Analog compute-in-memory (CIM) enables energy-efficient AI inference by performing in-situ matrix multiplication. The growing demand for deploying larger models motivates chiplet integration to expand CIM capacity. However, simulation abstractions based on local operator costs and aggregate communication metrics obscure the impact of weight-mapping topology when CIM system scales up. We develop a phase analysis that quantifies how strongly mapping topology affects system performance across hardware configurations and how mapping optimization translates into latency gains. To evaluate and exploit this design space, we present Isobar, a simulation platform integrating CIM decompositions, heterogeneous execution, and resource modeling. Validated against silicon measurements and cycle-accurate simulation, Isobar provides reliable cost estimates that enable effective mapping search across model and hardware configurations. For Gemma-7B prefill, Isobar-guided mapping reduces communication time by 23.3% and end-to-end latency by 18.6% relative to the strongest baseline. Isobar's system-level modeling show that analog CIM retains energy-efficiency potential relative to GPU when scaled to thousands of dies, and identify topology-aware mapping as a key design dimension for efficient execution.
Vimal William, Yuanzhe Li, Danella Zhao, Huanrui Yang
Parameter-efficient linearization turns a pretrained autoregressive Transformer into an efficient model by hybridizing sliding-window attention (SWA) with a constant-size recurrent memory. But does every layer's recurrent state contribute to long-context retrieval? Investigating an SWA and Test-Time Training (TTT) hybrid, we find that retrieval is sparse across depth, concentrated in a few anchor layers, and that the remaining states hold similar information expressed through layer-specific transformations. We exploit both observations in Anchor: TTT memory placed only at the anchor layers, and a cross-layer recurrent state written once and read by many layers, each through its own learned read map that realizes a layer-specific stretch and rotation, along with demonstrated generalization to other hybrid attentions. On Llama-3.1-8B and Mistral-7B, Anchor delivers 2.3-3.1x higher recall and 6.7-8.9 points higher MMLU accuracy than LoLCATs with a 10x smaller recurrent state, retains 91-95% of the teacher's MMLU accuracy, and cuts end-to-end latency by up to 4.3x relative to softmax attention.
Yuanzhe Li, Jianing Deng, Jingtong Hu, Tianlong Chen, Song Wang, Huanrui Yang
Large language models (LLMs) achieve strong performance in long-horizon agentic tasks through multi-step interaction and reasoning at test time. However, repeatedly invoking these models incurs substantial inference cost over an episode. Post-training quantization (PTQ) reduces this cost, but using a fixed low precision throughout an episode can sharply lower task success. We observe that interaction steps differ sharply in their sensitivity to quantization: given the same context, decoding a step at low precision turns a success into a failure at only about 1% of the steps, which we call critical steps. Based on this observation, we propose Dynamic Mixed-Precision Routing (DMR), a framework that adaptively selects between high-precision and low-precision versions of the same LLM at each decision step. Leveraging an entire-space factorization to jointly predict a step's divergence and its effect on the outcome, DMR learns to recognize these critical steps by training only a lightweight router on the model's hidden state. By reusing a shared high-precision KV cache and the hidden states, DMR avoids redundant context prefill and a separate text encoder, enabling precision switching with less than 1 ms of overhead per step and adding less than 0.11% to step latency for routing. This mixed-precision design also enables nested quantization, reducing the memory footprint by 39% on average compared with FP16. Across three long-horizon agentic benchmarks with nested quantization, DMR achieves an average 2.44x decode speedup over FP16 at comparable task success rates, without fine-tuning the underlying model. On ALFWorld with GPTQ, DMR achieves a 1.71x decode speedup with 86.9% success, compared with 87.2% for FP16.
Dongwei Wang, Jinhee Kim, Seokho Han, Denis Gudovskiy, Yohei Nakata, Tomoyuki Okuno, KhayTze Peong, Kang Eun Jeon 等 11 位作者
Dynamic runtime latency and memory constraints necessitate flexible large language model (LLM) deployment, where an LLM can be inferred with various quantization precisions based on available computational resources. Recent work on such any-precision quantization either relies on hardware-inefficient vector quantization or induces additional scaling factors when switching between bit-widths. Meanwhile, existing post-training quantization (PTQ) methods calibrated for a fixed low precision show poor generalizability under runtime precision change. In this work, we attribute the source of poor generalization across bit-widths to a precision-dependent outlier migration phenomenon where the distribution of PTQ-sensitive tokens changes across precisions. Motivated by this observation, we propose MoBiQuant, a novel any-precision Mixture-of-Bits quantization framework that adjusts weight precision for flexible LLM inference based on token sensitivity. Specifically, we propose a many-in-one recursive residual quantization that can iteratively reconstruct higher-precision weights at runtime and mitigates outlier migration with a token-aware router to dynamically select the optimal inference precision of each token.Extensive experiments show that MoBiQuant matches or surpasses frontier single-precision PTQ while exhibiting strong elasticity, achieving significant memory savings and throughput gains of up to $1.34\times$ over state-of-the-art any-precision methods. Our code is available at https://github.com/SimWangArizona/MoBiQuant.