Introduction

The MSI GeForce RTX 5090 SUPRIM Liquid SOC graphics card is designed to be the fastest custom-design RTX 5090, which in turn is the flagship next-generation graphics card that the new GeForce Blackwell generation kicks the door open with. The SUPRIM brand of graphics cards from MSI represent the company's highest tier of industrial design and thermal engineering. This line of cards are designed to go up against NVIDIA's Founders Edition in design, and take on the custom design market to top spots in out of the box performance, factory overclock, cooler performance, and noise, all while looking like a piece of jewellery. We have not one but two RTX 5090 SUPRIM series graphics cards that we're reviewing for you today, this one is the MSI RTX 5090 SUPRIM Liquid SOC, which comes with a factory-fitted closed-loop liquid cooling solution; while the other is the air-cooled MSI RTX 5090 SUPRIM SOC. There are two sub-variants for each of the two, SOC and OC, we have with us the faster SOC (super overclock) edition.



The GeForce RTX 5090 and the Blackwell graphics architecture it's based on, debut a fundamentally new chapter in real time 3D graphics, called neural rendering. By now you've experienced the awesome capability of generative AI to create photorealistic images and videos. NVIDIA and its allied researchers have created a way to combine objects that have been created in real time by a generative AI with conventional raster 3D graphics, much in the same way it figured out how to combine real time ray traced objects with raster 3D. This is done by creating a way for graphics applications to directly access Tensor cores, and the GPU having the ability to run 3D rendering and AI acceleration workloads in tandem, thanks to a new hardware scheduler NVIDIA calls the AI management processor.

The new Blackwell graphics architecture brings updates to all key areas of the GPU. The new generation CUDA cores present generational IPC uplifts, concurrent FP32 and INT32 execution capability on all cores in an SM, and introduce awareness for neural shaders, including an updated shader executing reordering that sends neural shader traffic to the Tensor cores. The new 5th Gen Tensor cores gain FP4 data format support for 32X the throughput over the original Tensor cores. The new 4th Gen RT core is ready for Mega Geometry, the ability for ray traced objects to have significantly higher poly counts thanks to the new Triangle Cluster Intersection engine, and Triangle Cluster Decompression engine. NVIDIA has also re-architected the power management engine of the GPU.

At the heart of the RTX 5090 is the new GB202 silicon. A surprising aspect of this generation is that it is built on the exact same foundry node as the RTX 40-series Ada. This would be the variant of the 5 nm EUV node that TSMC co-developed with NVIDIA, called the TSMC 4N. Whatever generational performance per watt gains you see are hence entirely due to the newer graphics architecture and power management engine. The GB202 is a 750 mm² Goliath with over 92 billion transistors, and 192 streaming multiprocessors, a PCI-Express 5.0 x16 host interface, and a 512-bit wide GDDR7 memory interface that nearly doubles the memory bandwidth over the previous generation.

The RTX 5090 is carved out of the GB202 by enabling 170 out of 192 SM, resulting in 21,760 CUDA cores, 680 Tensor cores, and 170 RT cores, across 11 GPCs, and this doesn't even max out the GB202. The card comes with 32 GB of 28 Gbps GDDR7 memory across a 512-bit wide memory bus, giving it 1.79 TB/s of memory bandwidth. You'll see large increases in memory bandwidth across the board, since neural shaders, like pretty much all accelerated AI, are memory sensitive.

The MSI RTX 5090 SUPRIM Liquid SOC is more compact than even the RTX 5090 Founders Edition, because the card lugs a 360 mm radiator. The card is designed to look like jewellery, made with premium materials and machining. Under the hood, the GB202 is powered by a 29-phase VRM using the highest-grade chokes, controllers, and DrMOS available. The liquid cooling solution pulls heat from the GPU and the sixteen GDDR7 memory chips. The cold plate makes contact with an additional network of heatsinks that cool the VRM solution. A dedicated 100 mm MSI StormForce axial airflow fan ventilates this heatsink. MSI has given the SUPRIM Liquid SOC its highest factory overclock, with the GPU Boost going up to 2565 MHz (vs. 2407 MHz reference). These are the exact same clock speeds the air-cooled MSI RTX 5090 SUPRIM SOC comes with, but the liquid cooling should provide better boost frequency residency and lower noise. Out of the box, though, both our MSI SUPRIM samples came with the BIOS switch in the "silent" setting, which boosts up to 2512 MHz. We test cards in the BIOS setting they come with, out of the box. MSI is pricing the MSI RTX 5090 SUPRIM Liquid SOC at $2,500, a 25% premium over the already steep $2,000 starting price for the RTX 5090.

