Semiconductor factories process many kinds of wafers organized into lots of various sizes and configurations. Different kinds of machines, each with its own operations, process wafers for hours to days, and the wafers undergo many incremental fabrication steps (with re-entry) and flows spanning days to weeks across connected machines. Fast and accurate predictions of throughput (TH) and turnaround time (TA) dynamics for individual machines and their combinations are essential for factory operations. To use machine learning methods for this purpose, this paper presents a Temporal Convolutional Network with Multi-head Attention (TCNA) architecture. It uses simulated datasets from a benchmark Parallel DEVS (PDEVS) semiconductor fabrication factory model. The factory processes configurable wafer lots with uniform and sinusoidal wafer arrival profiles and may undergo preventive maintenance. The wafer lot size, arrival profile, and factory maintenance policy are used as static covariates. TCNA is applied to a selected set of simulated discrete-event throughput and turnaround time trajectories. The TH and TA time series predicted by TCNA, evaluated with the common MAPE, MSE, and R2 measures, show comparable or higher accuracy relative to other model architectures, with similar parameter counts and epoch times; the TCNA model size and execution time are similar to those of other architectures such as the Temporal Fusion Transformer. Although time series predictions are less accurate than high-resolution simulations, they can be deployed at a much lower computational cost. The TCNA achieves orders-of-magnitude speedup over PDEVS simulation with an acceptable loss of accuracy, enabling frequent product scheduling, capacity planning, and maintenance planning.
Figures & tables
Figure 1: TCN with Multi-head Attention (TCNA) Model Architecture. The dashed future-time branch is an optional input for known-future (dynamic) covariates; it is not used in the experiments of this paper (see Section 7 ).
Scenario
Repair Type
Wafer Generation
A
No Repair
Sinusoidal
B
No Repair
Uniform
C
MTBF
Uniform
D
Processing Steps
Uniform
E
MTBF
Sinusoidal
F
Processing Steps
Sinusoidal
Table 1: Simulation Scenarios for Wafer Manufacturing
Figure 2: MAPE Values for an 8-Stage Cascade Semiconductor MiniFab Factory
Figure 3: TH Prediction Comparison for Medium Lot Size
Figure 4: TA Prediction Comparison for Medium Lot Size
Dataset
Scenario
Metrics
LSTM
TCN
TFT
TCNA
FEDformer
Autoformer
N-BEATS
Throughput
A
R2 Score
0.6754
0.9833
0.9998
0.9999
0.9984
0.9997
0.9990
MSE
1.22E-07
4.00E-08
5.67E-10
4.07E-10
5.22E-09
1.14E-09
3.32E-09
MAPE
32.59%
4.30%
0.54%
0.86%
3.28%
1.52%
1.90%
B
R2 Score
0.6810
0.9977
0.9998
0.9998
0.9983
0.9996
0.9989
MSE
1.20E-06
6.07E-09
5.77E-10
6.43E-10
5.22E-09
1.15E-09
3.35E-09
MAPE
32.56%
2.05%
0.57%
0.95%
3.36%
1.54%
1.98%
Table 2: Model Performance Comparison: Average Values for Scenarios A–F
Figure 5: TH Prediction Comparison for Small Lot Size
Figure 6: TH Prediction Comparison for Large Lot Size
Figure 7: TA Prediction Comparison for Small Lot Size
Figure 8: TA Prediction Comparison for Large Lot Size
Model Scores
Metrics
Lot Size
LSTM
TCN
TFT
TCNA
FEDformer
Autoformer
N-BEATS
t -statistic
Throughput
Small
3.37E+02
2.14E+02
1.84E+02
9.79E+01
1.02E+03
-2.08E+02
8.96E+01
Medium
-1.68E+03
1.59E+02
-2.12E+02
1.01E+01
8.02E+02
-1.51E+02
1.73E+02
Large
-4.07E+02
-1.20E+02
1.43E+02
1.76E+02
8.42E+02
-1.57E+02
1.22E+02
Turnaround Time
Small
1.12E+03
1.75E+03
-4.95E+02
5.04E+01
4.72E+02
-1.42E+02
4.15E+02
Medium
4.79E+02
-1.17E+00
2.96E+02
7.70E+01
4.42E+02
-1.32E+02
-1.79E+02
Table 3: Statistical Analysis for Scenario E
Timestamp
Lot Size
PDEVS Values
Percentage Difference from PDEVS Values
LSTM
TCN
TFT
TCNA
FEDformer
Autoformer
N-BEATS
Initial
Throughput
Small
1.14E-04
378.70%
31.71%
96.02%
94.38%
34.08%
155.16%
3.21%
Medium
1.10E-04
1931.99%
28.55%
87.50%
97.23%
32.03%
100.16%
23.99%
Large
1.12E-04
896.10%
12.20%
95.65%
92.52%
27.96%
107.36%
32.96%
Turnaround Time
Small
8.75E+03
13.28%
0.66%
3.47%
1.85%
28.10%
49.26%
20.54%
Medium
9.09E+03
8.06%
11.29%
0.92%
0.49%
28.05%
47.30%
16.87%
Table 4: Percentage Difference Between PDEVS and Model Predictions at Initial, Early-Horizon, and Final Time Steps for Throughput and Turnaround Time
