AMD Radeon RX 7990 XTX

The AMD Radeon RX 7990 XTX is a flagship offering within the Radeon RX 7000 series, built on the RDNA 3.0 architecture and the Navi 31 chip. With a 5 nm process from TSMC, it packs 57,700 million transistors into a 529 mm² die, resulting in a transistor density of 109.1M per mm². The card is positioned as an active production model, bridging the gap between the previous Navi II generation and the upcoming Navi IV, and this analysis draws solely on the provided data to assess its standing.

Benchmark Performance

The fact pack lists an average benchmark score of 0 and a percentile rank of 50 among all GPUs, with no specific rival scores or delta percentages provided. This creates an unusual situation where the card’s raw benchmark numbers are absent from the dataset. Consequently, any interpretation of its performance must rely on the architectural and specification data available. The FP32 compute throughput is 88.45 TFLOPS, while the FP16 rate matches at 88.45 TFLOPS with a 1:1 ratio, indicating that the card does not rely on packed math to boost half-precision performance. The pixel rate stands at 691.0 GPixel/s, and the texture rate is 1,382.0 GTexel/s, which are derived from the 192 ROPs and 384 TMUs respectively.

Given the absence of direct rival scores, the percentile field suggests that the RX 7990 XTX sits at the median of all GPUs tracked, which is a stark contrast to what one might expect from a flagship part. Without numeric deltas, the analysis must focus on the internal consistency of the specs. The shading units total 6,144, which, when combined with the 96 RT cores, forms the core compute architecture. The boost clock of 3599 MHz and game clock of 3300 MHz are high figures, but without benchmark scores, translating these into relative performance is impossible. The data indicates a 50th percentile placement, meaning that half of the GPUs in the database score higher and half score lower, but the lack of an actual score prevents a more granular comparison.

In practical terms, the FP32 throughput of 88.45 TFLOPS is a substantial number, but the benchmark data does not validate how this translates into real-world gaming or compute workloads. The texture rate of 1,382.0 GTexel/s and pixel rate of 691.0 GPixel/s suggest strong fill-rate capabilities, which typically benefit high-resolution rendering. However, the median percentile rank is a cautionary signal, it implies that, based on the available database, this card does not outperform the majority of GPUs, which is surprising for a top-tier model. The lack of nearestRivals data further complicates any effort to contextualize these numbers. The conclusion is that the benchmark performance cannot be quantified relative to competitors from this fact pack alone, and the 50th percentile stands as the only comparative metric.

Ray Tracing and Feature Set

The RX 7990 XTX includes 96 RT cores dedicated to ray tracing workloads. This is a hardware-level feature that accelerates BVH traversal and ray intersection calculations, which are fundamental to modern ray-traced effects in games. The API support is comprehensive: it supports DirectX 12 Ultimate with a feature level of 12_2, which encompasses ray tracing, mesh shaders, and variable rate shading. It also supports Vulkan 1.4 and OpenGL 4.6, covering a wide range of current and legacy applications. The DirectX 12 Ultimate compliance is critical for games that use DXR (DirectX Raytracing) or similar features, ensuring that the card can handle the latest rendering techniques.

There are no tensor cores listed in the fact pack, which means the card does not have dedicated AI acceleration hardware akin to what some competitors use for DLSS or other AI-based upscaling. This is a notable absence, as tensor cores are often leveraged for performance-enhancing features. The RT cores, however, are present and their count of 96 is a specific hardware resource. The card does not appear to have any proprietary upscaling technology mentioned in the data, so users must rely on standard rendering or vendor-agnostic solutions. The feature set is rounded out by the display outputs, which include 1x HDMI 2.1a, 2x DisplayPort 2.1, and 1x USB Type-C, allowing for high-bandwidth connections to modern monitors.

The combination of RT cores and DirectX 12 Ultimate support means the card is technically capable of ray-traced gaming, but the performance in such workloads is not quantified by any benchmark scores. The absence of tensor cores suggests that any AI-driven features are not hardware-accelerated, which could be a limitation in games that heavily rely on such technologies. The Vulkan 1.4 support is forward-looking and ensures compatibility with newer titles that adopt this API. Overall, the ray tracing feature set is present and functional, but without performance data, its efficiency remains unverified.

Memory Subsystem

The memory configuration is one of the most substantial aspects of this GPU. It comes with 24 GB of GDDR6 memory, which is a large capacity suited for high-resolution textures and large datasets. The memory bus width is 384 bit, and the memory clock runs at 3000 MHz with an effective data rate of 24 Gbps. This combination yields a memory bandwidth of 1.15 TB/s, which is a high figure that allows for rapid data transfer between the GPU and the frame buffer. For gaming at 4K or higher resolutions, this bandwidth is crucial for maintaining smooth frame rates, as the GPU needs to access textures and geometry data quickly.

The 24 GB capacity is particularly relevant for future-proofing, as recent games have begun to exceed 16 GB of VRAM usage at ultra settings with high-resolution texture packs. The 384-bit bus width is a wide interface that enables the 1.15 TB/s bandwidth, which is twice what many mid-range cards offer. This memory subsystem is designed to handle the most demanding scenarios, such as 8K rendering or heavy modding in games. The GDDR6 type is not the newer GDDR6X or HBM, but the effective speed of 24 Gbps compensates for this, delivering bandwidth that is competitive with faster memory types on narrower buses.

In terms of high-resolution performance, the memory bandwidth of 1.15 TB/s ensures that the GPU is not bottlenecked by memory access. The pixel rate of 691.0 GPixel/s aligns with the memory bandwidth, indicating that the card can sustain high fill rates without stalling. For users who play at 1440p or 4K with maximum settings, this memory subsystem provides ample headroom. The 24 GB capacity also makes this card suitable for professional workloads like video editing or 3D rendering, where large assets are common. The data shows a well-balanced memory architecture, but without benchmark scores, the actual impact on frame rates remains theoretical.