NVIDIA Blackwell Architecture

The GeForce Blackwell graphics architecture heralds NVIDIA's 4th generation of RTX, the late-2010s re-invention of the modern GPU that sees a fusion of real time ray traced objects with conventional raster 3D graphics. With Blackwell, NVIDIA is helping add another dimension, neural rendering, the ability for the GPU to leverage a generative AI to create portions of a frame. This is different from DLSS, where an AI model is used to reconstruct details in an upscaled frame based on its training date, temporal frames, and motion vectors. At the heart of the GeForce RTX 5090 we are reviewing today is the mammoth 5 nm GB202 silicon. This is one of the largest monolithic dies ever designed by NVIDIA, measuring 750 mm², compared to the 608.5 mm² of the AD102 die. The process is unchanged between the two generations—it's still an NVIDIA-specific variant of TSMC 5 nm EUV, dubbed TSMC 4N. The GB202 rocks 92.2 billion transistors, a 20% increase over the AD102.


The GB202 silicon is laid out essentially in the same component hierarchy as past generations of NVIDIA GPUs, but with a few notable changes. The GPU features PCI-Express 5.0 x16, making it the first gaming GPU to do so. PCIe Gen 5 has been around since Intel's 12th Gen Core "Alder Lake" and AMD's Ryzen 7000 "Zen 4," so there is a sizable install-base of systems that can take advantage of it. The GPU is of course compatible with older generations of PCIe. Whether this affects performance is a question we cover in our separate RTX 5090 PCIe Scaling Article. The GB202 is also the first GPU to implement the new GDDR7 memory standard, which doubles speeds over GDDR6 while lowering the energy cost of bandwidth. NVIDIA left no half-measures with the GB202, and gave it a broad 512-bit GDDR7 memory interface. On the RTX 5090, this is configured with 32 GB of 28 Gbps GDDR7. Upcoming RTX 50-series SKUs could have narrower memory interfaces but with higher memory speeds, and some professional graphics cards based on the GB202 could even use high-density memory chips.

The GigaThread Engine is the main graphics rendering workload allocation logic on the GB202, but there's a new addition, a dedicated serial processor for managing all AI acceleration resources on the GPU, NVIDIA calls this AMP (AI management processor). Other components at the global level are the Optical Flow Processor, a component involved in older versions of DLSS frame generation and for video encoding; and a vast media acceleration engine consisting of four NVENC encode accelerators, and four NVDEC decode accelerators. The new 9th Gen NVENC video encode accelerators come with 4:2:2 AV1 and HEVC encoding support. The RTX 5090 has 3 out of 4 NVENC and 2 out of 4 NVDEC units enabled. The central region of the GPU has the single largest common component, the 128 MB L2 cache. The RTX 5090 is configured with 96 MB of it.


Each graphics processing cluster (GPC) is a subdivision of the GPU with nearly all components needed for graphics rendering. On the GB202, a GPC consists of 16 streaming multiprocessors (SM) across 8 texture processing clusters (TPCs), and a raster engine consisting of 16 ROPs. Each SM contains 128 CUDA cores. Unlike the Ada generation SM that each had 64 FP32+INT32 and 64 purely-FP32 SIMD units, the new Blackwell generation SM features concurrent FP32+INT32 capability on all 128 SIMD units. These 128 CUDA cores are arranged in four slices, each with a register file, a level-0 instruction cache, a warp scheduler, two sets of load-store units, and a special function unit (SFU) handling some special math functions such as trigonometry, exponents, logarithms, reciprocals, and square-root. The four slices share a 128 KB L1 data cache, and four TMUs. The most exotic components of the Blackwell SM are the four 5th Gen Tensor cores, and a 4th Gen RT core.


