New reports provide an insight into the technical design of the NVIDIA GeForce RTX 5090, the flagship GPU of the upcoming “Blackwell” generation. In addition to information on the power supply phases, details on the printed circuit board (PCB) structure have been revealed, providing clues to the basic architecture of the GPU.
Power supply design and PCB details
According to a report by Benchlife, the GeForce RTX 5090 uses a 16 6 7 power phase design. This configuration differs from the RTX 4090, which was equipped with a 20 3 design. Current phase designs regulate the power supply to the processor and GPU, which has an impact on stability and efficiency. The PCB of the RTX 5090 is reportedly built in 14 layers. This structure allows for the integration of more components and can improve thermal and electrical performance. It is likely that this configuration is limited to the reference design, while third-party models may differ. Additionally, the RTX 5090 utilizes the new 12V-2×6 connector, which replaces the previous 12VHPWR connector. This connection meets the requirements of the PCIe 5.0 standard, which offers higher performance with improved signal stability.
Expected technical specifications
The following specifications are based on current reports and rumors:
- GPU chip: Blackwell GB202-300
- Memory: 32 GB GDDR7
- Power consumption: 600 Watt TDP
- Current phase design: 16 6 7
- Power connector: 1x 12V-2×6 (16-pin)
- Interfaces: PCIe 5.0, DisplayPort 2.1a
The memory of the RTX 5090 uses GDDR7 technology, which enables a higher data transfer rate compared to GDDR6X. This could have a particularly positive effect on graphics-intensive applications. The power consumption of 600 watts indicates the need for suitable cooling solutions to keep heat development under control.
Advances in design and thermal management
Initial images of the PCB design show an enlarged heatspreader (IHS). This component protects the GPU core and distributes the heat more evenly, which can contribute to more stable temperatures. The GPU itself is based on the new Blackwell chip, which uses smaller production nodes and optimized architectures. The combination of optimized power supply, PCIe 5.0 support and DisplayPort 2.1a interfaces aims to ensure compatibility with modern hardware and displays. It remains to be seen how these specifications will affect real-world applications such as gaming or professional workloads.
Release and outlook
The NVIDIA GeForce RTX 5090 is expected to be launched in late January or early February 2025. Further details on the RTX 50 series are expected at CES 2025, which will take place in a few weeks. While exact pricing and market availability have not yet been confirmed, the focus remains on developing energy-efficient and powerful GPUs. Initial tests after the market launch will show whether the planned innovations prove themselves in practice.
Sources: Videocardz
8 Antworten
Kommentar
Lade neue Kommentare
Urgestein
Veteran
Urgestein
Urgestein
Veteran
Veteran
Mitglied
Mitglied
Alle Kommentare lesen unter igor´sLAB Community →