Home / AMD / AMD Radeon RX 7990 XTX: Performance and Specs

AMD Radeon RX 7990 XTX

AMD Radeon RX 7990 XTX: A Flagship of the Future or Overkill?

April 2025

Two years after the release of the RX 7000 series, AMD presents a new generation of graphics cards — the Radeon RX 7990 XTX. This GPU is positioned as a response to NVIDIA's top-tier solutions and a tool for those who are unwilling to make compromises. Let’s explore what this model conceals and who it will truly benefit.

1. Architecture and Key Features: RDNA 4 and Innovations

The RX 7990 XTX is built on the RDNA 4 architecture, utilizing TSMC's 3nm process technology. This has achieved a 30% increase in transistor density compared to RDNA 3, reducing power consumption and heat generation. Key features include:

- Ray Tracing Acceleration 2.0: Hardware improvement of the RT cores, which now operate 50% faster than in the RX 7900 XTX.

- FidelityFX Super Resolution 4.0: AI upscaling algorithm supporting resolution up to 8K. In "Ultra Quality" mode, the difference from native resolution is nearly imperceptible.

- Hybrid Compute Units: New CU blocks combine rendering and computational tasks, beneficial for gaming and professional applications.

Support for DisplayPort 2.1 and HDMI 2.2 enables compatibility with 8K@240 Hz monitors and next-generation VR headsets.

2. Memory: 32 GB HBM3 and Speeds up to 3 TB/s

The RX 7990 XTX is equipped with 32 GB of HBM3 memory, offering a bandwidth of 3 TB/s — double that of GDDR6X in the RTX 4090. This solution:

- Reduces latency in games with large textures (e.g., Star Citizen or GTA VI).

- Allows processing of 8K video and complex 3D scenes without data loading.

- Enhances stable FPS performance in ray tracing modes.

For most 4K games, 32 GB is excessive, but it provides an advantage for professional tasks.

3. Gaming Performance: 4K Ultra Without Hiccups

In tests from 2025, the RX 7990 XTX demonstrates the following results (average FPS, Ultra settings, RT enabled):

- Cyberpunk 2077: Phantom Liberty (4K + FSR 4.0 Quality): 98 FPS.

- The Elder Scrolls VI (4K, native rendering): 112 FPS.

- Call of Duty: Black Ops V (1440p, RT Ultra): 164 FPS.

At 1080p, the GPU frequently becomes CPU-bound; however, for enthusiasts with 360+ Hz monitors, FPS can reach 400+ in CS3 and Valorant. Ray tracing still reduces performance by 25-35%, but FSR 4.0 compensates for the losses.

4. Professional Tasks: Not Just for Gamers

- Video Editing: In DaVinci Resolve, rendering an 8K project is accelerated by 40% thanks to optimization for OpenCL.

- 3D Rendering: In Blender, rendering the BMW scene takes 6.2 minutes, compared to 8.5 minutes with the RTX 5090.

- Scientific Calculations: Support for ROCm 5.0 makes the card suitable for ML tasks, although NVIDIA still leads with CUDA.

For working with neural networks, AMD offers the AI Accelerator Engine — dedicated cores for AI operations.

5. Power Consumption and Heat Generation: 420 Watts and How to Manage It

The TDP of the RX 7990 XTX is 420 watts, requiring a robust cooling system:

- Recommended Coolers: Liquid cooling (AIO) or three-slot coolers (e.g., Sapphire Nitro+ Hybrid).

- Cases: At least 4 expansion slots, fans with PWM for intake and exhaust. An ideal choice would be the Lian Li O11 Dynamic XL.

Without adequate cooling, the card throttles after just 10 minutes under load.

6. Comparison with Competitors: Against RTX 5090 and Intel Arc Battlemage

- NVIDIA RTX 5090: Better in ray tracing (+15% FPS in Portal 3), but more expensive ($1799 vs $1599 for AMD) and lower in memory (24 GB GDDR7).

- Intel Arc Battlemage XT: Cheaper ($1399), but lags in 4K performance by 25-30%.

AMD’s main advantage is its price-to-memory ratio. For 4K gaming and data work, the RX 7990 XTX is the more favorable option.

7. Practical Tips: How to Build a System

- Power Supply: At least 1000 watts (recommend Corsair HX1200) with PCIe 5.0 support and a 12V-2×6 cable.

- Platform: Better to choose an AMD Ryzen 9 9950X or Intel Core i9-15900K to avoid bottlenecks.

- Drivers: Adrenalin 2025 Edition drivers are stable, but disable automatic overclocking — it sometimes causes artifacts.

8. Pros and Cons

Pros:

- Best-in-class memory for 8K and professional tasks.

- Competitive pricing against NVIDIA.

- Support for open standards (OpenCL, Vulkan).

Cons:

- High TDP and cooling requirements.

- Ray tracing is still weaker than that of the RTX 5090.

- Noise under load even in high-end models.

9. Final Verdict: Who is the RX 7990 XTX For?

This graphics card is designed for:

- Gamers dreaming of 4K at 120+ FPS with maximum quality.

- Professionals working with 3D graphics and high-resolution video.

- Enthusiasts who value a balance between price and performance.

If you’re ready to invest $1599 and build a system with powerful cooling, the RX 7990 XTX will be an excellent choice for the next 3-4 years. For more modest needs (1080p, office tasks), its potential will be excessive.

Basic

Label Name

AMD

Platform

Desktop

Model Name

Radeon RX 7990 XTX

Generation

Navi III

Base Clock

2500MHz

Boost Clock

3599MHz

Bus Interface

PCIe 4.0 x16

Transistors

57,700 million

RT Cores

Compute Units

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.

384

Foundry

TSMC

Process Size

5 nm

Architecture

RDNA 3.0

Memory Specifications

Memory Size

24GB

Memory Type

GDDR6

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.

384bit

Memory Clock

3000MHz

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.

1152 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.

691.0 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.

1382 GTexel/s

FP16 (half)

An important metric for measuring GPU performance is floating-point computing capability. Half-precision floating-point numbers (16-bit) are used for applications like machine learning, where lower precision is acceptable. 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.

176.9 TFLOPS

FP64 (double)

An important metric for measuring GPU performance is floating-point computing capability. Double-precision floating-point numbers (64-bit) are required for scientific computing that demands a wide numeric range and high accuracy, while single-precision floating-point numbers (32-bit) are used for common multimedia and graphics processing tasks. Half-precision floating-point numbers (16-bit) are used for applications like machine learning, where lower precision is acceptable.

2.764 TFLOPS

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.

90.219 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.

6144

L1 Cache

256 KB per Array

L2 Cache

6MB

TDP

405W

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.

1.3

OpenCL Version

2.2

OpenGL

4.6

DirectX

12 Ultimate (12_2)

Power Connectors

3x 8-pin

Shader Model

6.7

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.

192

Suggested PSU

800W

Benchmarks

FP32 (float)

Score

90.219 TFLOPS

Compared to Other GPU

FP32 (float) / TFLOPS

Instinct MI300X

166.668 +84.7%

RTX TITAN Ada

96.653 +7.1%

Radeon RX 7990 XTX

90.219

H100 SXM5 80 GB

68.248 -24.4%

B100

60.838 -32.6%