How It Compares

The fact pack lists no nearest rivals, which means there are no direct comparison points available. This is a significant gap, as the analysis cannot delineate how the RX 7990 XTX stacks up against competing GPUs from other manufacturers or even within its own series. The percentile rank of 50 suggests that it is statistically average in the database, but this is a broad measure that does not account for specific performance tiers. Without rival names, scores, or deltaPct values, any comparison would be speculative and violate the rule to use only provided facts.

In the absence of rival data, the comparison must be internal. The RX 7990 XTX is positioned above its predecessor, Navi II, and below its successor, Navi IV, according to the generation field. The transistor count of 57,700 million and die size of 529 mm² indicate a large chip, which typically correlates with high performance. The FP32 throughput of 88.45 TFLOPS is a raw compute metric that can be compared to other cards only if their numbers were present. The RT core count of 96 is another spec that lacks context without rival figures.

The data shows that the card uses a triple-slot design and requires 3x 8-pin power connectors, which suggests it is a power-hungry component, but again, no wattage comparisons are possible. The bus interface is PCIe 4.0 x16, which is current but not the latest PCIe 5.0 standard. Overall, the comparison section is constrained by the lack of external data, and the only verifiable statement is that the card occupies a median position in the overall GPU landscape, which is a weak indicator for a flagship product.

FAQ

Q: What is the memory bandwidth of the RX 7990 XTX?

A: The memory bandwidth is 1.15 TB/s, derived from a 384-bit bus width and GDDR6 memory running at 3000 MHz with a 24 Gbps effective data rate.

Q: How many ray tracing cores does the card have?

A: The RX 7990 XTX features 96 RT cores, which are dedicated to accelerating ray tracing calculations.

Q: What is the FP32 compute performance?

A: The FP32 compute performance is 88.45 TFLOPS, and the FP16 performance is also 88.45 TFLOPS with a 1:1 ratio.

Q: What power supply is recommended for this GPU?

A: The suggested PSU is 800 W, and the card requires 3x 8-pin power connectors.

Q: What is the process node and transistor count?

A: The card is built on a 5 nm process from TSMC and contains 57,700 million transistors on a 529 mm² die.

Q: Which APIs are supported?

A: The card supports DirectX 12 Ultimate (12_2), OpenGL 4.6, and Vulkan 1.4.

Who Should Consider It

Based on the available data, the RX 7990 XTX is a card with high-end specifications, but its performance percentile of 50 complicates recommendations. The 24 GB of VRAM and 1.15 TB/s bandwidth make it suitable for 4K gaming with ultra-high-resolution textures, as the memory subsystem can handle large data sets without bottlenecking. The pixel rate of 691.0 GPixel/s and texture rate of 1,382.0 GTexel/s also support high-resolution rendering, suggesting that the card is built for demanding visual workloads. Users who play at 1440p or 4K and prioritize maximum settings would find the memory capacity and bandwidth appealing, but the lack of benchmark scores means the actual frame rates are unknown.

The FP32 throughput of 88.45 TFLOPS indicates strong compute capabilities, which could benefit users who engage in GPU-accelerated rendering or scientific computations. The 96 RT cores enable ray tracing, but without performance data, it is unclear whether the card can maintain playable frame rates with ray tracing enabled at high resolutions. The median percentile rank is a red flag, as it implies that the card does not outperform most other GPUs in the database, which is unusual for a flagship. This could suggest that the card is not optimized for standard gaming workloads or that the benchmark data is incomplete.

For users who need massive VRAM for modding or professional work, the 24 GB capacity is a compelling reason to consider this card. The triple-slot design and 405 W TDP indicate that it requires a spacious case and robust cooling, which is a practical consideration. The card is not recommended for budget builds, as the power and cooling requirements are substantial. Ultimately, the decision should hinge on whether the user values the memory and compute specs over the uncertain benchmark standing. Given the data, the card is best suited for users who prioritize VRAM capacity and raw compute over proven gaming performance, but the absence of scores makes a firm recommendation difficult.

Power and Cooling

The RX 7990 XTX has a TDP of 405 W, which is a high power draw that demands a robust cooling solution. The card is designed as a triple-slot unit, meaning it occupies three expansion slots in a case, which is necessary to dissipate the heat generated by the 405 W TDP. The dimensions are 287 mm in length, 135 mm in height, and 50 mm in width, making it a large card that may not fit in smaller cases. The power connectors are 3x 8-pin, which is a significant requirement, and the suggested PSU is 800 W, ensuring that the system has enough headroom for the GPU and other components.

The cooling solution is not specified in the fact pack, but the triple-slot width implies a substantial heatsink and multiple fans are used to manage thermals. The 405 W TDP is a hard number that indicates the maximum power the card is expected to draw under load, and users must ensure their power supply can handle this. The 800 W PSU recommendation is a guideline, and it assumes that the rest of the system components do not exceed typical power draw. The 3x 8-pin connectors are standard for high-end GPUs, but users must verify that their PSU has the necessary cables.

The card’s bus interface is PCIe 4.0 x16, which is compatible with most modern motherboards, but it does not support the newer PCIe 5.0 standard, which is a minor limitation. The dimensions are 287 mm in length, so users must check their case clearance, especially with front-mounted radiators or drive cages. The height of 135 mm and width of 50 mm also require attention, as some cases may have clearance issues with side panels. The triple-slot design is a critical factor for case compatibility, and users with multi-GPU setups or dense builds may need to plan accordingly. Overall, the power and cooling requirements are substantial, and the data suggests that this is not a card for small form factor builds or systems with limited power delivery.