--- title: 'LMM-IR: Netlist-Aware IR-drop Prediction' url: https://www.emergentmind.com/topics/lmm-ir type: topic --- # LMM-IR: Netlist-Aware IR-drop Prediction LMM-IR is a netlist-aware multimodal deep-learning framework for static IR-drop prediction in large-scale chip designs. The approach integrates transformer-based representations of netlist topology, processed as 3D point clouds, with conventional 2D circuit property maps. This architecture enables efficient, accurate prediction of per-pixel voltage drop (static IR-drop) across entire power distribution networks (PDNs) for circuits with hundreds of thousands to millions of nodes, substantially reducing analysis turnaround time while matching or exceeding prior state-of-the-art predictive performance [2511.12581]. ## 1. Architecture and Data Modalities LMM-IR consists of four major components: a Circuit Encoder, a Large-scale Netlist Transformer (LNT), a Multimodal Fusion Module, and a Decoder for IR-drop prediction. The input pipeline processes both the SPICE netlist (representing circuit connectivity and device attributes) and 2D circuit property maps, such as current demand, PDN density, effective distance, voltage-source plots, and resistor-density. The netlist is transformed into a sparse 3D point cloud reflecting inter-node electrical/physical properties. All input modalities are embedded into a high-dimensional latent space for integration via cross-modal attention before spatial upsampling yields the predicted voltage drop map. The architectural data flow is summarized by the following block diagram (ASCII form from the original source): ``` ┌────────────────┐ ┌──────────────────────────────┐ │ SPICE Netlist │─┬──────▶│ Netlist Preprocessor │─┬──▶ LNT ──┐ └────────────────┘ │ └──────────────────────────────┘ │ │ │ │ │ │ ┌────────────────┐ │ ┌──────────────────────────────┐ ▼ │ Current Map, │─┼──────▶│ Circuit Feature Preprocessor│───┼─▶ Fusion ──▶ Decoder ──▶ IR-drop map │ PDN Density, │ │ └──────────────────────────────┘ │ │ Eff. Distance │ │ │ └────────────────┘ │ │ └───────────────────────────────────────────┘ ``` This schema reflects the dual-stream embedding of spatial (image-like) and topological (point-cloud) data, followed by joint representation learning and upsampling. ## 2. Netlist Representation via Large-scale Transformers The key innovation is encoding netlist topology as a large-scale point cloud suitable for self-attention-based neural architectures. Each netlist element is represented as an 8-dimensional point: \[ p_i = [x_{i,1}, y_{i,1}, x_{i,2}, y_{i,2}, v_i, t_i, L_{i,1}, L_{i,2}] \in \mathbb{R}^8 \] with $x_{i,1}, y_{i,1}, x_{i,2}, y_{i,2}$ spatial coordinates of node terminals, $v_i$ the element value (resistance, current, or voltage), $t_i$ a one-hot-encoded device type, and $L_{i,1}, L_{i,2}$ the metal layer indices. The full point cloud is embedded to dimension $d$ by a trainable linear or MLP layer. Stacked point embeddings $E^{(0)} \in \mathbb{R}^{N\times d}$ (for $N$ elements) feed into an $L$-layer transformer with standard scaled dot-product self-attention and residual feed-forward blocks: \[ Q = E^{(\ell-1)}W_Q,\quad K = E^{(\ell-1)}W_K,\quad V = E^{(\ell-1)}W_V \] \[ A = \mathrm{Softmax}\left(\frac{QK^T}{\sqrt{d_k}}\right),\quad H = AV \] \[ E^{(\ell)} = \mathrm{LayerNorm}(E^{(\ell-1)} + \mathrm{Dropout}(H)) \] \[ E^{(\ell)} \leftarrow \mathrm{LayerNorm}\big(E^{(\ell)} + \mathrm{Dropout}(\mathrm{FFN}(E^{(\ell)}))\big) \] This approach scales to $N\sim 10^5$–$10^6$ points by exploiting the sparsity inherent in netlists, maintaining manageable memory scaling $\mathcal O(Nd)$. ## 3. Multimodal Fusion and Joint Embedding The circuit property maps are processed by a downsampling convolutional encoder (with four strided blocks, U-Net style), yielding a feature tensor $\mathbf F_\mathrm{circ}\in \mathbb{R}^{d_\mathrm{circ}\times h\times