GeForce RTX 6090
GeForce RTX 6090 is NVIDIA's upcoming flagship graphics card based on the architecture Ruby, which will be the most powerful solution in the RTX 6000 series. This is a card without compromises: maximum performance, hardware-level ray tracing, huge memory capacity and top performance in all scenarios - from 4K gaming to generative tasks and rendering.
Architecture
The RTX 6090 is based on the new Rubin architecture, designed from the ground up with a focus on increased compute density, efficient ray tracing scaling, and deep neural network integration. Rubin uses TSMC 3nm process technology и grid 4x, thanks to which the density of logical blocks has increased and the internal latency has decreased.
Main features of the architecture:
-
25 CUDA cores — that's almost twice as much as the RTX 5090 and 35% more than the most powerful cards of previous generations. This configuration delivers tremendous performance in rasterization, geometry, and shader processing.
-
225 RT cores 4th generation is a new bar for ray tracing, supporting reflections, refractions, global illumination, soft shadows, and real-time noise reduction—all with minimal performance overhead.
-
Next-generation Tensor cores Provides hardware acceleration of DLSS 5, frame generation, scaling, and AI filtering. DLSS is now built into every rendering step.
-
512-bit bus and GDDR7 — the card is equipped with 32 GB of video memory with a throughput of up to 1.5 TB/s, which completely eliminates bottlenecks when working with high-resolution textures, including 8K.
-
Advanced power and cooling logic - power consumption up to 600 W will require an appropriate cooling system and power supply, but the architecture is focused on stable operation at high temperatures.
Preliminary characteristics
Parameter | GeForce RTX 6090 |
---|---|
Architecture | Ruby |
Technical process | 3nm (TSMC) |
CUDA cores | 25 000 |
RT cores | 225 |
Tensor cores | ~ 600 |
Video memory | 32 GB GDDR7 |
Memory bus | 512-bit |
Throughput | ~1500 GB/s |
Frequency Boost | ~3100–3300 MHz |
Energopotreblenie (TDP) | up to 600 W |
Interface | PCIe 5.0 x16 |
Support | DLSS 5, RT Gen4, AV1, DP 2.1, HDMI 2.1 |
Performance (forecast)
The RTX 6090 is expected to be 80% faster than RTX 5090 in tasks with ray tracing and using DLSS 5. According to approximate data:
-
4K Ultra +RT+DLSS5: 120–160 FPS in the heaviest games
-
1440p Ultra + Frame Gen: 200–300+ FPS in competitive titles
-
8K: stable 60 FPS with FSR or DLSS in visually intense games
-
AI and generative tasks: high efficiency thanks to tensor units and increased throughput
Strengths
-
Record number of CUDA and RT cores - absolute maximum
-
Hardware ray tracing and DLSS 5 built into the render chain
-
Memory bandwidth is not limited GPU even in 8K
-
Full stack of modern interfaces and codecs
-
Accelerated work with 3D, video, texture and scene generation
Restrictions
-
Power consumption of 600 W requires advanced liquid cooling and a power supply of 1200 W or more
-
High price - probably $2000 and up
-
Huge dimensions, incompatibility with compact cases
-
Full potential is only revealed in 4K and workloads
-
DLSS 5 still requires support from games
Conclusion
The GeForce RTX 6090 is not a graphics card for everyone. It is not for those who save money, and not for those who play in Full HD. This is a solution for enthusiasts, engineers, 3D designers and gamers who value absolute image quality, real-time ray tracing and performance for years to come. Rubin as an architecture takes a big step forward - and the RTX 6090 becomes its most powerful implementation.