Papers by ilhami torunoglu
Three new fast constraint graph generation algorithms, PPSS-1D, PPSS-1Dk and PPSS-2D, are present... more Three new fast constraint graph generation algorithms, PPSS-1D, PPSS-1Dk and PPSS-2D, are presented for VLSI layout compaction. The algorithms are based on parallel plane sweep shadowing (PPSS). The PPSS-1D algorithm improves the time spent on searching processes from O(Nˆ1.5) to O(G*N) with extra O(G) memory where G is independent of N. PPSS-1Dk, the successor to PPSS-1D, eliminates the possibility of
Proceedings of SPIE, Jul 9, 2015
Sub-Resolution Assist Features (SRAFs) have emerged as a key technology to enable semiconductor m... more Sub-Resolution Assist Features (SRAFs) have emerged as a key technology to enable semiconductor manufacturing for advanced technology nodes. SRAF placement is required to adhere to manufacturability constraints (MRC). MRC specifications are distance and size constraints specified by the user to ensure SRAFs are not detrimental to the final target shapes being printed. Conceptually, SRAF placement can be divided into two steps - SRAF candidate generation and SRAF candidate cleanup or conflict resolution. SRAFs generated as candidates may not adhere to MRC constraints. It is during the cleanup/conflict resolution process that the MRC constraints are enforced. In this paper we focus on the latter phase - cleanup. The goal of the cleanup phase is to retain as much of the initial candidates as possible, and, if necessary, transform them to adhere to MRC conditions. An SRAF is said to be in conflict with another shape if it violates the distance MRC constraint. One can model these conflicts using a conflict graph G=(V,E), whose vertices V correspond to geometric shapes involved in a conflict and an edge is present in E, between two vertices if the corresponding shapes are involved in a conflict. A weight is associated with each vertex that could, for example, correspond to area of the corresponding shape. The goal of conflict resolution then, is to find a transformation of the vertices so that the resulting graph is conflict free while maximizing the weight of vertices retained. This can be viewed as a generalization of the computationally hard problem of finding the largest independent set of candidates, albeit allowing for transformation. The transformations we allow include deletion, splitting, resizing, merge, and bounded translation. In this paper, we describe an approach which classifies the conflicts and apply appropriate transformations to achieve effective SRAF placement. Further, we demonstrate that such a strategy reduces the number of rules to be specified by the user, reducing the user effort needed to achieve desirable imaging results.
In the semiconductor fabrication process, yield is negatively impacted by defects that appear sys... more In the semiconductor fabrication process, yield is negatively impacted by defects that appear systematically within specific patterns of the physical layout design. Those defective patterns are popularly known as hotspots, and they can arise due to various causes. There are several known approaches of hotspot detection. One approach for hotspot detection is Machine Learning (ML), where known hotspot and non-hotspot patterns are used for training the model to be used afterwards in prediction of new hotspots. The objective in ML approaches is to maximize the hit rate (i.e. finding all potential hotspots) and to minimize the false alarm rate (i.e. reduce the overhead of false positives). The model’s ability to correctly classify between hotspots and non-hotspots depends on the coverage of the training data set. The real-world challenge in training a ML system to classify hotspots/non-hotspots is the imbalanced nature of the problem, where the known hotspot patterns are always in the minority class. Another challenge specific to the problem of hotspot classification is the difficulty to correctly classify non-hotspots that are similar to hotspots. These “hard-to-classify” patterns are ones with high mask error enhancement factor (MEEF), as small variations in the pattern can make it change between hotspot and non-hotspot. These two challenges cause conventional methods of handling imbalanced training datasets to be inadequate to the problem of hotspot detection. This paper will present a flow for quantified training dataset selection approach and put extra focus on the patterns that are hard to classify due to close similarity with known hotspots. Improved model accuracy is illustrated when adopting the quantified sampling approach compared to conventional sampling approaches.
New design paradigms based on the concept of system- on-chip are gradually replacing printed circ... more New design paradigms based on the concept of system- on-chip are gradually replacing printed circuit board centric ap- proaches. This trend is mainly due to two factors: far higher running speeds and greater miniaturization. The new paradigms will accelerate design cycles, which in turn will force designers to reuse existing and ac- quire new circuits ready to be integrated. Such acceleration will be pos- sible only if highly specialized core authors, integrators, and foundries will be able to efficiently and safely exchange and handle their intel- lectual property. The field known as intellectual property protection is aimed at limiting all violat.ions to intellectual property rights through appropriate design methodologies, tools, and infringement detection techniques. The paper surveys all published aspects of the intellectual property protection problem. in the context of concerted VSIA efforts to define new standards and protocols.
