Thilo Pionteck
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Thilo Pionteck has a total of 130 co-authors. Here are the top co-authors by number of shared publications:
- Jan Moritz Joseph
RWTH Aachen University Germany
21 shared publications Christopher Blochwitz
Universität zu Lübeck
12 shared publicationsCarsten Albrecht
11 shared publications- Sven Groppe
Professorship of Artificial Intelligence, Technische Universität Bergakademie, Freiberg, 09599, Germany
11 shared publications Stefan Werner
Universität zu Lübeck
10 shared publicationsErik Maehle
University of Lübeck
9 shared publications
Latest Publications
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 168794-168810
Wave-Pipelined Interconnects for Network-on-Chips in Monolithic 3-D System-on-Chips
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 168794-168810
Open AccessWave-Pipelined Interconnects for Network-on-Chips in Monolithic 3-D System-on-Chips
- Open Access
Evaluating Rapid Makespan Predictions for Heterogeneous Systems with Programmable Logic
Heterogeneous computing systems, which combine general-purpose processors with specialized accelerators, are increasingly important for optimizing the performance of modern applications. A central challenge is to decide which parts of an application should be executed on which accelerator or, more generally, how to map the tasks of an application to available devices. Predicting the impact of a change in a task mapping on the overall makespan is non-trivial. While there are very capable simulators, these generally require a full implementation of the tasks in question, which is particularly time-intensive for programmable logic. A promising alternative is to use a purely analytical function, which allows for very fast predictions, but abstracts significantly from reality. Bridging the gap between theory and practice poses a significant challenge to algorithm developers. This paper aims to aid in the development of rapid makespan prediction algorithms by providing a highly flexible evaluation framework for heterogeneous systems consisting of CPUs, GPUs and FPGAs, which is capable of collecting real-world makespan results based on abstract task graph descriptions. We analyze to what extent actual makespans can be predicted by existing analytical approaches. Furthermore, we present common challenges that arise from high-level characteristics such as data transfer overhead and device congestion in heterogeneous systems.
Open AccessEvaluating Rapid Makespan Predictions for Heterogeneous Systems with Programmable Logic
Heterogeneous computing systems, which combine general-purpose processors with specialized accelerators, are increasingly important for optimizing the performance of modern applications. A central challenge is to decide which parts of an application should be executed on which accelerator or, more generally, how to map the tasks of an application to available devices. Predicting the impact of a change in a task mapping on the overall makespan is non-trivial. While there are very capable simulators, these generally require a full implementation of the tasks in question, which is particularly time-intensive for programmable logic. A promising alternative is to use a purely analytical function, which allows for very fast predictions, but abstracts significantly from reality. Bridging the gap between theory and practice poses a significant challenge to algorithm developers. This paper aims to aid in the development of rapid makespan prediction algorithms by providing a highly flexible evaluation framework for heterogeneous systems consisting of CPUs, GPUs and FPGAs, which is capable of collecting real-world makespan results based on abstract task graph descriptions. We analyze to what extent actual makespans can be predicted by existing analytical approaches. Furthermore, we present common challenges that arise from high-level characteristics such as data transfer overhead and device congestion in heterogeneous systems.
- Open Access
Static task mapping for heterogeneous systems based on series-parallel decompositions
Modern heterogeneous systems consist of many different processing units, such as CPUs, GPUs, FPGAs and AI units. A central problem in the design of applications in this environment is to find a beneficial mapping of tasks to processing units. While there are various approaches to task mapping, few can deal with high heterogeneity or applications with a high number of tasks and many dependencies. In addition, streaming aspects of FPGAs are generally not considered. We present a new general task mapping principle based on graph decompositions and model-based evaluation that can find beneficial mappings regardless of the complexity of the scenario. We apply this principle to create a high-quality and reasonably efficient task mapping algorithm using series-parallel decompositions. For this, we present a new algorithm to compute a forest of series-parallel decomposition trees for general DAGs. We compare our decomposition-based mapping algorithm with three mixed-integer linear programs, one genetic algorithm and two variations of the Heterogeneous Earliest Finish Time (HEFT) algorithm. We show that our approach can generate mappings that lead to substantially higher makespan improvements than the HEFT variations in complex environments while being orders of magnitude faster than a mapper based on genetic algorithms or integer linear programs.
