Cong Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Cong Li has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
Chang Zhang
Qufu Normal University,College of Engineering,Rizhao,Shandong,China
1 shared publicationDiancong Zhang
Qufu Normal University,College of Engineering,Rizhao,Shandong,China
1 shared publication- Dianguo Cao
School of Engineering, Qufu Normal University, Rizhao, China
1 shared publication - Feng He
Jiangnan University
1 shared publication - Haide Liu
School of computer science and engineering, North Minzu University, Yinchuan, China
1 shared publication Jin Niu
Institute for Advanced Materials & Technology, University of Science & Technology Beijing, Beijing, 100083, China
1 shared publication
Latest Publications
- Open Access
- 16 September 2024
- Published by Association for Computing Machinery (ACM)
- p. 53-67
Foreseer: Knowledge-Driven Acceleration of Memory-Bound Matrix Multiplications for Large Language Model Inference
The majority of the latency in large language model inference lies in a few memory-bound matrix multiplication kernels. Traditionally, those kernels are optimized by extensive offline experiments of different tiling, striding, and slicing parameter choices on the target hardware. Scaling to different models, inference settings, and GPU hardware then becomes difficult. In this paper, we propose Foreseer, a new policy to generate the high-performing parameters for different kernels on the fly without offline experiments on the hardware. Foreseer leverages the specific characteristics of memory-bound matrix multiplication kernels, and synthesizes various prior knowledge such as keeping the adequate memory access pressure, avoiding the potential tail effect from load imbalance in dispatching, etc., into a simple penalty function. For a given parameter choice, Foreseer provides an analytical estimation of the impacting factors such as the memory access pressure, the potential tail effect, etc., to score the choice. A high-performing parameter choice is then discriminated from the penalty function. Micro-benchmarks of matrix multiplication kernels from popular language models on two different high-end Intel Data Center GPUs demonstrate that Foreseer outperforms different baselines including the vendor libraries, achieving >90% of the best-known performance obtained from the exhaustive offline parameter search experiments. End-to-end inference experiments on those language models with different settings also show that Foreseer accelerates the baseline by 20% on average, being competitive with and quite often exceeding the performance of the comprehensively-optimized vendor inference engine.
- 16 September 2024
- Published by Association for Computing Machinery (ACM)
- p. 53-67
Open AccessForeseer: Knowledge-Driven Acceleration of Memory-Bound Matrix Multiplications for Large Language Model Inference
The majority of the latency in large language model inference lies in a few memory-bound matrix multiplication kernels. Traditionally, those kernels are optimized by extensive offline experiments of different tiling, striding, and slicing parameter choices on the target hardware. Scaling to different models, inference settings, and GPU hardware then becomes difficult. In this paper, we propose Foreseer, a new policy to generate the high-performing parameters for different kernels on the fly without offline experiments on the hardware. Foreseer leverages the specific characteristics of memory-bound matrix multiplication kernels, and synthesizes various prior knowledge such as keeping the adequate memory access pressure, avoiding the potential tail effect from load imbalance in dispatching, etc., into a simple penalty function. For a given parameter choice, Foreseer provides an analytical estimation of the impacting factors such as the memory access pressure, the potential tail effect, etc., to score the choice. A high-performing parameter choice is then discriminated from the penalty function. Micro-benchmarks of matrix multiplication kernels from popular language models on two different high-end Intel Data Center GPUs demonstrate that Foreseer outperforms different baselines including the vendor libraries, achieving >90% of the best-known performance obtained from the exhaustive offline parameter search experiments. End-to-end inference experiments on those language models with different settings also show that Foreseer accelerates the baseline by 20% on average, being competitive with and quite often exceeding the performance of the comprehensively-optimized vendor inference engine.
- 10 January 2022
- Published by Elsevier in Composites Science and Technology
- Vol. 220, 109266
Large mechanical-to-electric output originated from optimized configuration in P(VDF-TrFE)-based nanocomposite fibrous membrane as wearable nanogenerator
- 10 January 2022
- Published by Elsevier in Composites Science and Technology
- Vol. 220, 109266
Large mechanical-to-electric output originated from optimized configuration in P(VDF-TrFE)-based nanocomposite fibrous membrane as wearable nanogenerator
- 1 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 96-101
An Analysis of DevOps Architecture for EMIS based on jBPM
In recent years, DevOps(Combination of Development and Operations) is an emerging software engineering practice. More and more organizations have begun to adopt DevOps method. This paper develops an educational management information system(EMIS) based on jBPM initiated by the DevOps concept, and proposes a quantitative analysis method based on the analytic hierarchy process(AHP) for its system evaluation. Preliminary evaluation results show that this method can help us understand the maturity of the system's DevOps architecture.
- 1 August 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 96-101
An Analysis of DevOps Architecture for EMIS based on jBPM
In recent years, DevOps(Combination of Development and Operations) is an emerging software engineering practice. More and more organizations have begun to adopt DevOps method. This paper develops an educational management information system(EMIS) based on jBPM initiated by the DevOps concept, and proposes a quantitative analysis method based on the analytic hierarchy process(AHP) for its system evaluation. Preliminary evaluation results show that this method can help us understand the maturity of the system's DevOps architecture.
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 5267-5271
Research and Design of a Portable Intelligent ECG Monitoring System
With the increase of cardiovascular diseases, there is an increasing demand for instruments with cardiac health monitoring functions. Therefore, we designed a portable intelligent ECG monitoring system that can collect ECG signals from the wrists of both hands through a pair of repeatable electrode holders. The system mainly includes hardware circuit design and software preprocessing algorithm design. Among them, the hardware circuit design mainly includes the filtering and amplifying circuit based on the design of AD8232 chip, the system circuit based on STM32 microprocessor, Bluetooth transmission circuit and other peripheral circuits. The software and filtering algorithm design mainly includes the lower computer acquisition, analog-to-digital conversion and filtering program based on C language development, and the signal pre-processing program and monitoring interface based on LabVIEW software development. Among them, the preprocessing process of ECG signals mainly includes low-pass filtering, power frequency notching, empirical mode decomposition, R-wave detection and heart rate calculation. The experiment has proved that the system has good performance and can complete daily cardiac monitoring tasks.
- 1 November 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 5267-5271
Research and Design of a Portable Intelligent ECG Monitoring System
With the increase of cardiovascular diseases, there is an increasing demand for instruments with cardiac health monitoring functions. Therefore, we designed a portable intelligent ECG monitoring system that can collect ECG signals from the wrists of both hands through a pair of repeatable electrode holders. The system mainly includes hardware circuit design and software preprocessing algorithm design. Among them, the hardware circuit design mainly includes the filtering and amplifying circuit based on the design of AD8232 chip, the system circuit based on STM32 microprocessor, Bluetooth transmission circuit and other peripheral circuits. The software and filtering algorithm design mainly includes the lower computer acquisition, analog-to-digital conversion and filtering program based on C language development, and the signal pre-processing program and monitoring interface based on LabVIEW software development. Among them, the preprocessing process of ECG signals mainly includes low-pass filtering, power frequency notching, empirical mode decomposition, R-wave detection and heart rate calculation. The experiment has proved that the system has good performance and can complete daily cardiac monitoring tasks.
- 2 May 2018
- Published by Springer Nature in SICS Software-Intensive Cyber-Physical Systems
- Vol. 34 (1) , 35-43
Combining global regression and local approximation in server power modeling
- 2 May 2018
- Published by Springer Nature in SICS Software-Intensive Cyber-Physical Systems
- Vol. 34 (1) , 35-43
Combining global regression and local approximation in server power modeling