Perhaps the biggest change to the way the SM handles work introduced with Blackwell is the concept of neural shaders—treating portions of the graphics rendering workload done by a generative AI model as shaders. Microsoft has laid the groundwork for standardization of neural shaders with its Cooperative Vectors API, in the latest update to DirectX 12. The Tensor cores are now accessible for workloads through neural shaders, and the shader execution reordering (SER) engine of the Blackwell SM is able to more accurately reorder workloads for the CUDA cores and the Tensor core in an SM.


The new 5th Gen Tensor core introduces support for FP4 data format (1/8 precision) to fast moving atomic workloads, providing 32 times the throughput of the very first Tensor core introduced with the Volta architecture. Over the generations, AI models leveraged lesser precision data formats, and sparsity, to improve performance. The AI management processor (AMP) is what enables simultaneous AI and graphics workloads at the highest levels of the GPU, so it could be simultaneously rendering real time graphics for a game, while running an LLM, without either affecting the performance of the other. AMP is a specialized hardware scheduler for all the AI acceleration resources on the silicon. This plays a crucial role for DLSS 4 multi-frame generation to work.


The 4th Gen RT core not just offers a generational increase in ray testing and ray intersection performance, which lowers the performance cost of enabling path tracing and ray traced effects; but also offers a potential generational leap in performance with the introduction of Mega Geometry. This allows for ray traced objects with extremely high polygon counts, increasing their detail. Poly count and ray tracing present linear increases in performance costs, as each triangle has to intersect with a ray, and there should be sufficient rays to intersect with each of them. This is achieved by adopting clusters of triangles in an object as first-class primitives, and cluster-level acceleration structures. The new RT cores introduce a component called a triangle cluster intersection engine, designed specifically for handling mega geometry. The integration of a triangle cluster compression format and a lossless decompression engine allows for more efficient processing of complex geometry.


The GB202 and the rest of the GeForce Blackwell GPU family is built on the exact same TSMC "NVIDIA 4N" foundry node, which is actually 5 nm, as previous-generation Ada, so NVIDIA directed efforts to finding innovative new ways to manage power and thermals. This is done through a re-architected power management engine that relies on clock gating, power gating, and rail gating of the individual GPCs and other top-level components. It also worked on the speed at which the GPU makes power-related decisions.


The quickest way to drop power is by adjusting the GPU clock speed, and with Blackwell, NVIDIA introduced a means for rapid clock adjustments at the SM-level.


NVIDIA updated both the display engine and the media engine of Blackwell over the previous generation Ada, which drew some flack for holding on to older display I/O standards such as DisplayPort 1.4, while AMD and Intel had moved on to DisplayPort 2.1. The good news is that Blackwell supports DP 2.1 with UHBR20, enabling 8K 60 Hz with a single cable. The company also updated NVDEC and NVENC, which now support AV1 UHQ, double the H.264 decode performance, MV-HEVC, and 4:2:2 formats.

Neural Rendering

Neural Rendering promises to be as transformative to modern graphics as programmable shaders itself. 3D Graphics rendering evolved from fixed-function over the turn of the century, to programmable shaders, HLSL, geometry shaders, compute shaders, and ray tracing, over the past couple of decades. In 2025, NVIDIA is writing the next chapter in this journey with Blackwell neural shaders. This allows for a host of neural-driven effects, including neural materials, neural volumes, and even neural radiance fields. Microsoft introduced the new Cooperative Vectors API for DirectX in a recent update, making it possible to access Tensor cores within a graphics API. Combined with a new shading language, Slang, this breakthrough enables developers to integrate neural techniques directly into their workflows, potentially replacing parts of the traditional graphics pipeline. Slang splits large, complex functions into smaller pieces that are easier to handle. Given that this is a DirectX standard API feature, there is nothing that stops AMD and Intel from integrating Neural Rendering (Cooperative Vectors) into their graphics drivers.

RTX Neural Materials works to significantly reduce the memory footprint of materials in 3D scenes. Under conventional rendering, the memory footprint of a material is bloated from complex shader code. Neural materials convert shader code and texture layers into a compressed neural representation. This results in up to a 7:1 compression ratio and enables small neural networks to generate stunning, film-like materials in real-time. For example, silk rendered with traditional shaders might lack the multicolored sheen seen in real life. Neural materials, however, capture intricate details like color variation and reflections, bringing such surfaces to life with unparalleled realism—and at a fraction of the memory cost.


