AMD has added Navi 31 and Navi 33 codes to ROCm development tools
AMD’s Navi 3x series GPUs seem to be still far away from consumers, but AMD has begun to make relevant preparations for developers. At present, in the ROCm development tool repository, AMD has added the code for the RDNA 3 architecture GPU, which are Navi 31 and Navi 33. The Navi 31 is positioned as a flagship product, while the Navi 33 is a mid-range product.
AMD ROCm is the first open-source software development platform for HPC/Hyperscale-class GPU computing. AMD ROCm brings the UNIX philosophy of choice, minimalism, and modular software development to GPU computing.
According to previously leaked information, Navi 31 will use 5nm process manufacturing and MCM multi-chip packaging, with 15,360 stream processors, Infinity Cache is 512MB, and the video memory bit width is still maintained at 256 bits, still GDDR6. At the same time, WGP will be used as the main computing module. Each computing chip in the Navi 31 core will have 30 WGPs, and each WGP will have 256 stream processors. This means that each computing chip can provide 7680 stream processors, so a Navi 31 core will reach the scale of 15,360 stream processors. In addition, Infinity Cache will be added to the MCD small chip, similar to the principle of using 3D V-Cache on the Zen 3 architecture, and manufactured using a 6nm process.
The specifications of Navi 33 are similar to the current Navi 21. There will be 5120 stream processors, Infinity Cache is 128 or 256MB, and the video memory width is 128 bits. It is rumored that the RDNA 3 architecture will surpass any NVIDIA product in traditional rasterization performance, and it is expected to debut in the third or fourth quarter of 2022.
