ATI Radeon HD 5690

How It Compares

The ATI Radeon HD 5690 occupies a distinctly mid-pack position in the benchmark database, landing at the 50th percentile among all GPUs. This is a card that defined the middle of the road when it launched in early 2011, and the data reflects a product that was never designed to lead, but rather to populate the vast mainstream segment with competent, unremarkable performance.

No nearest rivals are listed in the fact pack for this GPU. That absence is itself telling: the HD 5690 sits in a performance tier where competitive pressure was diffuse rather than concentrated. Without direct comparison points in the dataset, the card's 50th percentile ranking becomes the primary reference — exactly half of all GPUs in the database perform better, and half perform worse. This is the statistical definition of average, though the architecture behind that average is worth examining.

The HD 5690 is built on AMD's TeraScale 2 architecture, using the Redwood chip fabricated on TSMC's 40 nm process. The die contains 627 million transistors packed into a 104 mm² area, yielding a transistor density of 6.0 million per square millimeter. For its era, this represented a mature, well-understood manufacturing node and a chip design that prioritized yield and cost efficiency over raw complexity. The card is a single-slot solution measuring 168 mm (6.6 inches) in length, making it physically unobtrusive in a chassis.

Ray Tracing and Feature Set

The Radeon HD 5690 does not include dedicated ray tracing cores or tensor cores — these hardware units are entirely absent from the specification. This is a fundamental architectural limitation: the TeraScale 2 design predates hardware-accelerated ray tracing by nearly a decade, and the card's rendering pipeline is purely rasterization-based. The absence of RT cores means any ray-traced workload would fall back to compute shaders running on the 400 shading units, with predictably poor results.

The feature set is anchored to its API support. The card exposes DirectX 11.2 with the 11_0 feature level, which means it supports the core DirectX 11 feature set: tessellation, compute shaders, and multithreaded rendering. OpenGL 4.4 is also supported, providing a reasonably modern graphics API surface for its generation. No Vulkan support is listed, which effectively locks the card out of modern cross-platform titles that rely on Vulkan for rendering.

Display connectivity is straightforward: one DVI port, one HDMI 1.3a port, and one VGA port. This trio of outputs covers the standard monitor configurations of the early 2010s, with HDMI 1.3a capping the card at 1080p output for consumer displays. The card connects via PCIe 2.0 x16, which provides adequate bandwidth for its 32.00 GB/s memory subsystem. The absence of tensor cores also means no AI-accelerated features like DLSS — the card has no hardware path for machine learning-based upscaling, leaving it dependent on raw resolution rendering.

Benchmark Performance

The benchmark data for the HD 5690 is sparse — the benchmark scores array is empty, and the average benchmark score is listed as zero. This makes direct performance quantification impossible from the dataset alone. However, the specifications provide a basis for understanding the card's computational throughput. The pixel rate is 6.200 GPixel/s, the texture rate is 15.50 GTexel/s, and the FP32 compute is 620.0 GFLOPS. These figures place the card in a performance class that was competitive for 720p gaming in its era, but would struggle with 1080p in demanding titles.

The lack of nearestRivals data means we cannot cite specific percentage deltas against competing products. What the specifications do tell us is that the 400 shading units operating at the listed clock rates (though base and boost clocks are absent from the fact pack, the memory clock is 1000 MHz, translating to 2 Gbps effective) produce a compute throughput that is entirely consistent with a mid-range 2011 product. The 8 ROPs are a notable bottleneck — this is a low count even for the era, limiting fill-rate-heavy workloads like high-resolution anti-aliasing. The 20 texture mapping units are better proportioned relative to the shading units, but the overall balance skews toward compute rather than pixel throughput.

The 50th percentile ranking, combined with the absence of direct rival data, suggests a card that was neither embarrassing nor impressive. It likely traded blows with other mainstream GPUs of its generation, winning some workloads and losing others, without ever establishing a decisive advantage. The TeraScale 2 architecture was known for decent DirectX 11 tessellation performance and solid driver maturity, but the HD 5690's modest shader count and narrow memory bus would have limited its ceiling.

FAQ

Q: Does the ATI Radeon HD 5690 support hardware ray tracing?

A: No. The card has no ray tracing cores or tensor cores listed in its specifications. Ray-traced workloads would have to run on the 400 shading units via compute shaders, which is not a practical path for real-time ray tracing.

Q: What is the memory configuration of the HD 5690?

A: The card ships with 1024 MB (1 GB) of GDDR3 memory on a 128-bit bus, providing 32.00 GB/s of memory bandwidth. The memory clock is 1000 MHz, with an effective data rate of 2 Gbps.

Q: What APIs are supported by this GPU?

A: The HD 5690 supports DirectX 11.2 (with the 11_0 feature level) and OpenGL 4.4. Vulkan is not supported, which limits compatibility with modern titles that require it.

Q: What is the power consumption of the HD 5690?

A: The card has a TDP of 64 W and a suggested power supply rating of 250 W. It is a single-slot card and does not list any power connectors, suggesting it draws power entirely from the PCIe slot.

Q: What display outputs does the HD 5690 provide?

A: The card offers one DVI port, one HDMI 1.3a port, and one VGA port. This covers standard analog and digital connections but lacks DisplayPort support.

Q: What is the production status of this GPU?

A: The HD 5690 is end-of-life. It was released on January 31, 2011, and has been succeeded by the Northern Islands generation, with the Radeon R700 series as its predecessor.

Memory Subsystem

The memory subsystem of the HD 5690 is a study in deliberate conservatism. The card ships with 1024 MB of GDDR3 memory, which was the standard capacity for mid-range GPUs in 2011. The memory type is GDDR3 rather than the faster GDDR5 that was increasingly common in higher-tier products of the same era — this is a significant differentiator, as GDDR5 typically offered substantially higher bandwidth per pin. The memory clock runs at 1000 MHz, translating to 2 Gbps effective data rate, and the bus width is 128 bits.

The resulting memory bandwidth is 32.00 GB/s. This is the critical bottleneck for the card's performance at higher resolutions. A 128-bit bus with GDDR3 simply cannot feed the 400 shading units enough data to keep them fully occupied in texture-heavy or resolution-intensive workloads. At 1080p, the card would likely run into bandwidth limitations in modern games that demand large texture pools and high-resolution assets. The 1024 MB capacity, while adequate for its time, would also prove restrictive as game memory requirements grew beyond that threshold.

For 720p gaming, the 32.00 GB/s bandwidth is more palatable — lower resolution means fewer pixels to fill and less texture data streaming per frame. The 6.200 GPixel/s pixel rate and 15.50 GTexel/s texture rate are better matched to this lower workload. However, the 8 ROPs remain a constraint even at 720p when anti-aliasing is enabled, as the fill-rate cost of MSAA multiplies with the number of samples. The memory subsystem is therefore best described as adequate for its intended mainstream segment but without headroom for future demands. The GDDR3 choice, in particular, marks this as a cost-optimized design rather than a performance-oriented one — a card that would hold up for esports titles and older games, but would quickly show its age with newer, more memory-hungry releases.