51WORLD(06651): SimOne4.0 has successfully completed systemic adaptation and deep optimization on the MTT S5000 GPU in Moore's Law thread.

51WORLD (06651) announced that in recent days, the next generation of intelligent driving from 51Sim and Siasun Robot & Automation's simulation platform SimOne4.0 has successfully completed systematic adaptation and deep optimization on the Moore Thread MTT S5000 GPU. This includes support for large model perception mining, 4DGS model training, 4DGS simulation reasoning, and synthetic data generation, fully supporting mass production in various intelligent driving technology routes such as end-to-end, VLA, and world models. This extends to training and intelligent upgrades in fields such as Siasun Robot & Automation, achieving a fully domestically produced stack of high-confidence closed-loop simulation and synthetic data in physical AI. 51Sim, as one of the company's three major businesses, is a leading synthetic data and simulation platform in China. Its core products include SimOne (intelligent driving simulation platform) and DataOne (data closed-loop platform). SimOne significantly reduces the testing and validation costs of intelligent driving systems, enabling comprehensive testing and verification of intelligent driving systems in various complex scenarios (such as different weather conditions, environmental conditions, and traffic dynamics), reducing the risks and costs of intelligent driving vehicle testing. The Moore Thread MTT S5000 is a full-featured GPU smart computing card designed for large model training, reasoning, and high-performance computing, built on the fourth-generation MUSA architecture "Pinghu." Its single-card AI dense computing power can reach up to 1000 TFLOPS, fully supporting full-precision calculations from FP8 to FP64. The successful adaptation of our company's SimOne4.0 and Moore Thread MTT S5000 GPU provides a complete domestic solution for the ecosystem of the intelligent driving industry (including OEMs, tier-one suppliers, testing agencies, etc.), demonstrating that domestically produced GPUs have entered a stage of high-precision and high-load tasks in autonomous driving. At the same time, the new version of the "Regulations on the Review of the Entry of Road Motor Vehicle Manufacturers" in 2026 has been upgraded, requiring intelligent driving-related vehicle companies to have verification capabilities in simulation, closed test sites, and actual roads, as well as safety evaluation capabilities. As advanced intelligent driving vehicles enter mass production, simulation testing is becoming a necessity before intelligent driving vehicles hit the road.