AMD Radeon Vega 11 Embedded

The AMD Radeon Vega 11 Embedded is a 14 nm integrated GPU from the Vega IGP (Great Horned Owl) generation, built on the Raven chip and manufactured by GlobalFoundries. The die contains 4,940 million transistors across 210 mm², giving a transistor density of 23.5M / mm². It was released on 2018-02-12 and is now end-of-life. The GPU is built on the GCN 5.0 architecture, uses an IGP slot width, and relies entirely on System Shared memory rather than dedicated VRAM.

Benchmark Performance

The benchmarks array for the Vega 11 Embedded is empty, and the average benchmark score field is 0. That zero is not the result of any recorded workload; it is simply the empty placeholder in the database. The only relative ranking supplied is the percentileVsAllGpus value of 50, which places the GPU at the 50th percentile of all tracked GPUs. In a database-wide sense, this means it sits at the median: half of all tracked parts are above it and half are below it. With no nearest-rival entries, this percentile is the sole comparative anchor.

Without workload scores, the compute and fill-rate numbers define the device’s theoretical performance envelope. The GPU has 704 shading units, 44 texture mapping units, and 8 ROPs. Those fixed hardware blocks produce a texture rate of 57.24 GTexel/s and a pixel rate of 10.41 GPixel/s. The pixel rate is constrained by the 8 ROPs, so fill-rate-bound scenes will hit that ceiling regardless of shader throughput. The texture rate, meanwhile, is exactly what the 44 TMUs support at the listed peak rates.

The compute side is expressed through FP32 and FP16 throughput. FP32 performance is 1.832 TFLOPS, while FP16 performance is 3.664 TFLOPS with a 2:1 ratio. The FP16 number is exactly double the FP32 number, which reflects the architecture’s packed rate behavior. That makes the Vega 11 Embedded capable of handling FP16 workloads at a higher rate than FP32, but the 1.832 TFLOPS FP32 figure remains the more general measure for everyday rendering and compute.

The clocks run from a 300 MHz base to a 1301 MHz boost. That large clock range allows the GPU to idle at a low frequency and move up when needed, but the database does not include sustained clock behavior. API support is modern: DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3 are all listed. This means the part is not functionally obsolete in terms of feature support, even though no benchmark data exists to quantify how those features perform.

Memory Subsystem

The Vega 11 Embedded has no dedicated memory of its own. The memory size, memory type, and memory bus width are all listed as “System Shared.” That triple is significant because it removes every fixed memory specification from the equation. There is no VRAM size to check, no memory type such as GDDR5 or HBM, and no bus width to divide into a bandwidth calculation.

The memory bandwidth field is listed as “System Dependent.” In practice, this means the GPU’s effective memory throughput is whatever the host platform supplies. A motherboard with a stronger memory subsystem will produce more available bandwidth, while a weaker platform will reduce it. For high-resolution workloads, this is the single most important variable in the part’s behavior. The GPU cannot fall back on a dedicated framebuffer; it must share system memory for both the CPU and the GPU.

Because the bus width is also System Shared, there is no fixed memory pipeline geometry. The pixel rate and texture rate are fixed, but memory bandwidth is a moving target. At high resolutions, that dependency creates a direct link between platform memory quality and GPU performance. The database does not specify a memory configuration, so any measured bandwidth from one platform may not transfer to another.

Power and Cooling

The TDP of the Vega 11 Embedded is 25 W. This is a low power ceiling and is consistent with the IGP slot width, which indicates that the GPU is meant to live on the motherboard rather than in an expansion slot. The power connector field is “None,” so the GPU does not require external power cabling. The suggested PSU field is null, meaning the database records no PSU recommendation for this part.

For system integration, the absence of power connectors removes one category of hardware planning. The GPU is not a discrete add-in card; it occupies no slot and requires no separate power feed. Cooling planning is also tied to the 25 W envelope, although the fact pack does not specify a cooler or heatsink. The 14 nm process, 4,940 million transistor count, and 210 mm² die provide the physical context for that 25 W target: the part is designed for low-power, integrated environments rather than high-current board designs.

How It Compares

The nearestRivals list for this GPU is empty. No rival names, scores, or deltaPct values are present, so direct percentage comparisons against competing GPUs cannot be computed. The only ranking signal is the 50th percentile, which places the Vega 11 Embedded at the median of the database’s GPU population.

Against the GCN 3.0 IGP predecessor, the Vega 11 Embedded is listed as its successor, but no benchmark scores exist for either part. The architecture names differ — GCN 5.0 versus GCN 3.0 — yet the database provides no numeric performance delta between the two. Without workload entries, any statement about generational improvement would exceed the facts.

Against the Vega II IGP successor, the Vega 11 Embedded is listed as its predecessor, and again no deltaPct values are present. The production status is end-of-life, and the release date of 2018-02-12 is earlier than the successor’s listing, but the relative performance remains unspecified. The empty nearestRivals array and empty benchmarks array leave the Vega 11 Embedded with an unusual profile: it has a percentile position but no scored comparisons.

Who Should Consider It

The Vega 11 Embedded is a 25 W integrated part with System Shared memory, motherboard-dependent display outputs, and no power connectors. It is built for embedded and motherboard-integrated systems where a discrete GPU is not an option. The 50th percentile standing means it is not at the bottom of the database, but the empty benchmark list prevents any concrete performance ranking beyond that midpoint.

For high-resolution work, the data is not encouraging in a specific way: memory bandwidth is System Dependent, and the fixed pixel rate is 10.41 GPixel/s with only 8 ROPs. Those numbers set a hard fill-rate ceiling, while the shared memory arrangement makes high-resolution behavior dependent on the host platform. Any user expecting predictable high-resolution results would need more data than this record provides.

For users who need modern API support in a low-power integrated GPU, the feature set is present: DirectX 12 (12_1), OpenGL 4.6, and Vulkan 1.3 are all listed. The 300 MHz base clock and 1301 MHz boost clock define the operating range, while the FP32 figure of 1.832 TFLOPS gives a peak compute reference. Actual settings recommendations cannot be derived from the empty benchmark array, so a buyer should treat this part as a known-quantity embedded GPU with a median database rank and no quantified workload results.