Nvidia RTX 5000 Ada AD102 Advantages

Suddenly Open: Nvidia’s RTX 5000 Ada AD102 with 32GB GDDR6

The AD102-based graphics board with its dimensions reduced is now available.

Nvidia’s partners have secretly begun selling graphics cards from Nvidia’s RTX 5000 Ada Generation lineup. These cards are intended for use in applications related to professional visualisation (ProViz). The AD102 GPU, which is the company’s flagship product for use in client computers, is integrated into the board in a configuration that has been significantly scaled down in order to bring the board’s power consumption down to 250W. PNY’s Nvidia RTX 5000 Ada boards are being sold at significantly inflated pricing by several merchants, which comes as a surprise.

The RTX 5000 Ada graphics card is now available for purchase from a variety of workstation-oriented shops, including PC Connection, eBay, ProVantage, ShopBLT, and ThinkMate, amongst others. However, none of them are presently selling this add-in board at the manufacturer’s suggested retail price of $4,000; instead, their prices are ranging between 10 and 20 percent over this amount. ProVantage even goes as far as to say that the card has a list price of $6,999 and that you are saving 37% by purchasing it from them. Good effort.

The RTX 5000 Ada Generation graphics card from Nvidia is based on the AD102 graphics processor and has a total of 12,800 CUDA cores that are distributed over 100 SMs. It has a maximum compute capability of 65.3 FP32 TFLOPS and is paired with 32 GB of GDDR6 memory over a 256-bit interface. Additionally, it has a large amount of memory.

In comparison, Nvidia’s top-of-the-line RTX 6000 Ada model, which is also powered by the same AD102 GPU, brags that it has 18,176 CUDA cores and is capable of an incredible 91.6 FP32 TFLOPS. In addition to that, the RTX 6000 is equipped with 48 gigabytes of GDDR6 RAM that is linked to the CPU through a 384-bit bus.

It may seem strange that thirty percent of the GPU was reduced in size, but it’s possible that this was the only option to design a product that would sit below the RTX 6000 Ada and yet give a performance disparity that’s apparent but not dramatic when compared to the most powerful ProViz graphics card available. The entire AD103 from Nvidia has a maximum of 10,240 CUDA cores that are split over 80 SMs; hence, its theoretical performance would be much lower than that of an AD102 with its core count reduced.

It is interesting to note that despite making significant changes to the GPU design, Nvidia has kept the part’s power rating at the relatively high level of 250W. Regarding the cooling system, the motherboard has a dual-slot cooler that only has one fan attached to it. Additionally, a 12VHPWR PCIe auxiliary power connection is used by the board.

In the meanwhile, it is still unknown which strategy Nvidia will choose to use in order to make use of the AD103 processor for its professional product offerings. The RTX 6000 Ada employs almost all of the available resources of the AD102, including 142 of the possible 144 SMs and the whole 384-bit memory interface. AD102 is used in the RTX 5000 Ada, which has 100 SMs but only a 256-bit interface.

The next step down is a significant one to the RTX 4500 Ada, which has a complete AD104 implementation with 60 SMs and a 192-bit interface. The RTX 4000 Ada and RTX 4000 SFF Ada both use a partial AD104 chip that is comprised of 48 SMs and a 160-bit interface to bring everything to a close.

To put it another way, there is space for an RTX Ada card to be placed between the 4500 Ada and the 5000 Ada, namely with the AD103. However, Nvidia has not yet made an announcement on such a product, and there is no space for it inside the conventional naming system (i.e., the RTX 5500 Ada would be the name of this card). It’s possible that Nvidia will completely avoid utilising AD103 on professional cards by doing nothing as all.

News source