Jinoh Cho

Jinoh Cho

Auto-generated profile, based on publicly available information

Jinoh Cho

Pohang University of Science and Technology (POSTECH), Pohang, Gyeongsangbuk-do, Republic of Korea

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 September 2024
Verified
Conference Paper
Open access
DKDonggyu Kim
Donggyu Kim
MKMinjae Kim
Minjae Kim
JHJunseok Hur
Junseok Hur
JLJakang Lee
Jakang Lee
JCJinoh Cho
Jinoh Cho
SKSeokhyeong Kang
Seokhyeong Kang
In the face of challenges posed by semiconductor scaling, 3D integration technology has emerged as a crucial solution to overcome the constraints of traditional 2D Integrated Circuits (ICs). Recent advancements in 3D vertical interconnection technology have enabled 3D interconnections with reduced area overhead, potentially enhancing 3D IC performance in terms of timing, power, yield, and cost. However, despite the increased inter-die connection capacity enabled by fine-pitch 3D vias, research on 3D IC partitioning methods for timing improvement remains insufficient. This paper proposes a partitioning and placement method that optimizes the timing of 3D ICs by leveraging the increased inter-die connection capacity provided by fine pitch 3D vias. By using the Leiden algorithm and unsupervised GNN, we achieve optimal partitioning without relying on the results of 2D placement or additional training data. Our method employs balanced COP-KMeans to incorporate critical path optimization into partitioning, while ensuring balanced cell allocation across dies, thereby maximizing the area benefits of 3D ICs. Additionally, we optimize the timing of 3D ICs through 3D cell placement with critical net weight adjustments and a novel 3D via placement approach that differs from traditional 3D Half-Perimeter Wire Length (HPWL) optimization perspectives.

20 January 2025
Verified
Conference Paper
Open access
JCJinoh Cho
Jinoh Cho
JIJaekyung Im
Jaekyung Im
JLJaeseung Lee
Jaeseung Lee
KMKyungjun Min
Kyungjun Min
SPSeonghyeon Park
Seonghyeon Park
SKSeokhyeong Kang
Seokhyeong Kang
As technology nodes advance and feature sizes shrink, the increasing complexity of design rules and routing congestion has resulted in greater design challenges and rising costs. Machine learning (ML) models offer significant potential to enhance design quality by enabling early prediction and optimization during the design flow. However, only a few works have validated the effectiveness of ML model when integrated to the traditional design flow. This paper will cover the effectiveness of ML-enhanced design workflow with some practical applications. Additionally, we will address which problems should be solved to achieve successful ML integration.

27 October 2024
Verified
Conference Paper
Open access
JCJinoh Cho
Jinoh Cho
SPSeonghyeon Park
Seonghyeon Park
JLJakang Lee
Jakang Lee
SLSung-Yun Lee
Sung-Yun Lee
JAJinmo Ahn
Jinmo Ahn
SKSeokhyeong Kang
Seokhyeong Kang

26 July 2026
Verified
Conference Paper
Open access
JCJinoh Cho
Jinoh Cho
JAJinmo Ahn
Jinmo Ahn
JLJakang Lee
Jakang Lee
JLJaeseung Lee
Jaeseung Lee
SPSeonghyeon Park
Seonghyeon Park
SKSeokhyeong Kang
Seokhyeong Kang

22 September 2025 in IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems
Verified
Journal Article
Closed access
JLJaeseung Lee
Jaeseung Lee
SLSung-Yun Lee
Sung-Yun Lee
JCJinoh Cho
Jinoh Cho
DHDaijoon Hyun
Daijoon Hyun
SKSeokhyeong Kang
Seokhyeong Kang
In advanced technology nodes, the back-end-of-line (BEOL) process is increasingly critical for determining chip timing and power consumption. This article presents CTRL-B, a transferable reinforcement learning (RL) framework that automates and accelerates this process. CTRL-B integrates two key innovations: 1) a graph attention network (GAT)-based encoder, pretrained on out-of-domain datasets to learn generalizable circuit representations and 2) a cross-domain transfer layer that reuses knowledge from previously optimized source domains. This architecture provides the RL agent with an effective initial policy, which significantly reduces redundant exploration and accelerates convergence to high-quality solutions. We demonstrate the effectiveness of CTRL-B against conventional methods, achieving a 0.4% power, 57.6% reduction in worst negative slack (WNS), and a 41.7% reduction in total negative slack (TNS), alongside 1.58% area reductions. The practical viability of the optimized BEOL configurations is confirmed through stable static IR drop and successful design rule (DR) violation checks. A series of ablation and sensitivity analyses further validates the robustness of our architecture across various designs. These results establish CTRL-B as a robust and scalable solution for automated BEOL optimization, highlighting its potential to shorten modern chip development cycles.

5 August 2026
Verified
Conference Paper
Open access
JCJinoh Cho
Jinoh Cho
MKMinhyuk Kweon
Minhyuk Kweon
JAJinmo Ahn
Jinmo Ahn
JPJeyeong Park
Jeyeong Park
SKSeokhyeong Kang
Seokhyeong Kang
Satisfying minimum implant area (MIA) rules is critical for improving yield in advanced technology nodes. Existing methods typically address MIA violations through VT reassignment for both filler and functional cells. Although these methods effectively reduce MIA violations, simultaneously optimizing power and timing remains challenging. In this work, we propose the first differentiable optimization framework for timing- and power-aware repair of MIA violations. At the core of our framework, we cast discrete VT reassignment as a continuous optimization problem by assigning VT probabilities to individual cells. Built upon this foundation, we propose novel differentiable formulations for MIA rules, timing, and power, unifying them into a single objective for simultaneous optimization. Moreover, to guarantee MIA-free solutions, we propose a lightweight MILP-based refinement stage. With these core components, our framework significantly outperforms the baselines, achieving an average reduction of 24.8% in leakage power while allowing minimal timing degradation.