At the MUSA Developer Conference 2025, Chinese GPU manufacturer Moore Threads introduced its next-generation graphics architecture, Huagang, alongside two specialized chips: Lushan for gaming and Huashan for AI workloads.
According to the company, the new Lushan GPU delivers an unprecedented leap in performance over its predecessors. Key claims from the presentation include:
- 15x performance increase in AAA gaming titles.
- 50x faster Ray Tracing capabilities.
- 64x boost in AI task performance on consumer cards.
- 4x increase in available memory volume.
On the enterprise side, the Huashan chip is designed to tackle massive neural network training. Moore Threads claims a 10-fold increase in power efficiency and a 50% boost in compute density. The platform supports scalable clusters of over 100,000 modules with an interconnect bandwidth of 1,314 GB/s. In direct comparisons shown by the company, Huashan is positioned between NVIDIA’s Hopper and Blackwell architectures in terms of raw compute power, while reportedly offering superior memory capacity.
Conclusion from HYPERPC:
"The GPU market thrives on innovation and competition. While these claims are ambitious, the potential for a new high-performance player in the global arena is exciting. At HYPERPC, we represent the pinnacle of gaming and professional computing. We are closely monitoring how Moore Threads' new technology translates into real-world performance for gamers and content creators alike."
Verified Sources
- Tom's Hardware: Moore Threads unveils next-gen gaming GPU with 15x performance — (General context on Moore Threads roadmaps).
- TechNode: Moore Threads unveils new Huagang GPU architecture at MUSA Developer Conference — (Confirmation of Huagang/Huashan/Lushan names and stats).
- Pandaily: Moore Thread Unveils Next-Gen "Huagang" GPU Architecture — (Verification of the 100k cluster scalability and specific performance claims).
Arabic 
English