Custom Integrated Circuits Conference, 2000
DTCO and Computational Patterning, Jun 13, 2022
Design-Process-Technology Co-optimization XV, 2021
In the semiconductor fabrication process, yield is negatively impacted by defects that appear sys... more In the semiconductor fabrication process, yield is negatively impacted by defects that appear systematically within specific patterns of the physical layout design. Those defective patterns are popularly known as hotspots, and they can arise due to various causes. There are several known approaches of hotspot detection. One approach for hotspot detection is Machine Learning (ML), where known hotspot and non-hotspot patterns are used for training the model to be used afterwards in prediction of new hotspots. The objective in ML approaches is to maximize the hit rate (i.e. finding all potential hotspots) and to minimize the false alarm rate (i.e. reduce the overhead of false positives). The model’s ability to correctly classify between hotspots and non-hotspots depends on the coverage of the training data set. The real-world challenge in training a ML system to classify hotspots/non-hotspots is the imbalanced nature of the problem, where the known hotspot patterns are always in the minority class. Another challenge specific to the problem of hotspot classification is the difficulty to correctly classify non-hotspots that are similar to hotspots. These “hard-to-classify” patterns are ones with high mask error enhancement factor (MEEF), as small variations in the pattern can make it change between hotspot and non-hotspot. These two challenges cause conventional methods of handling imbalanced training datasets to be inadequate to the problem of hotspot detection. This paper will present a flow for quantified training dataset selection approach and put extra focus on the patterns that are hard to classify due to close similarity with known hotspots. Improved model accuracy is illustrated when adopting the quantified sampling approach compared to conventional sampling approaches.
L'invention porte sur un systeme d'entree sensoriel (103) permettant de detecter le mouve... more L'invention porte sur un systeme d'entree sensoriel (103) permettant de detecter le mouvement des doigts d'un utilisateur sur une surface de travail inerte (204), et dans lequel au moins deux modes d'entree (par exemple un clavier (203A) et une souris (203B)) sont disposes dans un espace physique de chevauchement ou coextensif (203C). En fonction du mode actif, l'invention permet d'interpreter les mouvements des doigts selon un des modes d'entree. La commutation de mode automatique et/ou manuel est egalement installee.
L'invention se rapporte a une technique permettant de supprimer les plages floues dans une im... more L'invention se rapporte a une technique permettant de supprimer les plages floues dans une image capturee par un dispositif d'imagerie. Ledit dispositif d'imagerie peut comporter une lentille et un support d'imagerie compose d'une pluralite de fragments d'imagerie. Conformement a un mode de realisation, une distance est determinee entre les fragments d'imagerie individuels du support d'imagerie et une region de la surface de l'objet cible qui correspond au fragment d'imagerie individuel respectif. L'image de l'objet cible est capturee sur le support d'imagerie. Les plages floues de l'image capturee sont supprimees grâce a l'identification des distances associees aux fragments d'imagerie individuels.
Photomask Japan 2015: Photomask and Next-Generation Lithography Mask Technology XXII, 2015
Sub-Resolution Assist Features (SRAFs) have emerged as a key technology to enable semiconductor m... more Sub-Resolution Assist Features (SRAFs) have emerged as a key technology to enable semiconductor manufacturing for advanced technology nodes. SRAF placement is required to adhere to manufacturability constraints (MRC). MRC specifications are distance and size constraints specified by the user to ensure SRAFs are not detrimental to the final target shapes being printed. Conceptually, SRAF placement can be divided into two steps - SRAF candidate generation and SRAF candidate cleanup or conflict resolution. SRAFs generated as candidates may not adhere to MRC constraints. It is during the cleanup/conflict resolution process that the MRC constraints are enforced. In this paper we focus on the latter phase - cleanup. The goal of the cleanup phase is to retain as much of the initial candidates as possible, and, if necessary, transform them to adhere to MRC conditions. An SRAF is said to be in conflict with another shape if it violates the distance MRC constraint. One can model these conflicts using a conflict graph G=(V,E), whose vertices V correspond to geometric shapes involved in a conflict and an edge is present in E, between two vertices if the corresponding shapes are involved in a conflict. A weight is associated with each vertex that could, for example, correspond to area of the corresponding shape. The goal of conflict resolution then, is to find a transformation of the vertices so that the resulting graph is conflict free while maximizing the weight of vertices retained. This can be viewed as a generalization of the computationally hard problem of finding the largest independent set of candidates, albeit allowing for transformation. The transformations we allow include deletion, splitting, resizing, merge, and bounded translation. In this paper, we describe an approach which classifies the conflicts and apply appropriate transformations to achieve effective SRAF placement. Further, we demonstrate that such a strategy reduces the number of rules to be specified by the user, reducing the user effort needed to achieve desirable imaging results.
Uploads
Papers by ilhami torunoglu