Introduction

The ASUS ROG Astral GeForce RTX 5090 Liquid OC is the company's flagship graphics card, based on NVIDIA's top dog GPU from this generation, the RTX 5090. This card is designed to offer the highest possible performance for gaming, to climb up OC leaderboards, and for AI acceleration, where the relatively compact size comes in handy. The RTX 5090 is designed to plow through games at 4K Ultra HD with maxed out settings, including ray tracing. It could also unlock certain enthusiast use-cases, such as 4K high refresh-rate, or even 8K, by tapping into features such as DLSS 4 Multi Frame Generation. ASUS created the ROG Astral brand of graphics cards to be positioned above its popular ROG Strix brand. For the RTX 5090 and RTX 5080, ASUS won't have a ROG Strix product, but the ROG Astral Liquid, the air-cooled ROG Astral, which we've already reviewed, and straight to the TUF Gaming OC (also reviewed). The ROG Astral air-cooled cards come with a 4-fan setup, where the backplate has the fourth fan toward the tail-end, pulling air through the portion of the heatsink that's exposed completely. For the ROG Astral LC in this review, the air cooling solution makes way for a factory-fitted, all-in-one, closed-loop liquid cooling solution.



The ASUS ROG Astral RTX 5090 Liquid OC, as its name suggests, comes with a liquid cooling solution. The solution consists of a pump + water block located on the card's end, which pulls heat from the large RTX 5090 GB202 GPU and the sixteen GDDR7 memory chips surrounding it. The VRM solution, on the other hand, is cooled by a heatsink that's ventilated by a single fan. The GPU and memory are the main heat sources, for which the liquid cooling loop does the heavy lifting to cool; while the VRM is kept well within efficient operational temperatures by the air cooling solution. The card is a little bit more than two slots thick and about two-thirds the length of the air-cooled ROG Astral; but it lugs along a 360 mm radiator. This radiator has a trio of factory-fitted 120 mm fluid dynamic bearing fans. The fans are daisy-chained via contact points, a mechanism that minimizes cable clutter involved in interconnecting them. A single cable provides PWM and ARGB connections to all three, and the unified fan control cable discreetly vines its way around one of the coolant tubes to the card.

The GeForce RTX 5090 is the flagship GPU from the RTX 50-series Blackwell generation. It is based on the GB202 silicon, a mammoth monolithic chip with over 92 billion transistors on a die measuring 750 mm². The GPU introduces several firsts-to-market features, including the PCI-Express 5.0 x16 host interface, ATX 3.1 + PCIe Gen 5 power architecture, DisplayPort 2.1b UHBR20, and the new GDDR7 memory standard. The GPU features a wide 512-bit GDDR7 memory interface running 32 GB of 28 Gbps GDDR7 memory for a jaw-dropping 1,792 GB/s memory bandwidth. The GPU needs this for neural rendering, DLSS 4 multi frame generation, and other new technologies to work. The RTX 5090 enables 170 out 192 SM, and 96 MB out of the 128 MB L2 cache present on the silicon. It also gets 3 out of 4 NVENC accelerators, and 2 out of 4 NVDEC. This works out to 21,760 CUDA cores, 680 Tensor cores, and 170 RT cores, across 11 GPCs.

The GeForce RTX 5090 is powered by the Blackwell graphics architecture, which has new capabilities, too. To begin with, it implements Neural Rendering, a feature that brings the power of generative AI directly to the graphics pipeline. The GPU now has the ability to run a generative AI model in tandem with rendering graphics thanks to a new hardware scheduler on these chips, called the AI Management Processor (AMP). Neural Rendering allows certain objects created by the generative AI to be combined with conventional raster 3D graphics, just like RTX brings real time ray traced objects to it. This should vastly improve realism in games. The new Blackwell generation SM comes with concurrent FP32 and INT32 math capability on all CUDA cores, the previous generation Ada only had INT32 capability on half the cores in an SM. The shader execution reordering is now aware of neural shaders. The 5th generation Tensor cores come with FP4 data format capability, which should max out throughput by trading in precision. The new RT cores come with even more hardware accelerators, and are ready for Mega Geometry, which vastly improves the poly count of ray traced objects using hierarchical techniques resembling Mega Textures.

Then there are DLSS 4 and Multi Frame Generation. NVIDIA updated the AI model at the heart of the DLSS upscaler to one based on more advanced Transformers, instead of an older convoluted neural network (CNN). The new Transformer based model is more accurate, and hence there are image quality uplifts to be expected in all performance presets. NVIDIA introduced a new hardware component with Blackwell called Display Flip Metering, with which Blackwell implements Multi Frame Generation, or the ability for the GPU to generate up to three consecutive frames from a single conventionally rendered one, effectively quadrupling frame rates. While the Transformer models for upscaling and ray-reconstruction are available even for older RTX 40-series and RTX 30-series GPUs, Multi Frame Generation is exclusive to the RTX 50-series.

The ASUS ROG Astral RTX 5090 Liquid OC looks like a piece of jewellery, and is designed to visually match the flagship ASUS ROG Maximus and ROG Crosshair motherboards. In fact there's a special edition of this card in the works, called the Dhahab Edition, which is painted in gold, and available in the UAE. Even the standard ROG Astral Liquid OC comes at an eye-watering $3,400, or an 70% premium over the $2,000 baseline NVIDIA set for the RTX 5090.