Open AccessStatic task mapping for heterogeneous systems based on series-parallel decompositions
Modern heterogeneous systems consist of many different processing units, such as CPUs, GPUs, FPGAs and AI units. A central problem in the design of applications in this environment is to find a beneficial mapping of tasks to processing units. While there are various approaches to task mapping, few can deal with high heterogeneity or applications with a high number of tasks and many dependencies. In addition, streaming aspects of FPGAs are generally not considered. We present a new general task mapping principle based on graph decompositions and model-based evaluation that can find beneficial mappings regardless of the complexity of the scenario. We apply this principle to create a high-quality and reasonably efficient task mapping algorithm using series-parallel decompositions. For this, we present a new algorithm to compute a forest of series-parallel decomposition trees for general DAGs. We compare our decomposition-based mapping algorithm with three mixed-integer linear programs, one genetic algorithm and two variations of the Heterogeneous Earliest Finish Time (HEFT) algorithm. We show that our approach can generate mappings that lead to substantially higher makespan improvements than the HEFT variations in complex environments while being orders of magnitude faster than a mapper based on genetic algorithms or integer linear programs.
- Open Access
- 24 January 2025
- Published by Springer Nature
- p. 1-38
ADAMANT: Hardware-Accelerated Query Processing Made Easy
We present ADAMANT, an adaptive data management architecture for evolving heterogeneous hardware/software systems. The ADAMANT system enables plug’n’play integration of co-processors such as GPUs and FPGAs, provides a unified runtime that supports cross-device parallelization of SQL query execution on arbitrary co-processors at runtime, and supports multi-query processing. We discuss the concepts behind and performance of ADAMANT following an example query (TPC-H Q6), present different execution models for cross-device execution, and provide an FPGA hardware architecture that allows the implementation of query-specific hardware accelerators at runtime. In addition, we introduce a new approach for approximate query processing for FPGAs. Our performance analysis concludes that the use of hardware accelerators and device-specific implementations of query operators as part of a common system can provide significant speedups over state-of-the-art query execution engines.
- 24 January 2025
- Published by Springer Nature
- p. 1-38
Open AccessADAMANT: Hardware-Accelerated Query Processing Made Easy
We present ADAMANT, an adaptive data management architecture for evolving heterogeneous hardware/software systems. The ADAMANT system enables plug’n’play integration of co-processors such as GPUs and FPGAs, provides a unified runtime that supports cross-device parallelization of SQL query execution on arbitrary co-processors at runtime, and supports multi-query processing. We discuss the concepts behind and performance of ADAMANT following an example query (TPC-H Q6), present different execution models for cross-device execution, and provide an FPGA hardware architecture that allows the implementation of query-specific hardware accelerators at runtime. In addition, we introduce a new approach for approximate query processing for FPGAs. Our performance analysis concludes that the use of hardware accelerators and device-specific implementations of query operators as part of a common system can provide significant speedups over state-of-the-art query execution engines.
- 1 August 2024
- Published by Springer Nature
- p. 177-191
On-the-Fly CT Image Pre-processing on MPSoC-FPGAs
- 1 August 2024
- Published by Springer Nature
- p. 177-191
On-the-Fly CT Image Pre-processing on MPSoC-FPGAs
- Open Access
- 25 June 2024
- Published by Springer Nature in Information Systems Frontiers
- Vol. 28 (3) , 1-16
Exploiting Shared Sub-Expression and Materialized View Reuse for Multi-Query Optimization
Querying in isolation lacks the potential of reusing intermediate results, which ends up wasting computational resources. Multi-Query Optimization (MQO) addresses this challenge by devising a shared execution strategy across queries, with two generally used strategies: batched or cached. These strategies are shown to improve performance, but hardly any study explores the combination of both. In this work we explore such a hybrid MQO, combining batching (Shared Sub-Expression) and caching (Materialized View Reuse) techniques. Our hybrid-MQO system merges batched query results as well as caches the intermediate results, thereby any new query is given a path within the previous plan as well as reusing the results. Since caching is a key component for improving performance, we measure the impact of common caching techniques such as FIFO, LRU, MRU and LFU. Our results show LRU to be the optimal for our usecase, which we use in our subsequent evaluations. To study the influence of batching, we vary the factor - - which represents the similarity of the results within a query batch. Similarly, we vary the cache sizes to study the influence of caching. Moreover, we also study the role of different database operators in the performance of our hybrid system. The results suggest that, depending on the individual operators, our hybrid method gains a speed-up between 4x to a slowdown of 2x from using MQO techniques in isolation. Furthermore, our results show that workloads with a generously sized cache that contain similar queries benefit from using our hybrid method, with an observed speed-up of 2x over sequential execution in the best case.
