AMD Radeon HD 6990M

AMD Radeon HD 6990M

AMD Radeon HD 6990M: A Retrospective on a Mobile Graphics Legend

April 2025

In an era where graphics cards with ray tracing and neural network technologies have become the norm, interest in classic GPUs from the past continues to grow. The AMD Radeon HD 6990M is one of these legends that, in the early 2010s, set performance standards for gaming laptops. Let's explore what made this model memorable, how it compares to modern solutions, and who might find it useful today.


1. Architecture and Key Features

TeraScale 2 Architecture: The Power Foundation

The HD 6990M is based on the TeraScale 2 architecture, which debuted in 2009. This second generation of AMD technology focused on enhancing shader block efficiency. The chip was manufactured using a 40nm process, which was standard in 2011 but today seems archaic compared to the 5nm and 3nm norms.

Unique Features of the Era

- AMD Eyefinity: Support for up to 6 displays simultaneously—a revolution for multi-monitor setups.

- DirectX 11 and OpenGL 4.1: Relevant APIs at the time offering compatibility with games from the late 2000s to early 2010s.

- PowerPlay: Optimization of power consumption based on load—a crucial feature for mobile GPUs.

Technologies such as ray tracing (RTX) or FidelityFX Super Resolution (FSR) were not present—they emerged a decade later. However, for its time, the HD 6990M was a flagship, offering 1120 stream processors and a clock speed of up to 715 MHz.


2. Memory: Speed vs. Volume

GDDR5 and Limited Capacity

The graphics card was equipped with 2GB of GDDR5 memory on a 256-bit bus. The effective memory clock speed was 3600 MHz, ensuring a bandwidth of 115.2 GB/s. For games like Battlefield 3 or Crysis 2, this was sufficient for comfortable gaming at high settings in 1080p.

Limitations in 2025

Today, 2GB of video memory is critically low even for indie projects. Modern AAA games require at least 6–8GB, and high-resolution textures will quickly consume the HD 6990M's resources. While GDDR5 was fast for its time, it falls short against GDDR6X and HBM2 in terms of energy efficiency and data density.


3. Gaming Performance: Nostalgia in Frames

1080p: Once a Powerhouse

In 2011–2013, the HD 6990M delivered 30–60 FPS in games on high settings:

- The Witcher 2: ~35 FPS (Ultra);

- Skyrim: ~45 FPS (High);

- Metro 2033: ~25 FPS (Medium).

Modern Titles: A Test of Endurance

By 2025, even in CS2 or Fortnite at low settings in 1080p, the card barely reaches 20–30 FPS. Resolutions of 1440p and 4K are out of reach—lacking both power and memory. Ray tracing is, of course, not supported.

Conclusion: The HD 6990M is suitable only for retro gaming or non-demanding indie games like Stardew Valley.


4. Professional Tasks: Modest Capabilities

OpenCL and Hardware Limitations

The card supports OpenCL 1.2, theoretically enabling its use for rendering or computations. However, the 2GB of memory and the lack of optimization for modern applications (Blender 4.0, DaVinci Resolve 2025) make it largely impractical for work.

Comparison with Modern Solutions

Even a budget Radeon RX 7600M (8GB GDDR6, RDNA 3 architecture) is 10–15 times faster in editing and 3D modeling tasks. NVIDIA's CUDA cores, of course, are unavailable—this is exclusive to the "green team."


5. Power Consumption and Heat Dissipation

TDP 100W: Not the Most Power-Hungry

For a mobile GPU from 2011, a TDP of 100W was acceptable, but it required a serious cooling system. Such solutions are not used in today's slim laptops, which emphasize energy efficiency (for instance, the RTX 4050 Mobile has a TDP of just 35–115W with dynamic management).

Cooling Recommendations

If you’re using the HD 6990M in a custom PC (via an MXM adapter), a case with good ventilation and at least 2–3 fans will be necessary. It’s advisable to replace the thermal paste annually—older chips tend to overheat.


6. Comparison with Competitors

2011 Rivals

- NVIDIA GeForce GTX 580M: 2GB GDDR5, 384 CUDA cores. The HD 6990M excelled in multi-display configurations but lagged in optimization for DirectX 11.

- AMD Radeon HD 6970M: A slightly weaker model with 960 stream processors—about 15% less capable.

In 2025

Comparing the HD 6990M with modern GPUs is pointless. Even a budget Intel Arc A380 (2023) outperforms it by 3–4 times in performance while supporting HDMI 2.1 and AV1 encoding.


7. Practical Tips

Power Supply and Compatibility

- Minimum PSU: 450W (with headroom for older systems).

- Compatibility: only motherboards with PCIe 2.0 x16. For laptops—only models from 2010-2013 with an MXM slot.

