Core Insights - 3D Gaussian Splatting (3DGS) is an emerging method for novel view synthesis that utilizes a set of images with poses to iteratively train a scene representation composed of numerous anisotropic 3D Gaussian bodies, capturing the appearance and geometry of the scene [2][4] - The CLM system proposed by the team allows 3DGS to render large scenes using a single consumer-grade GPU, such as the RTX 4090, by addressing GPU memory limitations [6][8] Group 1: 3DGS Overview - 3DGS has shown revolutionary application potential in fields such as 3D modeling, digital twins, visual effects (VFX), VR/AR, and robot vision reconstruction (SLAM) [5] - The quality of images rendered using 3DGS depends on the fidelity of the trained scene representation, with larger and more complex scenes requiring more Gaussian bodies, leading to increased memory usage [5] Group 2: CLM System Design - CLM is designed based on the insight that the computation of 3DGS is inherently sparse, allowing only a small subset of Gaussian bodies to be accessed during each training iteration [8][20] - The system employs a novel unloading strategy that minimizes performance overhead and scales to large scenes by dynamically loading only the necessary Gaussian bodies into GPU memory while offloading the rest to CPU memory [8][11] Group 3: Performance and Efficiency - The implementation of CLM can render a large scene requiring 102 million Gaussian bodies on a single RTX 4090 while achieving top-tier reconstruction quality [8] - Each view typically accesses only 0.39% of the Gaussian points, with a maximum of 1.06% for any single view, highlighting the sparse nature of the data [23] Group 4: Optimization Techniques - The team utilized several unique characteristics of 3DGS to significantly reduce communication overhead associated with unloading, including pre-computing the accessed Gaussian sets for each view and leveraging spatial locality to optimize data transfer between CPU and GPU [12][17] - The microbatch scheduling optimization allows for overlapping access patterns between consecutive batches, enhancing cache hit rates and reducing redundant data transfers [24][25] Group 5: Results and Impact - CLM enhances the training capacity of 3DGS models by up to 6.1 times compared to pure GPU training baselines, enabling the training of larger models that improve scene reconstruction accuracy while lowering communication and unloading overhead [27]

Core Viewpoint - The company's 2024 performance met expectations, with a revenue of 5.57 billion and a net loss of 1.29 billion, while Q1 2025 exceeded expectations with a revenue of 2.02 billion and a net profit of 0.302 billion [1] Financial Performance - 2024 revenue was 5.57 billion, with a net loss of 1.29 billion, aligning with market forecasts [1] - Q1 2025 revenue reached 2.02 billion, surpassing the expected 1.45 billion, with a net profit of 0.302 billion, exceeding both internal and market expectations [1] - The company announced a cash dividend of 0.23 yuan per share for 2024 [2] Growth Drivers - Q1 2025 revenue growth was driven by a 10% increase in gaming revenue, attributed to the performance of "Zhu Xian World" and esports titles [3] - The film and television segment saw a significant revenue increase of 1616% in Q1 2025, primarily due to the confirmation of multiple TV series revenues [3] - Future focus includes the mid-year major update for "Zhu Xian World" and the planned overseas launch of "P5X" in Japan and Europe [3] Cost Management - The company effectively controlled costs, with sales and management expenses decreasing by 27% and 44% respectively in Q1 [4] - R&D expenses remained stable year-on-year, with expectations of a slight decrease in subsequent quarters due to improved internal management and cost control [4] Profit Forecast and Valuation - The net profit forecast for 2025 has been raised by 9.5% to 640 million, while the 2026 forecast remains unchanged [5] - The current stock price corresponds to 37/16 times the P/E for 2025/26, with a target price increase of 11% to 15.2 yuan, indicating a potential upside of 24% [5]