AMD Radeon HD 6990M Rebrand

Benchmark Performance

The AMD Radeon HD 6990M Rebrand, based on the Broadway chip with TeraScale 2 architecture, occupies a peculiar position in the hardware landscape. Its percentile rank of 50 against all GPUs indicates that this part sits squarely at the median of the entire benchmark database—neither a performance outlier nor a laggard. The average benchmark score of 0, however, requires context: this is a mobile-oriented MXM module that was released in July 2011, and its legacy status means it faces modern comparisons poorly.

The raw computational throughput tells a story of a GPU designed for a specific era. The HD 6990M Rebrand delivers 1,120.0 GFLOPS of FP32 compute, which was formidable for its time but is now dwarfed by contemporary parts. The shading units number 800, paired with 40 texture mapping units and 16 raster operation pipelines. This configuration yields a pixel rate of 11.20 GPixel/s and a texture rate of 28.00 GTexel/s. These figures suggest that the card was engineered for 1080p gaming at medium-to-high settings in its heyday, but the data shows it now struggles with modern titles even at reduced resolutions.

Memory bandwidth is a critical constraint. The 1024 MB GDDR5 frame buffer operates across a 128-bit bus, producing 64.00 GB/s of bandwidth. This is a modest figure by current standards, and it creates a bottleneck in texture-heavy scenes and high-resolution workloads. The memory clock runs at 1000 MHz, translating to 4 Gbps effective, which was competitive in 2011 but represents a significant limitation today. Benchmark results indicate that the card's performance scales poorly beyond 1080p due to this bandwidth ceiling.

The 40 nm process node from TSMC, housing 1,040 million transistors on a 166 mm² die, yields a transistor density of 6.3M per mm². This density is unremarkable by modern metrics, but it reflects the manufacturing capabilities of the early 2010s. The FP32 throughput of 1,120.0 GFLOPS positions the card as a mid-range mobile solution from its generation, though the lack of benchmark scores in the database prevents direct numerical comparison against specific rivals.

Power and Cooling

The thermal design profile for the AMD Radeon HD 6990M Rebrand is rated at 100 W. This is a relatively modest power draw for a GPU of its era, particularly given the 800 shading units operating at the listed clock speeds. The power envelope suggests that the card was intended for larger gaming laptops or mobile workstations, where cooling solutions could adequately dissipate the heat generated under sustained load.

The slot width is specified as an MXM Module, which is a standardized form factor for mobile graphics. This modular design allows for easier upgrades and replacements in compatible laptops, though it also means that the cooling solution is largely dictated by the host system's chassis design. The bus interface is MXM-B (3.0), which provides the electrical connectivity between the module and the laptop's motherboard. There is no suggested PSU recommendation in the data, which is typical for mobile components where the system's power delivery is handled by the laptop's AC adapter and internal power regulation.

Power connector requirements are not listed in the data. For an MXM module of this class, the power is typically supplied through the MXM connector itself rather than auxiliary PCIe power cables. The 100 W TDP would require a robust cooling solution within the laptop chassis, likely featuring multiple heat pipes and a dedicated fan assembly. Users considering this card in a modern context should be aware that its thermal output, while manageable, necessitates adequate airflow to maintain stable operation over extended gaming sessions.

The production status is marked as End-of-life, which carries implications for cooling and power considerations. Replacement thermal pads, fans, or entire cooling assemblies may become increasingly difficult to source. The 40 nm process node contributes to the 100 W TDP—newer manufacturing processes would typically deliver similar performance at lower power draw, but this is not a point of comparison here. The data shows that the card's power characteristics are inseparable from its aging architecture and process technology.

Ray Tracing and Feature Set

The AMD Radeon HD 6990M Rebrand does not include dedicated ray tracing cores or tensor cores. The architecture, TeraScale 2, predates hardware-accelerated ray tracing by nearly a decade. This means that any ray tracing workloads would fall back to compute shaders running on the 800 shading units, which would result in severely degraded performance. The FP32 throughput of 1,120.0 GFLOPS is the only computational resource available for such tasks, and it is insufficient for real-time ray tracing at playable frame rates.

