Nvidia RTX 50-Series
NVIDIA’s RTX 50 series indicates that the release has been pushed back to early 2025.
RTX 5090
CUDA Cores
Memory
Memory Bus
Memory Speed
Bandwidth
Power Consumption
Architecture
RTX 5080
CUDA Cores
Memory
Memory Bus
Memory Speed
Bandwidth
Power Consumption
Architecture
RTX 5070 ti
CUDA Cores
Memory
Memory Bus
Memory Speed
Bandwidth
Power Consumption
Architecture
21,760 CUDA cores
32GB GDDR7 VRAM
512-bit
up to 36 Gbps
1.728 TB/s
600W power
Blackwell architecture
10,752 CUDA cores
16GB GDDR7 VRAM
256-bit memory bus
32 Gbps memory speed
1 TB/s
400W power
Blackwell architecture
Approximately 8,960
16GB of GDDR7 VRAM
256-bit
Up to 32 Gbps
Up to 896 GB/s
300W
Blackwell architecture
Nvidia RTX 50-Series
As of December 31, 2024, the latest information on NVIDIA’s RTX 50 series indicates that the release has been pushed back to early 2025. Here’s what we know:
RTX 5090 and RTX 5080: These are expected to be the first models to launch in early 2025. They will be based on the Blackwell architecture and are anticipated to offer significant performance improvements over the previous generation.
RTX 5070 and RTX 5060: These models are scheduled to be released in early to mid-2025. They are also part of the Blackwell architecture and are expected to offer varying degrees of performance enhancements.
Blackwell Architecture: The RTX 50 series will feature the new Blackwell architecture, promising up to 2.5 times the performance of the 4000 series, with substantial improvements in ray tracing and DLSS 3.5 technology.
CUDA Cores and Memory: Rumors suggest that the RTX 5070 will have 6,400 CUDA cores and 12GB of GDDR7 memory, with a memory bandwidth of 672GB/s.
Additional Models: There are rumors about the RTX 5070 Ti and RTX 5050, though details on these models are less concrete. The RTX 5070 Ti is expected to have 1,200 more CUDA cores than the RTX 4070 Ti.
Release Dates: The initial rumors suggested a fall 2024 launch, but recent updates indicate a delay until CES 2025.