Shulou(Shulou.com)11/24 Report--

CTOnews.com October 1, the new "leather jacket" is ready, NVIDIA announced that the annual GTC developer conference will be held at the San Jose Convention Center in California from March 18 to 21 next year.

Considering the jet lag and the Nvidia practice, Beijing time is likely to be 23:00 on March 18. It is worth mentioning that Nvidia has been holding online digital conferences since 2020, so GTC 2024 will return to San Jose on behalf of Nvidia.

According to reports, the live meeting will bring together thousands of innovators, researchers, creators, thought leaders and decision makers to share ideas, discuss innovative strategies, and explore how accelerated computing and AI can change the way we work and live.

Traditionally, Nvidia will release its new generation of graphics technology on GTC and will bring a new generation of data center products.

Given the latest revelations, everyone at GTC should focus on the much-anticipated Blackwell series, which will also be extended to games, although the underlying architecture of game chips will be different.

CTOnews.com Note: Blackwell, the first black academician of the American Academy of Sciences and the first black tenured professor at the University of California, Berkeley, died of illness on July 8, 2010 at the age of 91.

Image source Pexels is worth mentioning that Kopite recently shared information about Nvidia's next-generation GeForce and data center GPU.

In the code name Blackwell product line, GB20X is mainly for consumer-grade GeForce RTX graphics cards, while GB10X is mainly for data centers.

The GB100 source Kopite indicates that there are 8 groups of GPC in GB100, and each group of GPC contains 10 TPC. It is speculated that there are 2 SM in each TPC, which is equivalent to 160 SM.

Each set of GPC contains 20 SM, assuming that each SM can contain 128 CUDA cores, then the GB100 may contain 20480 CUDA / FP32 cores.

The GB202 message source Kopite indicates that the GB202 has 12 groups of GPC, each group of GPC contains 8 TPC, and each TPC contains 2 SM, a total of 192 SM, equivalent to 24576 CUDA cores.

Kopite reports that GB100 and GB202 will include 8192-bit and 512-bit memory buses, respectively. CTOnews.com attaches the following information:

GB100 (data Center): 160 SM | 20480 CUDA cores | 8192-bit memory bus

GB202 (Game): 192 SM | 24576 CUDA cores | 48 GB G7 video memory

Since none of the above chips are taped, it is difficult to determine the performance. The initial leak touted RTX 5090 was 1.7 times faster than RTX 4090 and could not be verified.

Related readings:

"the next generation of Nvidia flagship graphics card RTX 5090 exposure: 3nm process, code name Blackwell"

"it is reported that Nvidia has tested Samsung 3nm GAA process and will produce as soon as 2025."

"it is reported that the number of CUDA cores of Nvidia RTX 5090 graphics card is 50% more than that of the previous generation, and the overall performance is improved by 70%."

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.