Home » Intel will water-cool its best GPU to maintain temperature

Intel will water-cool its best GPU to maintain temperature

AMD and NVIDIA presented their most innovative solutions in terms of GPUs and with this we already know the Instinct MI250X and the GH100 and its H100 module in versions PCIe 5.0 and SXM5, but what about Intel? Well, the best semiconductor company in the world is facing delays, both in desktop and laptop versions and in HPC, but what they showed was the system of GPU competitor cooling named. Ponte Vecchio will be cooled with a massive block of water and…

Artificial intelligence and high performance computing will be the sectors of this Ponte Vecchio GPU which must also be shown as such in operation. As we surely know, Ponte Vecchio will have a whopping 47 different components in an interposer to house a total of no less than 100 billion transistors with numbers really hard to believe: 1 PFLOP in FP16.

Intel Ponte Vecchio, a very curious water block

Undoubtedly the worst will be the consumption, even if it is at the height of its competitors with 600 watts that they will need to be cooled with a liquid cooling pad which surprised everyone as we weren’t expecting a design like this.

The first design of this looked oddly more elaborate, it implemented two inputs and two outputs in a rather rustic and curious block, worthy of a laboratory no doubt, so the final version that Intel showed now must have been more advanced . And while it is so, it is so simple that it has caught the eye because of the low throughput that can be implemented for so many watts to dissipate.

The block is made in two parts, both in PVC, a single inlet and a single outlet, but with a diameter that seems to be between 10mm and 8mm IO.

All very sober and without fanfare, with an immense size in the case of a water block only for the Ponte Vecchio GPU and not a full cover, which shows that space is a priority with its straight lines and its rectangular shape , but it’s curious how Intel works with compact inside and outside diameters.

The explanation comes from the point of view of the structure of the network, where space is essential and the pipes must be small and flexible to guarantee its implementation. Therefore, here first higher water pressure a large flow/speed, since both inside and outside the inside diameter is negligible. The question now is how much will this GPU and its block cost and especially how close or far it is compared to NVIDIA and AMD.

VIA: Tom’s Hardware

About the author


Add Comment

Click here to post a comment

Your email address will not be published. Required fields are marked *