AMD Radeon RX Vega M GH

The AMD Radeon RX Vega M GH is an integrated graphics processor built on the 14 nm GlobalFoundries process, featuring the Polaris 22 chip with GCN 4.0 architecture. It packs 5,000 million transistors on a 208 mm² die, yielding a transistor density of 24.0M per mm². The GPU operates at a base clock of 1063 MHz and a boost clock of 1190 MHz, with its HBM2 memory running at 800 MHz (1600 Mbps effective). This end-of-life part, released on January 31, 2018, holds a 73rd percentile ranking among all GPUs in the benchmark database, with an average benchmark score of 29197 across its two recorded tests.

Memory Subsystem

The RX Vega M GH is equipped with 4 GB of HBM2 memory, a configuration that stands out for an integrated part. The memory interface is notably wide at 1024 bit, which enables a peak bandwidth of 204.8 GB/s. This bandwidth figure is the defining characteristic of the memory subsystem; it allows the GPU to feed its 1536 shading units and 96 texture mapping units without the bottlenecks typically seen in narrower-bus designs.

For high-resolution workloads, the combination of 4 GB capacity and high bandwidth presents a mixed picture. The 204.8 GB/s bandwidth is sufficient for maintaining decent frame pacing at 1440p in many titles, as the data throughput is rarely the limiting factor at that resolution. However, the 4 GB frame buffer becomes a constraint at 4K, where texture-heavy scenes and high-detail settings can exceed this capacity, forcing the driver to spill over to system memory. The 64 ROPs produce a pixel rate of 76.16 GPixel/s, which aligns with the memory bandwidth to deliver consistent fill rates. In practice, the memory subsystem favors 1080p and 1440p gaming, where the high bandwidth compensates for the modest capacity, rather than 4K, where capacity limits become apparent.

Power and Cooling

The RX Vega M GH carries a TDP of 100 W, a figure that is modest for the performance level indicated by its benchmark scores. This power draw is managed within an integrated graphics package (IGP), meaning the GPU is designed to be soldered onto a motherboard or embedded in a portable device rather than installed in a standard PCIe slot. Consequently, the slot width is listed as "IGP," and no dedicated power connectors are required; the GPU draws all its power through the host platform.

Because it is an integrated solution, there is no suggested PSU rating in the data, and the cooling solution is entirely dependent on the host device. The 14 nm process node and 100 W TDP suggest that a capable air cooler, similar to what would be used for a mid-range laptop CPU, is sufficient to manage thermals. The absence of a discrete power connector simplifies system integration, but it also means that the GPU’s performance is inherently tied to the thermal and power delivery capabilities of the host device. Users should not expect to overclock or modify power limits, as the IGP form factor offers no such provisions.

Benchmark Performance

The benchmark data presents two specific tests: Geekbench OpenCL and Geekbench Vulkan, yielding scores of 27125 and 31268, respectively. The average benchmark score, which combines these results, is 29197. This places the GPU at the 73rd percentile of all GPUs in the database, indicating that it outperforms roughly three-quarters of the tested hardware.

Looking at the Vulkan score specifically, the 31268 result is notably higher than the OpenCL score, suggesting that the GCN 4.0 architecture handles modern, low-level graphics APIs with particular efficiency. The fp32 compute throughput is 3.656 TFLOPS, with fp16 performance at the same 3.656 TFLOPS (1:1 ratio), which indicates that the GPU does not offer double-rate half-precision compute, a feature that some competing architectures include. In terms of raw texture throughput, the 96 TMUs deliver a texture rate of 114.2 GTexel/s, which is adequate for detailed geometry at 1080p but may struggle with extreme anisotropic filtering at higher resolutions.

When compared to its nearest rivals, the RX Vega M GH’s average score of 29197 is remarkably close to the pack. It trails the AMD Radeon Pro Vega 20 by a mere 0.1% (which scores 29235), while it leads the Intel Arc A370M by 0.1% (which scores 29175). The margin against the AMD Radeon RX 6650 XT and the AMD Radeon RX 6800M is larger, with the Vega M GH ahead by 0.6% in both cases (those rivals score 29024 and 29011, respectively). These deltas are within the noise margin of typical benchmark variance, meaning that in real-world applications, the RX Vega M GH effectively trades blows with these discrete and integrated competitors.

