The Chinese company hopes to launch its first GPGPU in 2025; but Nvidia’s H100 is safe for now, as it’s 67x faster than its new AI challenger
Known in China as the granddaddy of domestic CPUs, Loongson is developing processors and chips in an effort to reduce the country’s dependence on foreign technology, especially companies like Intel and AMD.
While the company is probably best known for its Loongson series of processors (originally based on MIPS architecture), the company is expanding and has unveiled its first proprietary GPGPU (General-Purpose GPU), the Loongson 9A1000.
Announced at the recent Loongson Industrial Ecosystem Conference, and reported by Fast technology9A1000 aims to deliver competitive performance for AI computing in edge devices and unmanned systems. However, tech giants like Nvidia can breathe easy for now, as the Loongson 9A1000’s performance is expected to match that of the AMD RX 550, a graphics card best known for its 50W efficient power consumption and PCIe 3.0 interface, which was originally was launched in 2017.
Arrival next year
The new Loongson chip is not as dated as AMD’s card. It supports PCIe 4.0 and integrates LPDDR4X memory with a 128-bit interface. While detailed specifications such as core count, frequency and power consumption have yet to be revealed, an architecture diagram shown at the conference reveals eight compute arrays, an on-chip network and L2 cache.
In terms of performance, the 9A1000 supports OpenGL 4.0, OpenCL 3.0 and other standard APIs. It also includes video processing modules that can decode H.264 and H.265 and offers HDMI 2.1, DisplayPort 1.4 and, in another nod to the past, VGA output. The chip features a pixel fill rate of 16GPixel/s, a texture fill rate of 32GTexel/s, and a computing power of 1TFlops for FP32, 64GFlops for FP64, and 32TFlops for INT8 operations.
The Loongson 9A1000 is expected to complete the tape-out process and begin sampling in the first half of 2025.
Loongson also announced that it is developing its next-generation GPGPU chip, the Loongson 9A2000, which promises an 8 to 10x performance increase compared to the 9A1000. The company expects the new chip to deliver performance comparable to NVIDIA’s RTX 2080, which was launched in 2018.