- 25 June 2024
- Published by Springer Nature in Information Systems Frontiers
- Vol. 28 (3) , 1-16
Open AccessExploiting Shared Sub-Expression and Materialized View Reuse for Multi-Query Optimization
Querying in isolation lacks the potential of reusing intermediate results, which ends up wasting computational resources. Multi-Query Optimization (MQO) addresses this challenge by devising a shared execution strategy across queries, with two generally used strategies: batched or cached. These strategies are shown to improve performance, but hardly any study explores the combination of both. In this work we explore such a hybrid MQO, combining batching (Shared Sub-Expression) and caching (Materialized View Reuse) techniques. Our hybrid-MQO system merges batched query results as well as caches the intermediate results, thereby any new query is given a path within the previous plan as well as reusing the results. Since caching is a key component for improving performance, we measure the impact of common caching techniques such as FIFO, LRU, MRU and LFU. Our results show LRU to be the optimal for our usecase, which we use in our subsequent evaluations. To study the influence of batching, we vary the factor - - which represents the similarity of the results within a query batch. Similarly, we vary the cache sizes to study the influence of caching. Moreover, we also study the role of different database operators in the performance of our hybrid system. The results suggest that, depending on the individual operators, our hybrid method gains a speed-up between 4x to a slowdown of 2x from using MQO techniques in isolation. Furthermore, our results show that workloads with a generously sized cache that contain similar queries benefit from using our hybrid method, with an observed speed-up of 2x over sequential execution in the best case.
- Open Access
- 18 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 116219-116234
Enabling Plug-and-Play in Cyber-Physical Systems Using MPSoC-FPGAs
Cyber-Physical Systems (CPSs) combine computation, networking, and physical processes. A CPS consists of various independent subsystem components with different interfaces and protocols defined by vendors. Utilizing different interfaces and protocols with the necessity for flexible, dependable, and extensible CPSs poses new challenges for controlling, communicating, and synchronizing the components. In this paper, we propose a Centralized Control Unit (CCU) for CPSs and the associated Communication Infrastructure that enable the support of subsystem components as "plug-and-play" modules. Our CCU consists of a hardware-software architecture that handles physical signals as well as real-time and non-real-time tasks implemented on a single MPSoC-FPGA. To enforce the dependability and flexibility of the communication between components of different vendors, we model the Communication Infrastructure in layers with communication classes. In addition, we propose three vendor-agnostic application protocols, which are also implemented in the CCU. To validate our work, we have implemented and integrated the CCU with the communication infrastructure into an open-interface Computed Tomography (CT) scanner, where sensors/actuators can be added in a "plug-and-play" fashion. We have also evaluated the effort of integrating an additional detector into our scanner. The CCU runs on an AMD-Xilinx Zynq-7000 XC7Z045, and it uses only 10% of the hardware resources.
- 18 October 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 11, 116219-116234
Open AccessEnabling Plug-and-Play in Cyber-Physical Systems Using MPSoC-FPGAs
Cyber-Physical Systems (CPSs) combine computation, networking, and physical processes. A CPS consists of various independent subsystem components with different interfaces and protocols defined by vendors. Utilizing different interfaces and protocols with the necessity for flexible, dependable, and extensible CPSs poses new challenges for controlling, communicating, and synchronizing the components. In this paper, we propose a Centralized Control Unit (CCU) for CPSs and the associated Communication Infrastructure that enable the support of subsystem components as "plug-and-play" modules. Our CCU consists of a hardware-software architecture that handles physical signals as well as real-time and non-real-time tasks implemented on a single MPSoC-FPGA. To enforce the dependability and flexibility of the communication between components of different vendors, we model the Communication Infrastructure in layers with communication classes. In addition, we propose three vendor-agnostic application protocols, which are also implemented in the CCU. To validate our work, we have implemented and integrated the CCU with the communication infrastructure into an open-interface Computed Tomography (CT) scanner, where sensors/actuators can be added in a "plug-and-play" fashion. We have also evaluated the effort of integrating an additional detector into our scanner. The CCU runs on an AMD-Xilinx Zynq-7000 XC7Z045, and it uses only 10% of the hardware resources.
