In case you didn't know, AI and high-performance computing are big business, and there's a reason Nvidia's market cap is over $1 trillion. That reason is the exploding demand for the company's enterprise products, including the powerful H100 Hopper GPU. Yes, this monster processor, which costs over $30,000, shares much of its DNA with the humble GeForce gaming graphics card.

YouTuber Geekerwan (via Tom's Hardware) ran some gaming benchmarks with one of these monster cards as a fun experiment.

The H100 used by Geekerwan is the PCIe version, with 80GB of HBM2e memory, 14,592 CUDA cores, and a 350W TDP; compare that to the RTX 4090 with 24GB of GDDR6X, 16,384 CUDA cores, and a 450W TDP Compare that to the H100, which should be hard to beat.

In fact, the card was so poor in games that its 3DMark Time Spy graphics score was only 2,681 points. This score is lower than the Radeon 680M integrated graphics. In Red Dead Redemption 2, the card could not even achieve 30 FPS at 1080p.

In all seriousness, these results are not surprising: the H100 is a very powerful card, but it is not designed for graphics applications. In fact, it does not even have a display output. The system needed a secondary GPU to provide a display. It also lacks some other fixed hardware important for gaming.

Additionally, there is the fact that the drivers are not optimized for gaming at all. During the gaming test, the GPU consumed less than 100 W, indicating a significant underutilization.

In other words, if you have $30,000 to spend, buy a car. Or buy an RTX 4090 system and use the money you have floating around to buy a car. The powerful H100 card is not for the ultimate gaming rig.