Graphics cards/NVIDIA GeForce RTX 5090
NVIDIA · GeForce RTX 50 · Desktop
NVIDIA GeForce RTX 5090
The NVIDIA GeForce RTX 5090 is NVIDIA’s Blackwell (GB202) desktop graphics card with 32 GB of GDDR7 memory, 21,760 shading units and a boost clock of 2410 MHz.

Photo: PantheraLeo1359531 · CC BY 4.0 · via Wikimedia Commons
Overview
The NVIDIA GeForce RTX 5090 is NVIDIA’s Blackwell (GB202) desktop graphics card with 32 GB of GDDR7 memory, 21,760 shading units and a boost clock of 2410 MHz. These are the manufacturer’s published figures. They describe what the chip is designed to do, not what your particular machine does with it, and the gap between those two is usually cooling. A boost clock is a ceiling rather than a promise, and how long a graphics card holds it under continuous load is the one thing no specification sheet can tell you.
Specifications
- Architecture
- Blackwell (GB202)
- Shading units
- 21,760
- Base clock
- 2010 MHz
- Boost clock
- 2410 MHz
- Memory
- 32 GB
- Memory type
- GDDR7
- Memory bus
- 512-bit
- TDP
- 575 W
- Process
- 5 nm
- Launched
- 2025-01-30
Comparable within an architecture. Across vendors or generations it is not a like-for-like count.
Board power, not measured draw. A laptop GPU's power is a range the laptop maker chooses within.
Sources
Each figure, the label its source printed beside it, and the day it was read.
| Figure | As printed | Source | Checked |
|---|---|---|---|
| Shading units21,760 | NVIDIA CUDA Cores | nvidia.comSpecification page | |
| Base clock2010 MHz | Base Clock (GHz) | nvidia.comSpecification page | |
| Boost clock2410 MHz | Boost Clock (GHz) | nvidia.comSpecification page | |
| Memory32 GB | Standard Memory Config | nvidia.comSpecification page | |
| Memory typeGDDR7 | Standard Memory Config | nvidia.comSpecification page | |
| Memory bus512-bit | Memory Interface Width | nvidia.comSpecification page | |
| TDP575 W | Total Graphics Power (W) | nvidia.comSpecification page | |
| Launched2025-01-30 | The new GeForce RTX 5090 and GeForce RTX 5080 are out now | nvidia.comLaunch release |
Architecture, Process: recorded from en.wikipedia.org before each figure carried its own source. The maker's page does not print them in this form, so they do not count as verified.
Measured on SystemCheck: not enough runs yet. Run the test to add yours.
Compare
The 6 nearest graphics cards in the same segment, each with a published side-by-side.
- NVIDIA GeForce RTX 5090 vs NVIDIA GeForce RTX 4090
- NVIDIA GeForce RTX 5090 vs AMD Radeon RX 7900 XTX
- NVIDIA GeForce RTX 5090 vs NVIDIA GeForce RTX 5080
- NVIDIA GeForce RTX 5090 vs NVIDIA GeForce RTX 4080 SUPER
- NVIDIA GeForce RTX 5090 vs NVIDIA GeForce RTX 4080
- NVIDIA GeForce RTX 5090 vs NVIDIA GeForce RTX 5070 Ti
Other graphics cards in NVIDIA GeForce RTX 50
What these numbers mean
Every figure on this sheet is a design target. None of it is a measurement.
Nothing above describes how a graphics card behaves twenty minutes into a sustained workload, which is where two chips with nearly identical specifications routinely diverge. A boost clock is a ceiling the vendor permits, not a rate the chip sustains, and whether it holds depends on the cooling, the case, the power profile and the room. None of those appear on a specification sheet.
We deliberately do not reduce these figures to a single score. A composite invented from specifications would look authoritative and measure nothing, and choosing the weights would be choosing the winner.


