The next generation of high-end graphics cards from Nvidia is shaping up to be a significant leap in both performance and power consumption, if recent leaks are to be believed. Industry insiders have revealed tantalizing details about the upcoming GeForce RTX 5090 and RTX 5080, suggesting that Nvidia is pushing the envelope in terms of raw computing power and memory bandwidth.
RTX 5090: A Monster in the Making
According to reputable hardware leaker @kopite7kimi, the flagship RTX 5090 is rumored to boast:
- 21,760 CUDA cores
- 32GB of GDDR7 memory
- 512-bit memory bus
- 600W Total Graphics Power (TGP)
These specifications represent a substantial upgrade over the current RTX 4090, with a 33% increase in CUDA cores and a 50% jump in power consumption. The move to GDDR7 memory also promises a significant boost in bandwidth, though exact speeds have not been disclosed.
 |
|---|
| This sleek design represents the innovative technology and performance capabilities expected from the upcoming Nvidia GeForce RTX 5090 |
RTX 5080: Bridging the Gap
The RTX 5080 is not to be overlooked, with its own impressive leaked specs:
- 10,752 CUDA cores
- 16GB of GDDR7 memory
- 256-bit memory bus
- 400W TGP
This represents a notable increase in power draw compared to its predecessor, the RTX 4080, which has a 320W TGP.
Performance Chasm and Pricing Concerns
One striking aspect of these leaks is the vast performance gap between the RTX 5090 and RTX 5080. The flagship model appears to have exactly double the CUDA cores and memory interface width of its smaller sibling. This disparity is even more pronounced than in the current generation, raising questions about Nvidia's product segmentation strategy.
This significant performance difference may also hint at potential pricing issues. Given the already high price point of the RTX 4090, consumers might be concerned about how Nvidia will position the RTX 5090 in terms of cost.
Technical Challenges and Solutions
The massive core count of the RTX 5090 presents technical challenges. Nvidia may be exploring two potential solutions:
- Multi-chiplet design: Using two GB203 dies interconnected with high-speed links.
- Monolithic die: A single, enormous chip manufactured on TSMC's 4nm-class process.
Both approaches have their pros and cons in terms of manufacturing complexity, yield rates, and cost.
AI Influence and Market Positioning
The emphasis on such high-performance specifications may be influenced by the booming AI market. Nvidia could be prioritizing data center needs and then adapting those designs for the consumer market. This strategy might result in even higher prices for top-tier consumer cards.
Launch Timeline
While specifications are leaking, the release date remains uncertain. Some sources suggest we might hear more about the RTX 50-series at the Consumer Electronics Show in early 2025, indicating a potentially long wait for enthusiasts eager to upgrade.
As with all leaks and rumors, it's important to approach this information with a degree of skepticism. Specifications can change before product launch, and Nvidia has not officially confirmed any details about its next-generation graphics cards. However, if these leaks prove accurate, PC gaming and professional GPU users can look forward to a significant performance boost – provided they can manage the increased power requirements and potential cost.