How It Compares

Intel Arc A370M: The RX Vega M GH edges out the Arc A370M by a razor-thin 0.1% in average benchmark score (29197 vs. 29175). This parity is notable because the Arc A370M is a discrete mobile GPU, while the Vega M GH is an integrated part. The data suggests that the Vega M GH’s wider 1024-bit memory bus compensates for any architectural advantages the Arc might have, resulting in nearly identical overall performance.

AMD Radeon Pro Vega 20: This is the closest rival, with the Pro Vega 20 scoring 29235, which is 0.1% higher than the Vega M GH. The Pro Vega 20 is a professional-grade mobile GPU, and its slight lead indicates that the Vega M GH’s gaming-oriented design does not sacrifice much in compute-heavy workloads. The 0.1% delta is statistically insignificant, so the two can be considered performance equals in the aggregate.

AMD Radeon RX 6650 XT: The Vega M GH leads this desktop discrete GPU by 0.6% (29197 vs. 29024). The RX 6650 XT is a newer architecture, but the benchmark data shows the Vega M GH holding a slight edge in the averaged OpenCL and Vulkan tests. This is surprising given the 6650 XT’s newer process node, but the Vega M GH’s higher memory bandwidth (204.8 GB/s) likely contributes to its advantage in memory-bound synthetic tests.

AMD Radeon RX 6800M: Similarly, the Vega M GH is 0.6% ahead of the RX 6800M (29197 vs. 29011). The RX 6800M is a high-end mobile GPU, yet the data shows the integrated Vega M GH outperforming it in these specific benchmarks. This result underscores the efficiency of the HBM2 memory implementation, which allows the Vega M GH to achieve competitive scores despite its lower TDP and integrated nature.

Who Should Consider It

The RX Vega M GH is positioned for users who need solid 1080p performance without the space or power budget for a discrete GPU. The 73rd percentile ranking, combined with the 204.8 GB/s bandwidth, makes it well-suited for 1080p gaming at high settings, where the 4 GB HBM2 frame buffer is less likely to be exhausted. At 1440p, the GPU can still deliver playable frame rates in less demanding titles, but users should expect to lower texture quality to stay within the 4 GB capacity.

For 4K resolution, the data suggests this is not the intended use case. The 4 GB memory capacity is a hard limit, and the 76.16 GPixel/s pixel rate, while respectable, is not enough to drive 4K at high detail in modern games. The GPU is better suited for content creation tasks that leverage OpenCL or Vulkan acceleration, as the 3.656 TFLOPS fp32 compute is competitive with the rival cards listed. Users who prioritize portability, such as those in compact laptops or mini-PCs, will find the IGP form factor advantageous, as it eliminates the need for separate power connectors and reduces overall system complexity. Conversely, users targeting high-refresh-rate 1440p or 4K gaming should look elsewhere, as the benchmark margins over rivals are too small to suggest headroom for such demanding workloads.

FAQ

Q: What is the average benchmark score for the AMD Radeon RX Vega M GH?

A: The average benchmark score is 29197, which is the midpoint of its Geekbench OpenCL score of 27125 and its Geekbench Vulkan score of 31268.

Q: How does the RX Vega M GH compare to the Intel Arc A370M?

A: The RX Vega M GH has an average score of 29197, which is 0.1% higher than the Intel Arc A370M’s score of 29175. This indicates near-identical performance between the two.

Q: What is the memory bandwidth and bus width of this GPU?

A: The GPU features a 1024-bit memory bus with HBM2 memory, providing a total bandwidth of 204.8 GB/s. The memory operates at 800 MHz, which translates to 1600 Mbps effective.

Q: Does the RX Vega M GH require a dedicated power connector?

A: No, the GPU is an integrated part (IGP) with no power connectors listed. It draws power through the host platform and does not have a suggested PSU rating.

Q: What is the TDP of the RX Vega M GH, and what does it imply for cooling?

A: The TDP is 100 W, which is modest for its performance level. Given the integrated form factor, cooling is dependent on the host device, but a capable air cooler should suffice.

Q: Which rival does the RX Vega M GH trail by the largest margin?

A: The GPU trails the AMD Radeon Pro Vega 20 by 0.1% (29197 vs. 29235), which is the only rival with a higher average score. It leads all other listed rivals by 0.6% or less.