feature entirely absent from the GTX 1080, which has no such hardware. This aligns with their API support: the 660M is rated for DirectX 12 Ultimate (12_2), while the GTX 1080 only reaches DirectX 12 (12_1). Both support OpenGL 4.6 and Vulkan 1.4, so those APIs are not distinguishing factors.

Memory architecture is fundamentally different. The 660M relies on system shared memory, meaning its capacity, type, bus width, and bandwidth are all dependent on the host laptop's configuration. The GTX 1080 has 8 GB of dedicated GDDR5X memory on a 256-bit bus with 320.3 GB/s of bandwidth, and its memory clock is listed at 1251 MHz with 10 Gbps effective data rate. The 660M's memory clock is listed as "System Shared," underscoring its dependence on external hardware.

Power and physical design also separate these parts. The 660M has a TDP of 40 W, uses no power connectors, and is an integrated graphics processor (IGP) with a slot width of IGP. The GTX 1080 draws 180 W, requires a single 8-pin power connector, a suggested PSU of 450 W, and is a dual-slot card measuring 267 mm in length, 112 mm in height, and 40 mm in width. The bus interface differs as well, with the 660M using PCIe 4.0 x8 and the GTX 1080 using PCIe 3.0 x16.

The Verdict

The data directs a straightforward choice for most users. The NVIDIA GeForce GTX 1080 is the clear winner for anyone prioritizing raw compute performance. It beats the AMD Radeon 660M by 74.9% in OpenCL and 51.5% in Vulkan, and its FP32 throughput of 8.873 TFLOPS dwarfs the 660M's 1,459.2 GFLOPS. The GTX 1080's dedicated 8 GB of GDDR5X memory with 320.3 GB/s bandwidth provides a level of sustained performance that system-shared memory cannot guarantee. For gaming, content creation, or any workload that demands high fill rates and texture throughput, the GTX 1080 is the only rational pick, provided the system can accommodate its 180 W TDP and dual-slot footprint.

The AMD Radeon 660M is the better choice for a different set of constraints. Its 40 W TDP makes it suitable for thin-and-light laptops where power draw and heat are critical. Its 6 nm process and 13,100 million transistors on a 208 mm² die show a modern design that prioritizes efficiency. The inclusion of 6 ray tracing cores and DirectX 12 Ultimate support means the 660M can handle newer graphical features that the GTX 1080 cannot, despite the latter's raw power. The 660M also has a higher percentile ranking, 55th versus 51st, and a higher average benchmark score, 13,812 versus 11,960, though those figures reflect the limited test set rather than absolute superiority.

Users with existing desktop systems that can house a dual-slot card and supply a 450 W PSU should choose the GTX 1080. Users shopping for a new ultraportable laptop or a low-power mini PC should focus on the 660M. The GTX 1080 is end-of-life, as is the 660M, but the latter's modern architecture means it remains relevant for current API workloads. The GTX 1080's Pascal architecture is older, and its lack of ray tracing cores is a structural disadvantage in games that use that feature. In short, pick the GTX 1080 for performance, pick the 660M for efficiency and modern feature support.

Specification Differences

The two GPUs differ in nearly every measurable specification. The process node is 6 nm for the AMD Radeon 660M and 16 nm for the NVIDIA GeForce GTX 1080, both from TSMC. Transistor count is 13,100 million for the 660M versus 7,200 million for the GTX 1080, though the die sizes are 208 mm² and 314 mm² respectively, yielding transistor densities of 63.0M per mm² and 22.9M per mm².

Clock speeds differ, with the 660M running at a 1500 MHz base and 1900 MHz boost, while the GTX 1080 runs at 1607 MHz base and 1733 MHz boost. The 660M's memory is system shared, while the GTX 1080 has 8 GB of GDDR5X on a 256-bit bus with 320.3 GB/s bandwidth. Shading units number 384 for the 660M and 2,560 for the GTX 1080. Texture mapping units are 24 versus 160, and ROPs are 16 versus 64. The 660M has 6 ray tracing cores; the GTX 1080 has none.

Pixel rate is 30.40 GPixel/s for the 660M and 110.9 GPixel/s for the GTX 1080. Texture rate is 45.60 GTexel/s versus 277.3 GTexel/s. FP32 compute is 1,459.2 GFLOPS for the 660M and 8.873 TFLOPS for the GTX 1080. FP16 performance is 2.918 TFLOPS for the 660M and 138.6 GFLOPS for the GTX 1080, with the 660M using a 2:1 ratio and the GTX 1080 using a 1:64 ratio.

