AMD Radeon Pro 570X

The AMD Radeon Pro 570X is an end-of-life graphics processor built on the 14nm GCN 4.0 architecture (chip Ellesmere), featuring 1792 shading units, 112 TMUs, and 32 ROPs. It operates at a base clock of 1000 MHz and a boost clock of 1105 MHz, paired with 4GB of GDDR5 memory on a 256-bit bus delivering 217.0 GB/s bandwidth. In aggregate benchmarks, it achieves an average score of 31682, placing it in the 75th percentile of all GPUs. Its nearest rivals include the NVIDIA TITAN RTX, Intel Arc Pro A30M, NVIDIA Quadro RTX 8000, and NVIDIA GRID M60-1Q, with performance deltas ranging from 0% to 1.5%.

Memory Subsystem

The Radeon Pro 570X equips 4GB of GDDR5 memory across a 256-bit interface, resulting in 217.0 GB/s of bandwidth. The memory clock runs at 1695 MHz, which translates to 6.8 Gbps effective. For high-resolution workloads, the 4GB capacity is the more immediate constraint than the bandwidth; textures and geometry at ultra-high resolutions can easily exceed this buffer. The 256-bit bus and 217 GB/s are adequate for standard resolutions, but the data suggests that pushing to higher resolutions will likely hit the frame buffer ceiling first. The pixel rate of 35.36 GPixel/s and texture rate of 123.8 GTexel/s further indicate that the card can fill frames quickly at lower resolutions, but the memory size remains the limiting factor. With 5,700 million transistors on a 232 mm² die (24.6M / mm²), the memory subsystem is balanced for its era but not designed for massive texture sets.

Ray Tracing and Feature Set

The card does not list any ray tracing cores or tensor cores, meaning there is no dedicated hardware for real-time ray tracing or AI-accelerated features. Instead, it relies on the GCN 4.0 architecture's compute units for all rendering. API support includes DirectX 12 (12_0), OpenGL 4.6, and Vulkan 1.3, which covers modern graphics APIs but without the ray tracing extensions that some newer cards offer. The FP32 performance is 3.960 TFLOPS, and FP16 is identical at 3.960 TFLOPS (1:1), indicating no half-rate acceleration for FP16. This makes the card a straightforward compute device, but not one optimized for ray-traced effects. The lack of tensor cores also means no dedicated hardware for deep learning or denoising tasks, which are increasingly common in modern renderers.

Benchmark Performance

In Geekbench tests, the card scores 40066 in Metal, 27604 in OpenCL, and 27376 in Vulkan. The Metal score is notably higher than the other two, suggesting that the card is particularly well-tuned for Metal API workloads, consistent with its Radeon Pro Mac lineage. The average benchmark score is 31682. Against its nearest rivals, the performance is remarkably tight: it matches the NVIDIA TITAN RTX (31676, delta 0%), trails the Intel Arc Pro A30M by 0.7% (31894 vs 31682), leads the NVIDIA Quadro RTX 8000 by 0.9% (31401 vs 31682), and outperforms the NVIDIA GRID M60-1Q by 1.5% (31220 vs 31682). These deltas are small, indicating that the 570X sits in a highly competitive performance band despite its older architecture. The 75th percentile ranking further underscores that it outperforms the majority of GPUs in the database, while the near-tie with the TITAN RTX is especially notable given the latter's high-end positioning.

How It Compares

NVIDIA TITAN RTX: The two cards are effectively tied, with the 570X averaging 31682 and the TITAN RTX 31676 (0% delta). This is surprising given the TITAN RTX's flagship status, but the benchmark scores show parity in these specific tests. The delta of 0% suggests that for the workloads captured by Geekbench, the Radeon Pro 570X delivers equivalent performance.

Intel Arc Pro A30M: The 570X is 0.7% slower, with the Arc Pro A30M scoring 31894. The difference is negligible in real-world terms, but the data shows the Intel card holds a slight edge. The -0.7% delta indicates that the 570X trails by a fraction of a percent, which is within run-to-run variance.

NVIDIA Quadro RTX 8000: The 570X is 0.9% faster than the Quadro RTX 8000 (31401 vs 31682). Again, a small margin, but the Radeon Pro comes out ahead. The positive delta of 0.9% means the 570X outperforms the Quadro in these benchmarks, a notable result given the Quadro's professional positioning.

NVIDIA GRID M60-1Q: The 570X leads by 1.5%, scoring 31682 against the GRID's 31220. This is the largest delta among the nearest rivals, yet still within a couple of percent. The data shows the 570X holds a consistent advantage over this virtualized GPU.

Power and Cooling

The Radeon Pro 570X has a TDP of 150W. It requires no external power connectors, drawing power solely from the PCIe slot. The slot width is listed as IGP, indicating it is designed for integrated or proprietary mounting, typical of a Mac-oriented board. No suggested PSU is provided in the data, but the 150W TDP and lack of connectors imply that a standard system power supply should suffice, provided the motherboard can deliver the required power through the PCIe slot. The card is end-of-life, and its production status suggests it is no longer actively manufactured. The absence of a launch MSRP means no pricing information is available, but the power profile is modest and unlikely to stress typical power supplies.

FAQ

Q: What is the memory bandwidth of the AMD Radeon Pro 570X?

A: The memory bandwidth is 217.0 GB/s, achieved via 4GB of GDDR5 on a 256-bit bus.

Q: Does the Radeon Pro 570X support hardware ray tracing?

A: No. The card has no ray tracing cores or tensor cores listed; it relies on traditional compute units.

Q: What is the average benchmark score of the Radeon Pro 570X?

A: The average benchmark score is 31682, placing it in the 75th percentile of all GPUs.

Q: Which API yields the highest score for this card?

A: The Geekbench Metal test scores 40066, significantly higher than OpenCL (27604) and Vulkan (27376).

Q: What is the thermal design power (TDP) of this card?

A: The TDP is 150W, and it does not require any external power connectors.

Q: How does the Radeon Pro 570X compare to the NVIDIA TITAN RTX?

A: The two are effectively tied, with a delta of 0% (31682 vs 31676).

Who Should Consider It

The benchmark data indicates that the Radeon Pro 570X is a solid mid-range performer, sitting at the 75th percentile. Its Metal score is exceptional, making it a strong candidate for Mac-based compute and graphics workloads, particularly those optimized for Metal. However, the 4GB frame buffer and 217.0 GB/s bandwidth are modest by modern standards; for standard-resolution rendering, the card should handle most titles, but ultra-high-resolution or high-texture workloads may exceed the memory capacity. The 3.960 TFLOPS FP32 performance is adequate for general compute, but the lack of RT cores means it is not suited for ray-traced applications. Users who primarily run Metal-optimized software and stay within the memory limits will find the performance competitive, as evidenced by its near-parity with the TITAN RTX in these benchmarks. Those needing higher memory capacity or dedicated ray tracing hardware should look elsewhere, but for its intended niche, the 570X delivers consistent, if not spectacular, results.