The new Neural Radiance Cache, which dynamically trains a neural network during gameplay using the user's GPU, allowing light transport to be cached spatially, enabling near-infinite light bounces in a scene. This results in realistic indirect lighting and shadows with minimal performance impact. NRC partially traces 1 or 2 rays before storing them in a radiance cache, and infers an infinite amount of rays and bounces for a more accurate representation of indirect lighting in the game scene.

DLSS 4 and Multi Frame Generation

DLSS 4 introduces a major leap in image quality and performance. It isn't just a version bump with the introduction of a new feature, namely Multi Frame Generation, but introduces updates to nearly all DLSS sub-features. DLSS from its very beginning relied on AI to reconstruct details in super resolution, and with DLSS 4, NVIDIA is introducing a new transformer-based AI model to succeed the convolutional neural networks previous used, for double the parameters, four times the compute performance, and significantly improved image quality. Ray Reconstruction, introduced with DLSS 3.5, gets a significant image quality update with the new transformer-based model.


To understand Multi Frame Generation, you need to understand how DLSS Frame Generation, introduced with GeForce Ada, works. An Optical Flow Accelerator component gives the DLSS algorithm data to generate an entire frame using a neural network, using information from a previous rendered frame, effectively doubling frame rate. In Multi Frame Generation, AI takes over the functions of optical flow, to predict up to three frames following a conventionally rendered frame, effectively drawing four frames form the rendering effort of one.


Now, assuming this rendered frame is a product of Super Resolution, with the maximum performance setting generating 4x the pixels from a single rendered pixel, you're looking at a possibility where the rendering effort of 1/4th a frame goes into drawing 4 frames, or 15 in every 16 pixels being generated entirely by DLSS. When generating so many frames, Frame Pacing becomes a problem—irregular frame intervals impact smoothness. DLSS 4 addresses these issues by using a dedicated hardware unit inside Blackwell, which takes care of flip metering, reducing frame display variability by 5-10x. The Display Engine of Blackwell contains the hardware for flip metering.

NVIDIA Reflex 2

The original NVIDIA Reflex brought about a significant improvement to the responsiveness of maxed out graphics in competitive online gameplay, by compacting the rendering queue with the goal of reducing the whole system latency by up to 50%. Reflex is mandatory in DLSS 3 Frame Generation, given the latency cost imposed by the technology. Multi-frame generation calls for an equally savvy piece of technology, so we hence have Reflex 2. NVIDIA claims to have achieved a 75% reduction in latency with Frame Warp, which updates the camera (viewport) positions based on user inputs in real-time, and then uses temporal information to reconstruct the frame to display.

Packaging

The Card

MSI follows the design theme of their GeForce 40 Suprim lineup—metal silver with gray highlights—looks fantastic, I also like the way the cover is shaped with the sharp angle.


MSI's watercooling radiator is a classic triple-fan design using three 120 mm fans.


Dimensions of the card are 28.0 x 15.0 cm, and it weighs 2913 g.


Installation requires three slots in your system. We measured the card's width to be 53 mm.


Display connectivity includes three standard DisplayPort 2.1b and one HDMI 2.1b.

Standard for all GeForce RTX 50-series Blackwell cards is a new display engine that supports three DisplayPort 2.1b outputs, each capable of UHBR20; and one HDMI 2.1a. Both interfaces support DSC (display stream compression). With DSC enabled, a single DisplayPort on this card can drive 4K 12-bit HDR at 480 Hz; or 8K 12-bit HDR at up to 165 Hz. The RTX 5090 features an updated media acceleration engine with support for 4:2:2 video formats, AV1 UHQ, and MV-HEVC. There are three independent NVENC units, and two NVDEC.


The card uses a single 16-pin connector, which allows a maximum power draw of 600 W.


MSI has installed adjustable RGB lighting zones near the fans, the Suprim text logo is illuminated and the Suprim logo on the corner, too.


This dual BIOS switch lets you toggle between the default "silent" BIOS and a "gaming" BIOS with a more aggressive fan curve. The Gaming BIOS will also increase the default power limit to 600 W (up from 575 W).