NVIDIA’s most powerful consumer GPU generation with the new Blackwell architecture is now listed online. The price for this “super product,” the ASUS ROG Astral GeForce RTX 5090 OC Edition, is not cheap—up to $3,000.
The RTX 5090 from ASUS ROG with the Astral series is positioned as the super high-end segment, with many exclusive cooling technologies and an excellent design and finish. At first glance, the ASUS ROG Astral GeForce RTX 5090 OC Edition looks similar to the RTX 5080. However, a closer look reveals several key differences.
ASUS ROG Astral GeForce RTX 5090 OC Edition Hands-on: First Impression
Design and Build Quality
First, the RTX 5090 Astral OC is heavier than the RTX 5080 (3.03kg vs. 2.9kg), and its backplate is slightly thicker. The mysteries of the universe inspire the design. Both have an extremely high level of finishing, and the outer frame and backplate are both metal, ensuring the card’s solidity and durability.
The Astral RTX 5090 OC’s overall design is beautiful but straightforward. ASUS’s famous RGB Aura LED system does not appear too much. Mainly, small LED strips run along the card body and in other locations.
Power Consumption and Cooling
Of course, RTX 5090’s power consumption is higher, at 575W, and in this Astral OC Edition version, users can overclock to increase its performance further.
The ASUS ROG Astral GeForce RTX 5090 OC Edition takes up 3.8 slots in the case and is also one of the largest and heaviest RTX 5090 series models. ASUS also has a version that uses water cooling instead of air cooling, but the price is much higher.
ASUS’ four fans include three on the front and one on the back on the backplate. The fourth fan acts as a “pull” fan, increasing airflow over the heatsink. This design creates a powerful vertical airflow, increasing air pressure by up to 20%. The axial-tech fans have 7 blades to enhance airflow and optimize cooling efficiency.
The vapor chamber cooling system is designed with milled pathways, allowing the heat pipes to lie neatly inside. This design increases the contact area between the heat pipes and the vapor chamber assembly, improving the cooling efficiency of the entire system. MaxContact Technology is also an ASUS-exclusive technology. It increases the contact area between the GPU heatsink and GPU by 5%, further improving cooling efficiency.
ASUS also uses a premium phase-change GPU thermal pad to fill the gap between the GPU and the heatsink module, enhancing thermal conductivity.
Included Accessories
The accessories inside the ASUS ROG Astral GeForce RTX 5090 OC Edition box are identical to those in the RTX 5080 version (Hands-On). They include a 3-head, 8-pin to 16-pin converter cable for users using the old standard power supply. They also come with a ruler, a miniature model keycap, and other accessories.
Connectivity
The Astral RTX 5090 OC’s I/O ports are the same as those on the RTX 5080 version, including 3 DisplayPort 2.1b ports and 2 HDMI 2.1b ports.
Performance
At the heart of the ASUS ROG Astral GeForce RTX 5090 OC Edition is the GB202 GPU, manufactured on a 5nm process optimized specifically for NVIDIA by TSMC. It boasts a transistor count of 92.2 billion. The RTX 5090 also features over 21,000 CUDA cores, 680 Tensor cores, and 170 RT cores, delivering over 3,000 TOPS of AI computing power.
Compared to the Founders Edition, the ASUS ROG Astral GeForce RTX 5090 OC Edition has a 2580MHz clock speed, a 512-bit memory interface, and 1.79TB/s memory bandwidth.
It also features a 32GB GDDR7 memory capacity. Both the RTX 5090 and RTX 5080 use 5th-generation Tensor Cores, supporting the FP4 data format.
AI and Tensor Core Advancements
The RTX 5090 has more than double the number of Tensor Cores than the RTX 5080. This directly affects the ability to handle AI tasks, as Tensor Cores are specialized parts that accelerate calculations related to AI and deep learning. Therefore, if you need to build an AI system, the RTX 5090 will be more suitable than the RTX 5080.
The Blackwell architecture introduces the concept of neural shaders, allowing applications to access Tensor Cores directly via the graphics API. This means that the RTX 5090 and RTX 5080 can use Tensor Cores to accelerate AI-related graphics tasks, such as neural rendering, DLSS, and more.
Both cards feature an AI Management Processor (AMP), a dedicated processor that manages AI acceleration resources on the GPU. This allows the GPU to handle AI and graphics tasks simultaneously without affecting the performance of the other. This will be especially important for the two key technologies on the RTX 50 Series: DLSS 4 Multi Frame Generation.
Final line
Overall, the RTX 5090 has significantly better AI processing power than the RTX 5080.
This is thanks to more Tensor Cores, a larger SM count, more L2 cache, and other more powerful components. Both cards support NVIDIA’s advanced AI technologies, but the RTX 5090 excels in tasks that demand higher AI performance.
B&HPhotoVideo, Newegg, and Amazon now list it for $3,079.99, but no release date details are available.
