A few months ago, popular enthusiast der8auer took a look at the RTX Titan Ada, an unreleased card from Nvidia's Ada Lovelace lineup that was supposed to take full advantage of the AD102 GPU, purportedly featuring 18,432 cores and 48 GB VRAM. While this card never made it to market, a working prototype of it featuring dual 12VHPWR connectors does exist. And, today, we get to see the insane adapter that ships with it to make the whole thing compatible with older power supplies (that don't adhere to the ATX 3.0/3.1 spec).
The Titan series from Nvidia always represented the best-in-class mainstream GPU the company had to offer. Since 2018's RTX Titan, Nvidia has replaced this with its 90-class GPUs like the RTX 3090 and 5090. The RTX Titan Ada was, therefore, supposed to be the flagship offering of the Ada Lovelace lineup, superceding even the RTX 4090. It was the largest reference card Nvidia had ever built, but alas, we never got to enjoy the brutality of it.
All that performance requires a lot of power, and this card came well-equipped. With not only one but two 12VHPWR connectors, the Titan Ada can technically handle up to 1200W of power, but due to driver and BIOS limitations, it only peaked at around 600W (with overclocking) when der8bauer tested it earlier. Now, in his latest video, he specifically breaks down this power delivery system and, most interestingly, the mystery dual 12VHPWR to 6x 8-pin adapter.
As the name suggests, Nvidia made an adapter that can take the power from six standard 8-pin PCIe connectors, so 900W, and supply that across two 12VHPWR connectors. This adapter has no load balancing, and there are no sense pins on the connector either, so the GPU will run regardless of how many 8-pins are actually plugged in. Der8auer demonstrates this by literally unplugging the 8-pin connectors one by one while the card is running, and it doesn't shut down even when down to just two 8-pins.
Design-wise, the Titan Ada adapter, despite featuring an angled housing, is much stiffer compared to the newer 12V-2x6 to quad 8-pin adapter that ships with the RTX 5090. There's no smart functionality in that either. Still, at least the 5090's adapter has an extra communication wire that can detect whether every 8-pin is properly plugged in or not—unlike the Titan Ada, which will just run willy-nilly. However, this behavior could be attributed to its prototype nature. The Titan adapter also features much longer cables that are not braided.
Der8auer otherwise praised the construction of the adapter, noting how it's all one piece of plastic instead of smaller bits attached together. That could indicate Nvidia was quite far into the process of finalizing it, and this is proven by the date of manufacture. Der8auer disassembled the 12VHPWR side of the adapter and found "20220121" mentioned on the PCB—corresponding to 21 January, 2022—and reiterated that Nvidia likely made a new mold specifically for the connectors.
That date tells us this adapter existed just around 9 months before the launch of the RTX 4090, which again suggests that Nvidia was seriously considering this for the mass market. What ultimately stopped them at the end remains unanswered. It's no question that eventually we will see a flagship card that needs more than 600W, and that may be when Nvidia will bring these dual 16-pin connectors back, but with the updated 12V-2x6 designs instead.
Follow Tom's Hardware on Google News to get our up-to-date news, analysis, and reviews in your feeds. Make sure to click the Follow button.
