ISO: Overlap of Computation and Communication within Seqenence For LLM Inference
This addresses resource underutilization in multi-GPU LLM inference, offering a more efficient method for practitioners, though it is incremental over prior overlap techniques.
The paper tackles the inefficiency of sequential computation and communication in LLM inference by introducing a novel sequence-level overlap strategy, which reduces time consumption by approximately 35% on 4090 GPU and 15% on A800 GPU during the prefill stage.
In the realm of Large Language Model (LLM) inference, the inherent structure of transformer models coupled with the multi-GPU tensor parallelism strategy leads to a sequential execution of computation and communication. This results in substantial underutilization of computing resources during the communication phase. To mitigate this inefficiency, various techniques have been developed to optimize the use of computational power throughout the communication process. These strategies primarily involve overlapping matrix computations and communications, as well as interleaving micro-batches across different requests. Nonetheless, these approaches either fall short of achieving ideal overlap or impose certain limitations on their application. To overcome these challenges, this paper introduces a novel strategy for computation-communication overlap that operates at the sequence level. This method not only enhances the degree of overlap but also minimizes the constraints on its applicability. Experimental evaluations conducted using 30b/70b models have demonstrated significant improvements in efficiency. Specifically, the proposed technique has been shown to reduce time consumption by approximately 35% on 4090 GPU and by roughly 15% on A800 GPU during the prefill stage of LLM inference.