TDP is 40 W for the 660M and 180 W for the GTX 1080. The 660M is an IGP with no power connectors, while the GTX 1080 is dual-slot with a single 8-pin connector and a suggested PSU of 450 W. The bus interface is PCIe 4.0 x8 for the 660M and PCIe 3.0 x16 for the GTX 1080. Display outputs are portable-device dependent for the 660M, whereas the GTX 1080 offers 1x DVI, 1x HDMI 2.0, and 3x DisplayPort 1.4a. DirectX support is 12 Ultimate (12_2) for the 660M and 12 (12_1) for the GTX 1080. Release dates are January 2022 for the 660M and May 2016 for the GTX 1080.

Head-to-Head Benchmarks

Two shared benchmark tests are recorded in the database, and the NVIDIA GeForce GTX 1080 wins both. In geekbench_opencl, the GTX 1080 scores 51,204 against the AMD Radeon 660M's 12,876. This is a delta of 74.9%, meaning the GTX 1080 outperforms the 660M by nearly three-quarters in OpenCL compute workloads. The margin is substantial enough that no system configuration could close the gap; the GTX 1080's 2,560 shading units and dedicated memory simply overwhelm the 660M's 384 shading units and shared memory setup.

In geekbench_vulkan, the GTX 1080 scores 30,398, while the 660M manages 14,748. The delta here is 51.5%, still a decisive win for NVIDIA but a closer margin than the OpenCL result. The 660M's Vulkan score is 14.5% higher than its OpenCL score, suggesting that its RDNA 2.0 architecture handles modern APIs relatively well. The GTX 1080, however, still doubles the 660M's output in this test. The GTX 1080's Vulkan score of 30,398 is lower than its OpenCL score of 51,204, which is unusual and indicates that the Pascal architecture is better optimized for OpenCL than Vulkan.

The head-to-head record shows 2 wins for the GTX 1080 and 0 for the 660M. The 660M's average benchmark score of 13,812 is higher than the GTX 1080's 11,960, but that average is computed from only two tests, both of which the 660M loses. The GTX 1080's average includes additional tests such as PassMark DirectX 9 (211), DirectX 10 (93), DirectX 11 (124), DirectX 12 (55), G2D (888), G3D (15,586), GPU compute (7,614), and 3DMark Steel Nomad DX12 (1,560). These legacy and varied tests pull the GTX 1080's average down, but they also demonstrate its versatility across a wider range of workloads.

Where Each One Wins

The NVIDIA GeForce GTX 1080 wins in every test where the two are directly compared. It is the superior part for OpenCL compute, with a 74.9% advantage, and for Vulkan rendering, with a 51.5% advantage. It also holds the edge in raw specification metrics: FP32 performance, texture rate, pixel rate, and memory bandwidth. Any workload that relies on these capabilities, such as high-resolution gaming, 3D rendering, or GPU-accelerated compute, favors the GTX 1080. Its 8 GB of dedicated GDDR5X memory ensures that data transfers do not compete with the CPU for system bandwidth, which is a structural advantage over the 660M's shared memory design.

The AMD Radeon 660M wins in efficiency and modern feature support. Its 40 W TDP is a fraction of the GTX 1080's 180 W, making it viable for laptops and compact systems where the GTX 1080 cannot physically fit. The 660M's 6 nm process and 63.0M transistors per mm² density indicate a far more efficient design. It also has 6 ray tracing cores, which the GTX 1080 lacks entirely. For any application that uses DirectX 12 Ultimate features, the 660M is the only one of the two that qualifies, as the GTX 1080 is limited to DirectX 12 (12_1). The 660M's FP16 performance of 2.918 TFLOPS is also much higher than the GTX 1080's 138.6 GFLOPS, which matters for workloads that can exploit half-precision arithmetic.

The percentile rankings and average scores favor the 660M in the database's aggregate metrics. The 660M sits at the 55th percentile versus the GTX 1080's 51st, and its average benchmark score of 13,812 is higher than the GTX 1080's 11,960. These figures reflect the 660M's focus on modern API benchmarks, while the GTX 1080's average is diluted by older DirectX tests. For users who only care about current-generation workloads, the 660M's relative standing is better than its raw loss in the head-to-head would suggest. For users who need maximum performance across all workloads, the GTX 1080 is unequivocally the stronger choice.