AMD Radeon Pro 460

The AMD Radeon Pro 460 is a mobile-class discrete GPU built on the GCN 4.0 architecture, using the Baffin chip fabricated on a 14 nm process at GlobalFoundries. Released on 2016-10-29, this part belongs to the Radeon Pro Mac (400 Series) generation and has reached end-of-life production status. The benchmark data shows an average score of 17575 across Geekbench tests, placing it at the 59th percentile of all GPUs. It packs 3,000 million transistors on a 123 mm² die, resulting in a transistor density of 24.4M per mm². Its base clock of 850 MHz can boost to 907 MHz, and it offers 1024 shading units, 64 texture mapping units, and 16 ROPs. With a 35 W TDP and an IGP slot width, this is a low-power part designed for integration into portable systems, and its display outputs are portable device dependent.

Memory Subsystem

The Radeon Pro 460 comes with 4 GB of GDDR5 memory on a 128-bit bus, delivering a bandwidth of 81.28 GB/s. The memory clock runs at 1270 MHz, translating to 5.1 Gbps effective. This configuration is modest by current standards, and the data indicates that the 128-bit bus width is the primary constraint on memory throughput. For high-resolution workloads, particularly at 1440p or 4K, the 4 GB frame buffer will fill quickly with detailed textures and geometry. The 81.28 GB/s bandwidth becomes a limiting factor when streaming large amounts of texture data, potentially causing stutter or reduced frame rates in memory-intensive scenes.

At lower resolutions, such as 1080p, the memory subsystem is adequate for most titles, but the pixel rate of 14.51 GPixel/s and texture rate of 58.05 GTexel/s suggest that the card's fill-rate capabilities align with mid-range expectations. The 1:1 FP16 to FP32 ratio of 1.858 TFLOPS means no compute advantage for half-precision workloads, which is relevant for certain professional applications that rely on FP16 acceleration. Overall, the memory subsystem is balanced for its class but will not handle high-resolution textures or large datasets without compromise. Users targeting 4K gaming or heavy multi-monitor setups should consider that the 4 GB capacity and 128-bit bus will likely bottleneck performance, while 1080p and lower resolutions will see more consistent results.

Ray Tracing and Feature Set

The Radeon Pro 460 does not include dedicated ray tracing cores or tensor cores, as these fields are null in the specification data. Consequently, any ray tracing effects must be handled through compute shaders or software fallbacks, which will incur a significant performance penalty. The card supports DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3, providing broad API compatibility for modern games and applications. The DirectX 12 (12_0) feature level ensures access to features like bindless resources and asynchronous compute, but the lack of hardware RT acceleration means that ray-traced shadows, reflections, and global illumination are not practical in real-time scenarios.

The FP32 compute performance is 1.858 TFLOPS, which is the same as the FP16 rate (1:1). This symmetry is unusual and indicates that the GPU does not offer specialized half-precision throughput, which is often used in machine learning or certain scientific workloads. For users relying on Vulkan 1.3 or DirectX 12, the card can run modern APIs, but its feature set is fundamentally rasterization-focused. The absence of tensor cores also rules out any AI-accelerated features like DLSS or similar upscaling technologies. In summary, the Radeon Pro 460 is a capable rasterization GPU with solid API support, but it is not designed for ray tracing or AI-driven features.

Who Should Consider It

The benchmark results indicate that the Radeon Pro 460 performs best in Metal-based workloads, scoring 20560 on Geekbench Metal, which is notably higher than its OpenCL score of 15290 and Vulkan score of 16875. This suggests that the card is optimized for Apple's Metal API, making it a suitable choice for users within the macOS ecosystem who rely on Metal-accelerated applications. The 59th percentile ranking places it in the middle of the GPU landscape, meaning it can handle mainstream tasks but will not excel in demanding professional or gaming scenarios.

