List of GPUs with GDDR7 Memory – Graphics from Nvidia and AMD

The upcoming graphics processing units (GPUs) equipped with GDDR7 memory, specifically focusing on the Nvidia GeForce RTX 5080 and RTX 5090 models.

These next-generation GPUs are set to deliver unprecedented levels of performance, with enhancements in memory bandwidth, CUDA core counts, and architectural advancements.

The RTX 5080 and RTX 5090 are designed to meet the increasing demands of high-end gaming, professional rendering, and AI-driven applications, offering significant improvements over their predecessors in terms of speed, efficiency, and visual fidelity.

Nvidia GeForce RTX 5080: The Nvidia GeForce RTX 5080 is expected to launch in the fourth quarter of 2024 with a possibility of showcasing it at CES in January 2025. It will feature GDDR7 memory at 32 Gbps. This new GPU builds on the success of its predecessors by offering significant enhancements in features, performance, and visual quality, particularly in ray tracing. This is due to its specialized RT cores which create very realistic visuals and reflections.

Key Improvements:

• The RTX 5080 might boost performance by up to 30% compared to previous models, allowing for quicker rendering and smoother gameplay in demanding applications.
• It includes an increase in RT cores and tensor cores, improving rasterization and AI capabilities through features like DLSS and frame generation.
• The new GPU will have more CUDA cores and a higher clock speed, contributing to its enhanced gaming and rendering abilities.
• It also features a new structure for streaming multiprocessors (SMs) and more VRAM, improving overall efficiency and performance.

Specifics on the Hardware:

• The RTX 5080 will have a 192-bit bus width and provide a bandwidth of 768GB/s, a slight increase from the RTX 4080’s 717GB/s.
• It is likely to consume between 250W and 300W, which is less than the RTX 4080, while offering up to 20% more CUDA cores.

Gaming and VR:

• The RTX 5080 is expected to deliver enhanced ray tracing capabilities and better support for VR applications, with the potential for 8K gaming.

Compatibility:

• This GPU should fit well with most current motherboards and power supplies, but it’s advised to check the manufacturer’s specs to ensure compatibility.

Nvidia GeForce RTX 5090: The release of the Nvidia GeForce RTX 5090 is anticipated by the end of the year. It is expected to include GDDR7 VRAM operating at 28 Gbps, which could potentially reach up to 48 Gbps.

Hardware Capabilities:

• It might feature a 512-bit memory bus, significantly increasing the memory bandwidth compared to the RTX 4090.
• The expected memory bandwidth is 1536 GB/s, a 50% improvement from the RTX 4090, which aligns with a nearly 60% increase in shader density.

Blackwell Architecture:

• The RTX 5090 will likely introduce the new Blackwell architecture, which could feature a multi-chiplet design rather than a single monolithic chip. This could allow for denser configurations and potentially a significant performance leap.

Specifications:

• The RTX 5090 is projected to have up to 24,576 CUDA cores and 204 Streaming Multiprocessors, leading to an overall core count of 26,112.
• It could come with a 128MB cache, although it’s not specified if this is a shared cache or specific to different cache levels.

Performance:

• The GPU’s boost clock could reach up to 2.9 GHz, and it’s designed to be about 70% faster than the RTX 4090 in terms of overall performance.

Overall, both GPUs represent significant advancements in graphics technology, promising substantial improvements in gaming, rendering, and AI applications.