The NVIDIA GeForce RTX 5080, based on the Blackwell architecture, is the highly anticipated successor to the RTX 4080. Here's a roundup of everything we know so far about this enthusiast-grade GPU.
Architecture and Design
The RTX 5080 utilizes NVIDIA's Blackwell architecture, built on TSMC's 5nm (N5B) process node. It features the GB203-300-A1 die, succeeding the AD103-300-A1 found in the RTX 4080.
Specifications
Rumored specifications suggest a significant upgrade over the RTX 4080:
| Specification | RTX 5080 (Rumored) | RTX 4080 |
|---|---|---|
| CUDA Cores | 10,752 | 9,728 |
| Memory | 16GB/24GB GDDR7 | 16GB GDDR6X |
| Memory Speed | 32 Gbps | 22.4 Gbps |
| Memory Bandwidth | 1024 GB/s | 716.8 GB/s |
| TBP | 400W | 320W |
Expected Performance
Early rumors suggest the RTX 5080 could offer up to 10% better performance than the RTX 4090, especially in ray tracing. A significant performance boost over the RTX 4080 is expected, potentially enabling 60+ fps at 4K resolution in most games.
Price and Availability
Pricing for the RTX 5080 remains uncertain, but it's expected to be around $1000, similar to the RTX 4080 Super. NVIDIA is likely to reveal the RTX 5080, along with the RTX 5090 and RTX 5070, in early 2025, possibly at CES in January.