Drivers: The Point of No Return

AMD officially discontinued driver support in 2018. A community of enthusiasts releases custom versions, but stability isn’t guaranteed. For Windows 10/11, it is better to use drivers from 2017.


8. Pros and Cons

Pros:

- Historical value: one of AMD’s first mobile flagships.

- Eyefinity support for multi-display setups.

- Low price on the second-hand market ($30–50).

Cons:

- Outdated architecture and lack of video memory.

- Absence of support for modern APIs (DirectX 12 Ultimate, Vulkan 1.3).

- High power consumption for mobile tasks.


9. Final Verdict: Who Is the HD 6990M for?

This graphics card is an artifact of an era when Crysis 3 was the benchmark for graphics. In 2025, it can be recommended for:

1. Collectors — as part of the history of gaming hardware.

2. Owners of Older Laptops — for upgrades while maintaining authenticity.

3. Retro Gaming Enthusiasts — to play Dragon Age: Origins or Mass Effect 2 as they were intended.

For all other scenarios—ranging from gaming to work—it's better to choose modern budget GPUs like the Radeon RX 7600M or GeForce RTX 3050 Mobile. The HD 6990M will remain in the hearts of gearheads as a symbol of the struggle for mobile power, but its time has irrevocably passed.

Basic

Label Name
AMD
Platform
Mobile
Launch Date
July 2011
Model Name
Radeon HD 6990M
Generation
Vancouver
Bus Interface
MXM-B (3.0)
Transistors
1,700 million
Compute Units
14
TMUs
?
Texture Mapping Units (TMUs) serve as components of the GPU, which are capable of rotating, scaling, and distorting binary images, and then placing them as textures onto any plane of a given 3D model. This process is called texture mapping.
56
Foundry
TSMC
Process Size
40 nm
Architecture
TeraScale 2

Memory Specifications

Memory Size
2GB
Memory Type
GDDR5
Memory Bus
?
The memory bus width refers to the number of bits of data that the video memory can transfer within a single clock cycle. The larger the bus width, the greater the amount of data that can be transmitted instantaneously, making it one of the crucial parameters of video memory. The memory bandwidth is calculated as: Memory Bandwidth = Memory Frequency x Memory Bus Width / 8. Therefore, when the memory frequencies are similar, the memory bus width will determine the size of the memory bandwidth.
256bit
Memory Clock
900MHz
Bandwidth
?
Memory bandwidth refers to the data transfer rate between the graphics chip and the video memory. It is measured in bytes per second, and the formula to calculate it is: memory bandwidth = working frequency × memory bus width / 8 bits.
115.2 GB/s

Theoretical Performance

Pixel Rate
?
Pixel fill rate refers to the number of pixels a graphics processing unit (GPU) can render per second, measured in MPixels/s (million pixels per second) or GPixels/s (billion pixels per second). It is the most commonly used metric to evaluate the pixel processing performance of a graphics card.
22.88 GPixel/s
Texture Rate
?
Texture fill rate refers to the number of texture map elements (texels) that a GPU can map to pixels in a single second.
40.04 GTexel/s
FP32 (float)
?
An important metric for measuring GPU performance is floating-point computing capability. Single-precision floating-point numbers (32-bit) are used for common multimedia and graphics processing tasks, while double-precision floating-point numbers (64-bit) are required for scientific computing that demands a wide numeric range and high accuracy. Half-precision floating-point numbers (16-bit) are used for applications like machine learning, where lower precision is acceptable.
1.57 TFLOPS

Miscellaneous

Shading Units
?
The most fundamental processing unit is the Streaming Processor (SP), where specific instructions and tasks are executed. GPUs perform parallel computing, which means multiple SPs work simultaneously to process tasks.
1120
L1 Cache
8 KB (per CU)
L2 Cache
512KB
TDP
100W
Vulkan Version
?
Vulkan is a cross-platform graphics and compute API by Khronos Group, offering high performance and low CPU overhead. It lets developers control the GPU directly, reduces rendering overhead, and supports multi-threading and multi-core processors.
N/A
OpenCL Version
1.2
OpenGL
4.4
DirectX
11.2 (11_0)
Power Connectors
None
Shader Model
5.0
ROPs
?
The Raster Operations Pipeline (ROPs) is primarily responsible for handling lighting and reflection calculations in games, as well as managing effects like anti-aliasing (AA), high resolution, smoke, and fire. The more demanding the anti-aliasing and lighting effects in a game, the higher the performance requirements for the ROPs; otherwise, it may result in a sharp drop in frame rate.
32

Benchmarks

FP32 (float)
Score
1.57 TFLOPS

Compared to Other GPU

FP32 (float) / TFLOPS
1.671 +6.4%
1.618 +3.1%
1.508 -3.9%
1.457 -7.2%