Yesterday Intel announced its new Intel Data Center Max 1000 GPUs and it’s now known that they will feature the 16-pin 12VHPWR connector. The ATX 3.0 connector is currently exclusive to NVIDIA until 2023 but will be included with Intel’s data center GPUs when they launch next year. Intel has said that the Max Series 1100 GPU is a 300 Watt card, which means it may use only as much as half of the 600 Watt max limit specification for the 12VHPWR connector.
An introduction video showcasing the physical design of the new Intel GPUs clearly displays the connector on the PCB, and from the rear of the card, in an example of pairing multiple GPUs.
“We’re pleased to announce and spotlight our newest GPU, the Intel Data Center GPU Max Series. Formerly codenamed Ponte Vecchio, Intel Data Center GPU Max Series offers the industry’s highest-density processor with more than 100 billion transistors on 47 active tiles and up to 128 Xe-HPC cores. Enjoy this close-up trailer to see how it looks.“
12VHPWR issues
The 12VHPWR connector has recently come under scrutiny though. Numerous reports of NVIDIA adapters, and graphics card connectors, melting, have surfaced since the launch of the GeForce RTX 4090. NVIDIA is currently investigating the matter but has not issued a statement on it. Meanwhile, AMD’s Scott Herkelman (SVP & GM of AMD Radeon) has confirmed that it will not be using the new connector for its upcoming Radeon RX 7000 Series graphics cards. It is presently not known if Intel plans on adopting this connector for future releases of its Alchemist line of consumer graphics cards. There have also been suggestions from PCI-SIG experts that a revision of the adapter could be in the works and that user error may be to blame for some of these reports. However, these suggested revisions would be for the four data “sense” pins and not the pins used for power delivery.