AMD Radeon HD 7690M

The AMD Radeon HD 7690M is a mobile graphics processor from the London generation, built on the TeraScale 2 architecture using a 40 nm process at TSMC. This end-of-life part, released in late 2011, was designed for the MXM-A form factor and targets the mainstream notebook segment. The data indicates a hardware configuration centered on 480 shading units, 24 texture mapping units, and 16 ROPs, paired with 1024 MB of DDR3 memory on a 128-bit bus.

Benchmark Performance

The benchmark data for the HD 7690M shows no recorded average benchmark score, with a percentile ranking of 50 among all GPUs. This places it in the absolute middle of the performance distribution, though the lack of a specific score means the percentile is derived from the hardware's structural capabilities rather than direct testing. The theoretical compute throughput of 696.0 GFLOPS (FP32) and a texture fill rate of 17.40 GTexel/s provide the raw performance envelope. The pixel rate stands at 11.60 GPixel/s, which is consistent with a 16-ROP design operating at the given clock frequencies.

The memory subsystem contributes a bandwidth of 28.80 GB/s, derived from a 900 MHz memory clock (1800 Mbps effective) across the 128-bit interface. This bandwidth is a critical constraint for the architecture, as the DDR3 memory type is slower than GDDR5 solutions found in higher-tier parts. In practical terms, the data suggests the HD 7690M will be heavily limited by memory bandwidth in texture-heavy scenes, despite having a reasonable compute-to-texture ratio. The 6.1 million transistors per square millimeter density reflects the mature 40 nm process, with a total of 716 million transistors on a 118 mm² die.

Given the absence of direct rival scores, the positional analysis relies on the percentile field. A 50th percentile ranking indicates this GPU sits at the median of all GPUs tracked in the database. This implies it outperforms roughly half of all known graphics processors while trailing the other half, a positioning typical of a mainstream mobile part from its era. The lack of a benchmark score suggests the data collection process did not yield a representative performance figure, so the percentile is the primary quantitative anchor. The FP32 throughput of 696.0 GFLOPS, when contextualized against the 480 shading units, yields a per-shader efficiency that is characteristic of TeraScale 2 designs.

Ray Tracing and Feature Set

The HD 7690M does not include dedicated ray tracing cores or tensor cores, as these hardware units are absent from the specification. The architecture relies on the TeraScale 2 shader array for all compute tasks, including any ray tracing workloads handled in software. This means ray tracing performance, if attempted, would be entirely dependent on the 480 shading units executing intersection and traversal routines, which is computationally expensive and would result in poor frame rates. The data indicates no hardware acceleration for this feature, making it unsuitable for any modern ray-traced content.

The API support is limited to DirectX 11.2 (feature level 11_0) and OpenGL 4.4. There is no Vulkan support listed, which restricts the GPU to older graphics APIs. DirectX 11.2 provides access to tessellation and compute shaders, which are the primary advanced features available. The absence of Vulkan means the GPU cannot leverage modern cross-platform APIs that offer lower overhead and better multi-threading. OpenGL 4.4 support is adequate for legacy applications but lacks the extensions found in newer versions.

The memory configuration of 1024 MB DDR3 on a 128-bit bus is a notable limitation for feature-rich workloads. Textures and geometry buffers must fit within this capacity, and the 28.80 GB/s bandwidth will bottleneck large data transfers. The display outputs are listed as "Portable Device Dependent," meaning the connectivity options vary by laptop manufacturer. The MXM-A (3.0) bus interface is a standardized module format, allowing for potential upgrades in compatible systems, though the 20 W TDP and lack of power connectors indicate a power-efficient design intended for thin notebooks.

Who Should Consider It

The HD 7690M is suited for users running older games at lower resolutions and settings, given its 50th percentile ranking and modest compute resources. For 720p resolution, the 696.0 GFLOPS of FP32 compute and 17.40 GTexel/s texture rate can handle esports titles and games from the early 2010s at medium to high settings. The 1024 MB memory is sufficient for these workloads, though texture-heavy games may require reduced detail levels to stay within the 28.80 GB/s bandwidth ceiling.

At 1080p, the GPU will struggle with modern titles, but older or less demanding games can run at low to medium settings. The 11.60 GPixel/s pixel rate limits fill-rate-heavy effects such as high-resolution shadows and post-processing. Users should expect to disable anti-aliasing and reduce shadow quality to maintain playable frame rates. The 20 W TDP makes this a good fit for basic productivity and light gaming, but it is not recommended for any form of high-refresh-rate or high-fidelity gaming.

This GPU is not for users interested in ray tracing, as the absence of RT cores and lack of Vulkan support preclude any modern rendering techniques. It is also unsuitable for compute-intensive tasks like machine learning or video encoding, which would require tensor cores or dedicated hardware. The end-of-life status means no driver optimizations for newer titles are expected, further limiting its relevance. The primary audience is retro gaming enthusiasts or users with legacy software that runs well within the DirectX 11.2 feature set.

FAQ

Q: What is the memory configuration of the AMD Radeon HD 7690M?

A: The GPU comes with 1024 MB of DDR3 memory on a 128-bit bus, operating at 900 MHz (1800 Mbps effective), providing a bandwidth of 28.80 GB/s.

Q: Does the HD 7690M support hardware ray tracing?

A: No, the specification lists no ray tracing cores. The TeraScale 2 architecture relies on its 480 shading units for all compute, making ray tracing possible only in software with very low performance.

Q: What is the thermal design power of this mobile GPU?

A: The TDP is rated at 20 W, with no power connectors required, as it draws power directly from the MXM-A (3.0) slot.

Q: Which graphics APIs are supported?

A: The GPU supports DirectX 11.2 (feature level 11_0) and OpenGL 4.4. Vulkan is not listed as supported.

Q: What is the pixel fill rate?

A: The pixel rate is 11.60 GPixel/s, derived from 16 ROPs at a given clock speed, while the texture rate is 17.40 GTexel/s from 24 TMUs.

Q: What is the process node and transistor count?

A: The chip is fabricated on a 40 nm process at TSMC, containing 716 million transistors on a die size of 118 mm², yielding a density of 6.1 million transistors per square millimeter.

How It Compares

The HD 7690M has no direct rivals listed in the nearestRivals field, which means the comparison must be framed against the broader GPU landscape. The 50th percentile ranking places it exactly at the median of all GPUs in the database. This indicates that half of all tracked graphics processors are faster, and half are slower, a typical position for a mainstream mobile part from its generation.

Without rival names or scores, the HD 7690M's position is defined by its absolute specifications. The 696.0 GFLOPS FP32 compute is a mid-pack figure for 2011-era mobile GPUs, while the 28.80 GB/s bandwidth is on the lower end due to the DDR3 memory choice. The absence of Vulkan support further separates it from newer integrated solutions that offer modern API compatibility, even if their raw compute is lower.

The 20 W TDP is a competitive advantage for thin-and-light laptops, but it caps performance under sustained load. The predecessor "Vancouver" and successor "Solar System" are mentioned in the lineage, but no comparative data is provided. The end-of-life production status means this GPU is now a legacy part, likely outperformed by modern integrated graphics in most scenarios, though its dedicated 480 shading units can still handle basic 3D acceleration tasks more efficiently than early iGPUs.