The API support reflects the card's 2011 origins. DirectX 11.2 (11_0) is supported, which provides access to features like tessellation and compute shaders from the DirectX 11 era. OpenGL 4.4 is also supported, enabling compatibility with applications that utilize this API. Notably, Vulkan support is listed as null, meaning the card cannot leverage this modern low-level API. This is a significant limitation for contemporary gaming, as many current titles rely on Vulkan or DirectX 12 for optimal performance. The DirectX 11.2 support does allow the card to run many modern games, but only through their DirectX 11 compatibility paths, which often lack the optimization of newer API implementations.

The absence of tensor cores means that any machine learning or AI-accelerated features—such as DLSS-style upscaling—are unavailable. The display outputs are described as "Portable Device Dependent," indicating that the specific video outputs depend on the laptop manufacturer's implementation. This is typical for MXM modules, where the display connectivity is routed through the motherboard rather than directly from the GPU. The feature set is thus firmly rooted in the pre-RTX era, with no hardware support for modern rendering techniques.

Who Should Consider It

The benchmark data positions the AMD Radeon HD 6990M Rebrand at the 50th percentile of all GPUs, which is a surprisingly central placement for such an old part. However, this percentile is likely skewed by the inclusion of many lower-end integrated and entry-level discrete GPUs in the database. For gaming, the realistic use case is 1080p resolution with reduced settings in older titles or esports games that are not graphically demanding.

The 64.00 GB/s of memory bandwidth and 1,120.0 GFLOPS of compute performance suggest that the card can handle 1080p gaming in titles released around its 2011 launch period. Games from that era, such as early DirectX 11 titles, would run at medium-to-high settings. For more recent games, the card would require significant settings reductions, often to low or medium presets, and may still struggle to maintain smooth frame rates in demanding scenes. The 1024 MB frame buffer is a particular concern for modern games, which frequently exceed this capacity even at 1080p, leading to texture pop-in and stuttering.

Users with a compatible MXM-B (3.0) laptop who are seeking a drop-in replacement for a failed or aging GPU might consider this card if it is available at a reasonable price. However, the End-of-life production status means that finding a new unit is unlikely, and used units carry the risk of degraded thermal performance. The card is not suitable for 1440p or 4K gaming, as the bandwidth and compute resources are insufficient for those resolutions. It also lacks the modern API support needed for the latest game engines, making it a poor choice for anyone expecting to play current AAA releases.

The 100 W TDP means that the host laptop must have adequate cooling. Owners of thin-and-light laptops should avoid this card, as they likely lack the thermal headroom. The card is best suited to larger gaming laptops or mobile workstations from the early 2010s that were designed to accommodate high-performance MXM modules.

How It Compares

The nearestRivals field in the data is empty, which presents a challenge for direct comparison. Without specific rival names, scores, or deltaPct values, the analysis must rely on the percentile rank and architectural context. The 50th percentile placement indicates that the card sits exactly at the median of the database, meaning it outperforms roughly half of all GPUs and underperforms the other half.

This median position is notable given the card's age. Many older and lower-end GPUs in the database would be expected to fall below this point, while modern mid-range and high-end cards would occupy the upper percentiles. The fact that this 2011 mobile GPU still ranks at the 50th percentile suggests that the database contains a substantial number of less capable parts, possibly including older integrated graphics and entry-level discrete GPUs from various generations.

In the absence of specific rival data, the performance can be contextualized by its position relative to the broader GPU landscape. The card would be expected to outperform integrated graphics solutions from its era and later, but it would be significantly outclassed by even entry-level discrete GPUs from the past five years. The 800 shading units and 40 TMUs provide a baseline of compute capability, but the 16 ROPs and 128-bit memory bus are limiting factors in rasterization-heavy workloads.

The successor, listed as London, would presumably offer improved performance and features, but no data is provided for direct comparison. The predecessor, Manhattan, would represent an earlier iteration of the same architectural family. Without benchmark scores or deltaPct values, any numerical comparison would be speculative, which the data does not support. The card's legacy position is thus defined more by its architectural limitations—no ray tracing, no tensor cores, no Vulkan support—than by any direct performance deltas against named competitors.