Figure 9: TH Prediction Around the First Non-Zero Value for Medium Lot Size
Figure 10: TA Prediction Around the First Non-Zero Value for Medium Lot Size
Automated material handling systems in semiconductor fabs are operated by a material control system (MCS) that must schedule a relay route for every transport command online, before execution. This is a data-driven scheduling problem in which route cost is dominated in the upper tail by queueing at heterogeneous, partially observable relay equipment, so route selection requires estimating both delivery time and congestion risk at the decision moment. This paper proposes a transport-network-aware dynamic congestion representation (TN-DCR). Built on a static directed transport graph induced by historically observed relay segments, TN-DCR combines structural route priors, multi-window network-wide congestion context, route-level bottleneck exposure, and an inductive graph-aware route embedding, all constructed under a prediction-time-safety invariant that admits only information observed strictly before the prediction moment. The representation feeds separate queue- and transfer-time regressors and an ordinal multi-label classifier producing calibrated multi-threshold exceedance scores, with an empirical-Bayes stock-key residual correction reducing systematic queue-time underprediction. The predictions serve as costs in a risk-constrained route-scheduling rule that minimizes predicted delivery time subject to a bound on extreme-congestion probability, embedding the learned predictors within a lightweight operations-research decision model. In a controlled closed-loop evaluation, mean delivery time falls by 16.4% and internal resource waiting time by 22.6% while throughput remains essentially unchanged.
Hao Yin, Meiqi Tu, Anbang Liu +2
Department of Data and Systems Engineering, The University of Hong Kong · The Research Center for Digital Production and Smart Logistics, Tsinghua University
The drive toward heterogeneous 2.5D chiplet packages promises unmatched performance and customization, yet the resulting density amplifies thermal challenges. To facilitate rapid design space exploration, we present ThermLeT, a multimodal transformer-based, scale invariant surrogate that integrates power density maps, spatial coordinate embeddings, material properties and heat transfer rate to predict steady state temperature in 13 ms, about four orders of magnitude faster than conventional finite element solvers such as ANSYS. To train the transformer model, we also create a dataset of 6,221 chiplet layouts perturbed from three representative architectures. On a 150 sample test set, ThermLeT achieves an RMSE of 3.16°C and an MAE of 2.24°C, improving accuracy by 35% over ThermEDGe and by 71% over HotSpot, while running approximately 6x and 11x lower inference latencies respectively. Ablation studies confirm robust generalization to unseen floorplans. Code and dataset available at https://github.com/MicroSystemsDesignLab/ThermLeT
Runtime thermal management of high-performance chips depends on fast and accurate full-chip thermal maps. Conventional simulators typically estimate power traces from performance metrics first, which adds overhead. This work proposes TherMapNet, an attention-guided thermal simulator that predicts full-chip thermal maps directly from performance metrics. A Transformer encoder captures temporal evolution by treating the time series of each metric as a token, improving modeling of dynamic workloads. A CNN then extracts fine-grained spatial features. For the CNN, a dual-branch channel-spatial attention convolution module (DACM) and a triplet loss are used to improve spatial learning and reconstruction accuracy. TherMapNet is applied to a multi-core CPU (AMD Ryzen 7 4800U) and a many-core GPU (NVIDIA GeForce RTX 4060). Experiments show that it outperforms prior thermal simulators, with RMSE below 0.26 C and inference under 2.4 ms on an NVIDIA GeForce RTX 3090 GPU. These results indicate that TherMapNet can support high-quality runtime thermal management of modern multi-core chips.
Qin Gu, Chaofang Ma, Mingyu Yang +4
College of Information Science and Engineering, Northeastern University, Shenyang, China · Department of Electronic and Computer Engineering, The Hong Kong University of Science and Technology, Hong Kong