Intel HD Graphics 6EU Mobile

Intel HD Graphics 6EU Mobile is an integrated graphics processor built by Intel around the Generation 7.0 architecture, using the Ivy Bridge GT1 chip and the HD Graphics-M (Ivy Bridge) generation. The 22 nm Intel process produces a 94 mm² die containing 392 million transistors, for a transistor density of 4.2M / mm². The release date is September 30, 2012, and the production status is end-of-life. It connects through a Ring Bus interface, uses the IGP slot width, and lists display outputs as portable-device dependent. No series name, codename, predecessor, successor, or launch MSRP is present in the data.

Memory Subsystem

VRAM size is listed as System Shared, and memory type is also System Shared. Bus width is likewise System Shared, while bandwidth is listed as System Dependent. The memory clock is System Shared rather than a fixed rate. This means the part does not define a dedicated graphics memory pool; the framebuffer is carved from the host system’s memory, and the available memory path depends on the platform around it.

Because bandwidth is System Dependent, high-resolution performance cannot be characterized with a single number. A high-resolution workload requires more data to be moved into and out of the framebuffer. With a System Shared design, the same memory also serves the host processor, so the effective bandwidth available to the GPU is a platform-specific variable rather than a fixed property of the graphics die. The system memory type, the memory controller, and the rest of the platform determine how much bandwidth the GPU can actually use.

The absence of a fixed memory bus width reinforces that point. A discrete part with a defined bus width can be evaluated for worst-case memory throughput; this part cannot, because the width is not owned by the GPU. In practical terms, the memory subsystem will scale with the host system. A higher-capability platform with a faster memory configuration will provide more headroom; a weaker platform will constrain the GPU. At high resolutions, the System Shared and System Dependent fields are the limiting variables.

Ray Tracing and Feature Set

The data lists no RT core count and no tensor core count. That places this design outside any hardware-accelerated ray tracing or tensor acceleration category in the database. Any ray-traced or tensor-oriented workloads would need to rely on software paths, which the listed feature set does not directly document.

The API support is DirectX 11.1 (11_0), OpenGL 4.0, and Vulkan 1.0. These three entries define the software interface surface. DirectX 11.1 (11_0) indicates the feature level is 11_0. OpenGL 4.0 and Vulkan 1.0 provide two further API levels, both listed alongside the DirectX entry. This is the complete API listing for the part.

Because the data includes no RT or tensor hardware counts, the feature set is best described as conventional rasterization-focused. The available throughput resources are small: 48 shading units, 6 texture mapping units, and 1 ROP. The API list is more expansive in software terms than the raw hardware counts might suggest, but the hardware still determines how much useful work can be pushed through those APIs.

Benchmark Performance

The benchmark array is empty. No workload scores, no frame-rate measurements, and no comparison deltas are supplied. The only summary benchmark figures are an average benchmark score of 0 and a percentile rank of 50 among all GPUs in the database.

The average score of 0 is the sole numeric performance summary, but it does not carry any rival-relative meaning because the nearestRivals list is also empty. There are no rival names, no rival scores, and no percentage differences in the data. Consequently, no exact deltaPct statements can be made. The 50th percentile placement is a global position across the entire GPU database, not a head-to-head result against a specific competitor.

The fixed throughput figures provide the strongest numeric anchor for performance. The pixel rate is 1.000 GPixel/s, the texture rate is 6.000 GTexel/s, and FP32 arithmetic is 96.00 GFLOPS. These figures align with the small execution resources: 48 shading units, 6 TMUs, and 1 ROP. With one ROP, pixel output is funneled through a single raster output unit, so pixel-fill-bound scenes will be especially constrained. With six TMUs, texture work is capped at 6.000 GTexel/s. The 96.00 GFLOPS FP32 figure represents the maximum shader arithmetic rate for the listed clock configuration.

The clock table also supplies a 350 MHz base and a 1000 MHz boost, with no separate game clock listed. The boost clock provides the upper operating point for the shading resources, while the base clock establishes the lower bound. Because there are no benchmark scores, the practical effect of these clocks can only be inferred from the raw rates, not from measured application results.

Power and Cooling

The TDP is 45 W. This is the only power figure in the data. The part is an IGP, with slot width listed as IGP, so it is not configured as a discrete expansion card.

No power connector is listed. The suggested PSU field is null. These two missing values indicate that the product does not carry a discrete graphics power connector requirement and does not come with a database-recommended power supply. Power delivery is therefore a matter for the host platform rather than a separate graphics card power input.

Cooling data is also absent. No cooler size, cooler type, or thermal solution is listed. The 45 W TDP is the only thermal design anchor. Since this is an integrated part, the surrounding platform must handle that 45 W within its own thermal envelope. The absence of a suggested PSU and power connectors makes the integrated nature clear: this is not a card that plugs into a slot and draws through its own power connectors.

How It Compares

The nearestRivals list contains no entries. There is no rival product name, no rival score, and no percentage delta available for comparison. Because the comparison set is empty, a per-rival section cannot be assembled from the data.

The part’s only positional metric is the 50th percentile among all GPUs. That places it in the middle of the database distribution, but it does not identify a specific competitor or class of competitors. A 50th percentile rank is not the same as a 50% advantage or disadvantage against any named GPU. It is simply the median point in the global list.

The predecessor and successor fields are also null. That removes another layer of positioning: the product is not tied to an earlier or later model in the data. The architecture generation is Generation 7.0, and the generation descriptor is HD Graphics-M (Ivy Bridge), but no immediate family neighbors are recorded. Within the database, then, the only supported comparison is the global percentile placement.

Who Should Consider It

Because the benchmark array is empty, any recommendation must rest on feature support and raw throughput rates rather than measured scores. The part is most appropriate for workloads that fit inside its listed API set: DirectX 11.1 (11_0), OpenGL 4.0, and Vulkan 1.0. Software written for those interfaces is within the documented feature scope.

High-resolution use is the weakest fit. The memory path is System Shared and System Dependent, so high-resolution framebuffer traffic cannot be guaranteed. The pixel rate is also 1.000 GPixel/s, and the ROP count is 1. That combination makes high-resolution rendering especially difficult, because the pixel output stage is very small. Modest resolutions and visually simple scenes are the better fit, since those workloads require less framebuffer data and produce fewer pixels to fill.

The 350 MHz base clock and 1000 MHz boost clock define how fast the 48 shading units can operate. The 96.00 GFLOPS FP32 figure is the resulting arithmetic ceiling. The 6.000 GTexel/s texture rate places a separate limit on texture-heavy work. Any workload that needs substantial texturing will run into that 6 TMU structure.

The product is end-of-life, so long-term availability is not part of the current data. Its relevance is limited to systems already using this integrated part or applications that explicitly need the listed API levels. It should not be selected for ray-traced workloads, since the data lists no RT cores and no tensor cores. It is a small, integrated rendering solution whose measured position in the database is a 50th percentile rank with an empty benchmark history.