- Asus RTX 5080 is now coupled with AI equipment and an SSD bay surprisingly cool
- The PCIe tracks are no longer only for GPUs, thanks to the hybrid material approach of Asus
- Asus develops a multi-LM so that AI developers can unload the models directly to SSD on board
The unusual GPFORCE RTX 5080 GPUROCE, spotted for the first time a few weeks ago with an SSD slit, is no longer just a hardware quirk.
Asus North America has now confirmed that the SSD Proart RTX 5080 edition is a real product.
As unusual as it may seem, the decision to associate high performance graphics treatment with solid state storage is part of a wider plan involving artificial intelligence and the development of local models.
A hybrid GPU storage design targeting the development of AI
Although the Asus was tight on an release date and a shipping configuration, certain technical details have started to emerge, as well as some unanswered questions.
One of the largest surprises is the GPU arrangement turned to 90 degrees – Asus has changed the design of the printed circuit card (PCB) so that the GPU is mounted perpendicular to the usual orientation.
This is not a configuration found in any other RTX 5080 model, suggesting a personalized approach that adapts to the SSD and improves air flow.
Asus claims that this provision helps the SSD to benefit from “Airflow provided by card fans”, although the SSD is not in direct contact with the main thermal dissipator.
Rather, it is cooled via a secondary thermal dissipator and a passive air flow, keeping the SSD at around 10 degrees Celsius fresher than a typical M.2 training mounted on a motherboard.
The SSD integration also uses the pccie bifurcation, which means that the unused PCI ways are reassigned to support the reader.
The company previously applied this approach to the cards of the RTX 4060 series, which did not completely saturate the PCIE interface, leaving enough bandwidth for a M.2 location.
Now, with the RTX 5080 PROART edition, the idea is refined to adapt to high-end IA workflows, although it introduces compromises, the GPU may not always have access to all PCIE tracks, which potentially limits advanced performance under certain workloads.
Perhaps the most important revelation lies in the ambitions of Asus software-the company is developing a tool called multi-LM, intended for developers working with important languages of languages.
Asus says that this tool “will allow AI developers to unload the models directly in storage, potentially allowing the development of local models without the need for constant memory swaps to slower system readers.”
This will interest those looking for the best LLM for coding, and positions the RTX 5080 SSD edition as more than a simple game card; It could become one of the best AI tools for local inference and prototyping.
Asus published only one rendering of the card and has not confirmed performance measures or availability.
The question of whether this concept provides significant gains in the real world or simply adds complexity remains unanswered.
Via video
