Nvidia GeForce RTX 5090

Name: Nvidia GeForce RTX 5090
Brand: Nvidia
SKU: 9397

Manufacturer: Nvidia

The Nvidia GeForce RTX 5090 is a high-end graphics card based on the Blackwell architecture, manufactured on a 5 nm process and housing 92,200 million transistors. It operates with a base GPU clock of 2010 MHz and boosts up to 2410 MHz, while its 104.9 TFLOPS of floating-point performance reflects the considerable compute capability packed into this design. The card supports ray tracing, DLSS, and DirectX 12 Ultimate, and can drive up to four displays simultaneously.

On the memory side, the RTX 5090 is equipped with 32GB of GDDR7 VRAM across a 512-bit bus, delivering a maximum bandwidth of 1792 GB/s at an effective memory speed of 28000 MHz. The card features 21,760 shading units, 680 texture mapping units, and 176 render output units, producing a texture rate of 1638.8 GTexels/s and a pixel rate of 424.2 GPixels/s. Connectivity includes one HDMI 2.1b port and three DisplayPort outputs, with Intel Resizable BAR support and a rated TDP of 575W. ECC memory and Double Precision Floating Point are both supported, and the card uses PCIe 5.0 with physical dimensions of 304 mm by 137 mm.

Pros

Equipped with 32GB of GDDR7 VRAM across a 512-bit bus, providing substantial memory capacity and bandwidth for demanding workloads
Supports up to four simultaneous display outputs through a combination of one HDMI 2.1b port and three DisplayPort connections
Ray tracing and DLSS are both supported, enabling access to modern rendering features
ECC memory support adds a layer of data reliability for compute-sensitive tasks
Double Precision Floating Point is supported, broadening the card's suitability beyond pure graphics use
Intel Resizable BAR is included, allowing the CPU more flexible access to the full GPU memory pool

Cons

A TDP of 575W places significant demands on system power delivery and cooling infrastructure
Air-water cooling is not included, meaning cooling is dependent entirely on the system or third-party solutions
The card lacks USB-C, DVI, and mini DisplayPort outputs, limiting connectivity options for older or specialized display setups
At 304 mm in length, physical fitment may be a constraint in smaller or mid-tower cases
No RGB lighting is present for users who consider aesthetics as part of their build

Who is this for?

This card is well-suited to users who regularly work with compute-intensive or graphics-heavy workloads, given its 104.9 TFLOPS of floating-point performance, Double Precision Floating Point support, and ECC memory — all of which benefit tasks such as 3D rendering, simulation, and professional visualization. Content creators and technical users who need to drive up to four displays simultaneously will find the output configuration practical, while the 32GB of GDDR7 VRAM across a 512-bit bus makes it a solid fit for applications that involve large datasets or high-resolution assets. Those running modern titles or applications that rely on ray tracing and DLSS will also be able to take full advantage of the card's feature set.

Who is this NOT for?

Users working within systems that have limited power delivery will likely find the 575W TDP a significant constraint, as it demands a robust power supply and well-ventilated chassis. The card's 304 mm length means it may not physically fit in compact or smaller form-factor cases, making it unsuitable for space-constrained builds. Additionally, those relying on older display interfaces such as DVI, or who need USB-C or mini DisplayPort connectivity, will find no support for those outputs on this card.

Performance:

GPU clock speed	2010 MHz
GPU turbo	2410 MHz
pixel rate	424.2 GPixel/s
floating-point performance	104.9 TFLOPS
texture rate	1638.8 GTexels/s
GPU memory speed	1750 MHz
shading units	21760
texture mapping units (TMUs)	680
render output units (ROPs)	176
Has Double Precision Floating Point (DPFP)	✓

The RTX 5090 runs a base GPU clock of 2010 MHz, boosting up to 2410 MHz under load, while the GPU memory operates at 1750 MHz. With 104.9 TFLOPS of floating-point performance, the card also delivers a texture rate of 1638.8 GTexels/s and a pixel rate of 424.2 GPixel/s, backed by 680 texture mapping units and 176 render output units. Its 21,760 shading units contribute to the overall throughput, and Double Precision Floating Point is supported, making the card capable of handling compute workloads alongside graphics tasks.

Memory:

effective memory speed	28000 MHz
maximum memory bandwidth	1792 GB/s
VRAM	32GB
GDDR version	GDDR7
memory bus width	512-bit
Supports ECC memory	✓

The RTX 5090 is equipped with 32GB of GDDR7 VRAM running across a 512-bit memory bus, reaching an effective memory speed of 28000 MHz and a maximum bandwidth of 1792 GB/s. ECC memory is supported, which helps maintain data integrity in error-sensitive workloads.

Features:

DirectX version	DirectX 12 Ultimate
OpenGL version	4.6
OpenCL version	3
Supports multi-display technology	✓
supports ray tracing	✓
Supports 3D	✓
supports DLSS	✓
has XeSS (XMX)	✕
AMD SAM / Intel Resizable BAR	Intel Resizable BAR
has LHR	✕
has RGB lighting	✕
supported displays	4

The RTX 5090 supports DirectX 12 Ultimate, OpenGL 4.6, and OpenCL 3, covering a broad range of graphics and compute APIs. Ray tracing and DLSS are both supported, alongside stereoscopic 3D and multi-display technology spanning up to four outputs. Intel Resizable BAR is included, while XeSS (XMX) and LHR are not supported. The card does not feature RGB lighting, and Lite Hash Rate limiting is absent.

Ports:

has an HDMI output	✓
HDMI ports	1
HDMI version	HDMI 2.1b
DisplayPort outputs	3
USB-C ports	0
DVI outputs	0
mini DisplayPort outputs	0

The RTX 5090 offers a total of four video outputs: one HDMI 2.1b port and three DisplayPort outputs. There are no USB-C, DVI, or mini DisplayPort connections on this card.

General info:

GPU architecture	Blackwell
Thermal Design Power (TDP)	575W
PCI Express (PCIe) version	5
semiconductor size	5 nm
number of transistors	92200 million
Has air-water cooling	✕
width	304 mm
height	137 mm

The RTX 5090 is built on the Blackwell architecture, fabricated on a 5 nm process and integrating 92,200 million transistors. It connects via PCIe 5.0 and carries a rated TDP of 575W, with no built-in air-water cooling solution included. The card measures 304 mm in width and 137 mm in height.

Final Verdict

The Nvidia GeForce RTX 5090 is a graphics card built around a clear set of priorities: raw compute throughput, substantial memory capacity, and support for modern rendering and API standards. Its 32GB of GDDR7 VRAM paired with 1792 GB/s of memory bandwidth positions it firmly within demanding professional and high-end creative workloads, while ray tracing, DLSS, ECC memory, and Double Precision Floating Point support round out a technically comprehensive feature set. The card does come with real practical considerations — a 575W TDP and 304 mm length require careful system planning — but for users whose workloads and environments can accommodate those requirements, the RTX 5090 delivers a well-specified package that leaves little on the table in terms of features and capability.