For gaming, the card is best suited for 1080p resolution with medium to high settings, where the 4 GB VRAM and 81.28 GB/s bandwidth are sufficient. At 1440p, users will need to lower texture quality and disable anti-aliasing to maintain playable frame rates. The card is not recommended for 4K gaming or for workloads that require large memory footprints, such as 3D rendering with high-resolution textures or video editing with multi-layer timelines. For professional users, the Radeon Pro 460 can handle light CAD, photo editing, and general productivity tasks, especially in Mac environments where Metal performance is prioritized. However, for compute-heavy tasks like machine learning or scientific simulation, the 1.858 TFLOPS FP32 performance is a limiting factor.

FAQ

Q: What is the average benchmark score of the AMD Radeon Pro 460?

A: The average benchmark score is 17575, based on Geekbench tests for Metal, OpenCL, and Vulkan.

Q: Does the Radeon Pro 460 support DirectX 12?

A: Yes, it supports DirectX 12 (12_0), along with OpenGL 4.6 and Vulkan 1.3.

Q: Does this GPU have dedicated ray tracing or tensor cores?

A: No, the specification data lists no ray tracing cores and no tensor cores for this card.

Q: What is the memory bandwidth of the Radeon Pro 460?

A: The memory bandwidth is 81.28 GB/s, provided by 4 GB of GDDR5 on a 128-bit bus.

Q: What is the TDP of this GPU?

A: The thermal design power is 35 W, with no power connectors required.

Q: Is the Radeon Pro 460 still in production?

A: No, its production status is listed as end-of-life, with a release date of 2016-10-29.

How It Compares

The Radeon Pro 460 sits in a tightly contested performance bracket, with its nearest rivals all within a 2% margin. Against the AMD Radeon Pro 560, the Pro 460 posts an average score of 17575 versus 17497, making it 0.4% faster. This near-tie means the two cards are effectively interchangeable in real-world performance, though the Pro 460 edges ahead in the aggregate benchmark.

The AMD Radeon HD 7790 is the closest competitor, scoring 17666, which puts the Pro 460 at a 0.5% deficit. The deltaPct of -0.5 indicates the Pro 460 is slightly slower than the HD 7790, but the difference is negligible and within run-to-run variance. Both cards deliver similar rasterization performance, making the choice dependent on other factors like power efficiency or driver support.

Against the NVIDIA Tesla K40c, the Pro 460 achieves a 0.6% advantage, with scores of 17575 and 17468 respectively. Despite the Tesla's professional compute orientation, the Pro 460 holds a slight lead in the Geekbench aggregate, though the Tesla may excel in specific compute workloads that are not reflected in these benchmarks.

Finally, the AMD FirePro W7000 scores 17790, giving it a 1.2% lead over the Pro 460. This is the largest gap among the rivals, but still modest. The FirePro W7000 outperforms the Pro 460 by a small margin, likely due to its different memory configuration or driver optimizations. Overall, the Pro 460 is firmly in the middle of this group, with no single rival dominating by a significant margin.

Power and Cooling

The Radeon Pro 460 has a thermal design power of 35 W, making it a highly efficient part suitable for thin and light laptops. Its slot width is listed as IGP (integrated graphics processor), meaning it is soldered directly onto the motherboard rather than being a replaceable module. This design eliminates the need for external power connectors, as the card draws all its power from the system's internal power delivery. The absence of a suggested PSU rating further confirms that this is not a card for desktop builds; it is intended for portable devices where power and thermal budgets are tight.

The bus interface is PCIe 3.0 x8, which provides adequate bandwidth for the card's performance level, though it is half the lanes of a typical x16 slot. Since the display outputs are portable device dependent, the actual video output options vary by the host laptop or all-in-one system. Cooling is handled by the host device's thermal solution, which is not specified in the data, but the 35 W TDP ensures that even a modest cooling solution can manage the heat output. Users should note that the card is end-of-life, so replacement or upgrade paths may be limited to the host system's design constraints.