w}$. This is flattened to a sequence of $M=h\cdot w$ tokens of dimension $d$. Each layout is padded or rescaled to $512\times512$ spatial resolution. Multimodal fusion is achieved by mutual cross-attention between the netlist and circuit feature tokens. For example, circuit tokens $C$ query netlist tokens $N$ in cross-attention layers: \[ Q_c = C W^c_Q,\quad K_n = N W^n_K,\quad V_n = N W^n_V \] \[ A = \mathrm{Softmax}\big(Q_c K_n^T/\sqrt{d}\big),\;\; C' = AV_n \] This bidirectional attention captures complex relationships between spatial circuit attributes and topological connectivity. The final representation concatenates or merges these token sequences before decoding. ## 4. Static IR-Drop Prediction and Loss The decoder reconstructs a $512\times512$ pixel per-point IR-drop map through four upsampling (deconvolution) layers and a final $1\times1$ convolution. The training loss is mean square error (MSE) between predicted $\hat V$ and true voltage-drop maps $V^{\mathrm{true}}$: \[ \mathcal L_{\mathrm{MSE}} = \frac{1}{HW}\sum_{i=1}^{H}\sum_{j=1}^{W} (\hat V_{ij} - V^{\mathrm{true}}_{ij})^2 \] No cross-entropy or F1-based losses are used during training; F1 is strictly an evaluation metric. ## 5. Training Procedures and Dataset Specification Training employs the ICCAD 2023 dataset, comprising 100 synthetic and 10 real layouts, with 10 hidden real test circuits (sizes ranging from $15{,}436$ to $181{,}206$ netlist nodes). Data augmentation is applied by oversampling synthetic (×10) and real (×20) layouts, and importing 2,000 additional layouts, for a total of 3,310 training samples. Key training parameters are: - Input tile size: $512 \times 512$ - Batch size: $16$ - Optimizer: Adam, initial learning rate $1\mathrm{e}{-3}$ - 200 total epochs - Channel-wise normalization to zero mean, unit variance Each circuit is processed in two modalities: (1) spatial circuit property maps, and (2) 3D netlist point cloud. The pseudocode in the original source outlines preprocessing, forward pass, fusion, decoding, and backpropagation [2511.12581]. ## 6. Empirical Performance and Comparative Evaluation LMM-IR is evaluated on concealed real-world circuits from ICCAD 2023, using three primary metrics: - F1 score for top 10% IR-drop (“hot nodes”) - Mean Absolute Error (MAE) in volts: \[ \mathrm{MAE} = \frac{1}{N_\mathrm{pts}} \sum_i |\hat V_i - V^{\mathrm{true}}_i| \] - Turn-Around Time (TAT) in inference seconds per GPU Performance on the hidden test circuits (average across 10 cases): | Model | F1 | MAE (×10⁻⁴ V) | TAT (s) | |-------------|-------|---------------|---------| | 1st Place | 0.46 | 1.35 | 14.8 | | 2nd Place | 0.45 | 1.50 | 3.04 | | IREDGe | 0.13 | 6.28 | 2.02 | | IRPnet | 0.03 | 3.98 | 2.54 | | **LMM-IR** |**0.58**|**1.35** | 3.05 | LMM-IR achieves the highest F1 and the lowest MAE on average, with competitive inference speed (TAT). It wins 6/10 test cases by F1 [2511.12581]. ## 7. Significance and Implications LMM-IR demonstrates that direct multimodal integration of explicit netlist topology and spatial property maps enables scalable and accurate IR-drop prediction in modern chip design contexts. Its transformer-based netlist point-cloud module efficiently manages extremely large netlists while preserving crucial circuit-theoretic relationships, in contrast to prior image- or grid-based solutions. The result is not only improved MAE and hot-node F1 but also a substantial reduction in required analysis time, suggesting applicability for iterative design closure workflows or real-time design feedback. A plausible implication is that similar multimodal point-cloud+image transformer methods could be adapted for other EDA analysis tasks involving large, non-Euclidean circuit structures. For full architecture specification, formulas, and reproducibility guides, see [2511.12581]. Source: https://www.emergentmind.com/topics/lmm-ir