- 16 September 2023
- Published by Springer Nature
- p. 361-365
A Control Data Acquisition System Architecture for MPSoC-FPGAs in Computed Tomography
- 16 September 2023
- Published by Springer Nature
- p. 361-365
A Control Data Acquisition System Architecture for MPSoC-FPGAs in Computed Tomography
- 28 August 2023
- Published by Springer Nature
- p. 74-87
What Happens When Two Multi-Query Optimization Paradigms Combine?
- 28 August 2023
- Published by Springer Nature
- p. 74-87
What Happens When Two Multi-Query Optimization Paradigms Combine?
- Open Access
- 26 May 2023
- Published by Springer Nature in Knowledge and Information Systems
- Vol. 65 (10) , 4359-4377
Hybrid CPU/GPU/APU accelerated query, insert, update and erase operations in hash tables with string keys
Modern computer systems can use different types of hardware acceleration to achieve massive performance improvements. Some accelerators like FPGA and dedicated GPU (dGPU) need optimized data structures for the best performance and often use dedicated memory. In contrast, APUs, which are a combination of a CPU and an integrated GPU (iGPU), support shared memory and allow the iGPU to work together with the CPU on pointer-based data structures. First, we develop an approach for dGPU to accelerate queries in libcuckoo and robin-map and when looking at accelerating insert, updates and erase operations in the original libcuckoo using OneAPI on an APU. We evaluate the dGPU against the CPU variants and our dGPU approach adapted for the CPU and also in a hybrid context by using longer keys on the CPU and shorter keys on the dGPU. In comparison with the original libcuckoo algorithm, our dGPU approach achieves a speed-up of 2.1, and in comparison with the robin-map a speed-up of 1.5. For hybrid workloads, our approach is efficient if long keys are processed on the CPU and short keys are processed on the dGPU. By processing a mixture of 20% long keys on the CPU and 80% short keys on dGPU, our hybrid approach has a 40% higher throughput than the CPU only approach. In addition, we develop a hybrid APU approach for insert, update and erase operations in the original libcuckoo structure focusing on shared memory with iGPU accelerated look-ups of the positions for insert, update and erase operations.
- 26 May 2023
- Published by Springer Nature in Knowledge and Information Systems
- Vol. 65 (10) , 4359-4377
Open AccessHybrid CPU/GPU/APU accelerated query, insert, update and erase operations in hash tables with string keys
Modern computer systems can use different types of hardware acceleration to achieve massive performance improvements. Some accelerators like FPGA and dedicated GPU (dGPU) need optimized data structures for the best performance and often use dedicated memory. In contrast, APUs, which are a combination of a CPU and an integrated GPU (iGPU), support shared memory and allow the iGPU to work together with the CPU on pointer-based data structures. First, we develop an approach for dGPU to accelerate queries in libcuckoo and robin-map and when looking at accelerating insert, updates and erase operations in the original libcuckoo using OneAPI on an APU. We evaluate the dGPU against the CPU variants and our dGPU approach adapted for the CPU and also in a hybrid context by using longer keys on the CPU and shorter keys on the dGPU. In comparison with the original libcuckoo algorithm, our dGPU approach achieves a speed-up of 2.1, and in comparison with the robin-map a speed-up of 1.5. For hybrid workloads, our approach is efficient if long keys are processed on the CPU and short keys are processed on the dGPU. By processing a mixture of 20% long keys on the CPU and 80% short keys on dGPU, our hybrid approach has a 40% higher throughput than the CPU only approach. In addition, we develop a hybrid APU approach for insert, update and erase operations in the original libcuckoo structure focusing on shared memory with iGPU accelerated look-ups of the positions for insert, update and erase operations.