Multi-Domain IVRs for Chiplet-Based Processors: Towards Energy-Efficient Power Delivery
DOI:
https://doi.org/10.63282/3050-9416.IJAIBDCMS-V6I1P106Keywords:
Relevant technical terms, Multi-Domain IVRs, Chiplet-Based Processors, Power Delivery, Energy Efficiency, On-Chip Voltage RegulatorsAbstract
The use of advanced processors in the contemporary world results in the implementation of the chiplet architectural design for better scalability, higher performance, and relatively affordable prices. However, power delivery is still problematic in such architectures because interconnects consume more power, the power distribution is nonuniform and this architecture has high power density. Coping with these challenges, the present paper aims to describe the design and implementation of the multi-domain integrated voltage regulators (IVRs) concerning chiplet-based processors with an emphasis on the aspects of energy-saving power delivery. Introducing a new multi-domain IVR which regulates the voltage across chiplets both for getting the better performance of the system and for reducing its power consumption. In order to optimize conversion efficiency and maintain thermal stability, our design adopts DVS, load awareness and adaptive feedback mechanisms. Other than that the conventional mono-voltage regulators, our methods provide discrete control of voltage over many domains leading to lower power wastage in the system. To confirm the effectiveness of our strategy, we perform realistic and experimental results using standard processor benchmarks. The outcomes show that the adopted IVR architecture can increase the power conversion efficiency by up to 20% and decrease thermal hotspots by 15% as compared to traditional power delivery methods. Furthermore, the proposed adaptive power management approaches results in turning off voltage droops and improvements in load voltage balancing, so that the total computational throughout increases by 10%. From these results, it is clear that the application of multi-domain IVRs open to opportunities in managing power in chiplet-based processors for the development of subsequent generation energy-dependent computing systems
References
[1] Li, H., Wang, X., Xu, J., Wang, Z., Maeda, R. K., Wang, Z., ... & Wang, Z. (2016). Energy-efficient power delivery system paradigms for many-core processors. IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems, 36(3), 449- 462.
[2] Shan, G., Zheng, Y., Xing, C., Chen, D., Li, G., & Yang, Y. (2022). Architecture of computing system based on chiplet. Micromachines, 13(2), 205.
[3] Chen, S., Zhang, H., Ling, Z., Zhai, J., & Yu, B. (2024). The Survey of Chiplet-based Integrated Architecture: An EDA perspective. arXiv preprint arXiv:2411.04410.
[4] Dave, A., & Dave, K. (2023). Chiplet-Based Architecture for Next-Generation Vehicular Systems. J Artif Intell Mach Learn & Data Sci, 1(4), 915-919.
[5] Chéour, R., Jmal, M. W., Khriji, S., El Houssaini, D., Trigona, C., Abid, M., & Kanoun, O. (2021). Towards hybrid energy-efficient power management in wireless sensor networks. Sensors, 22(1), 301.
[6] Zappone, A., Sanguinetti, L., Bacci, G., Jorswieck, E., & Debbah, M. (2015). Energy-efficient power control: A look at 5G wireless technologies. IEEE Transactions on Signal Processing, 64(7), 1668-1683.
[7] Latini, A. (2024). Validation of the Framework and IVR. In Immersive Virtual Reality for a Building Occupant-Centric Design: Defining, Validating and Applying an Innovative Framework (pp. 35-48). Cham: Springer Nature Switzerland.
[8] Tian, W., Li, B., Li, Z., Cui, H., Shi, J., Wang, Y., & Zhao, J. (2022). Using chiplet encapsulation technology to achieve processing-in-memory functions. Micromachines, 13(10), 1790.
[9] Empower Semiconductor to Present at Chiplet Summit 2024 on Eliminating External Regulators in Chiplet Architectures, Empower, 2024. online. https://www.globenewswire.com/newsrelease/2024/01/25/2816831/0/en/EmpowerSemiconductor-to-Present-at-Chiplet-Summit-2024-onEliminating-External-Regulators-in-ChipletArchitectures.html
[10] Willis, H. L., Tram, H., Engel, M. V., & Finley, L. (1996). Selecting and applying distribution optimization methods. IEEE Computer Applications in Power, 9(1), 12-17.
[11] Panciatici, P., Campi, M. C., Garatti, S., Low, S. H., Molzahn, D. K., Sun, A. X., & Wehenkel, L. (2014, August). Advanced optimization methods for power systems. In 2014 power systems computation conference (pp. 1-18). IEEE.
[12] Mounce, G., Lyke, J., Horan, S., Powell, W., Doyle, R., & Some, R. (2016, March). Chiplet based approach for heterogeneous processing and packaging architectures. In 2016 IEEE Aerospace Conference (pp. 1-12). IEEE.
[13] Voltage-Regulator ICs Close the Power Gap with NextGen Chips, Electronic Design, 2024. online. https://www.electronicdesign.com/technologies/power/ article/55021283/electronic-design-can-integratedvoltage-regulators-take-on-ai-power-delivery
[14] Vinnakota, B., & Shalf, J. M. (2023). Modular HighPerformance Computing Using Chiplets. Computing in Science & Engineering, 25(6), 39-48.
[15] Salah, A., & Guirguis, M. N. (2024, September). A Proposed Framework for integrating IVR Technology in Architectural Design courses; Application on architectural schools in Egypt. In IOP Conference Series: Earth and Environmental Science (Vol. 1396, No. 1, p. 012016). IOP Publishing.
[16] Amelifard, B., & Pedram, M. (2009). Optimal design of the power-delivery network for multiple voltage-island system-on-chips. IEEE Transactions on ComputerAided Design of Integrated Circuits and Systems, 28(6), 888-900.
[17] Zimmermann, J., Bringmann, O., & Rosenstiely, W. (2012, March). Analysis of multi-domain scenarios for optimized dynamic power management strategies. In 2012 Design, Automation & Test in Europe Conference & Exhibition (DATE) (pp. 862-865). IEEE.
[18] Manchester, K., & Bird, D. (1980). Thermal resistance: A reliability consideration. IEEE Transactions on Components, Hybrids, and Manufacturing Technology, 3(4), 580-587.
[19] Lawson, C. P., & Pointon, J. M. (2008, September). Thermal management of electromechanical actuation on an all-electric aircraft. In 26th ICAS Congress, Anchorage, Alaska (pp. 14-19).
[20] Empower Semiconductor to Present at Chiplet Summit 2024 on Eliminating External Regulators in Chiplet Architectures, Empower Semiconductor, online. https://www.empowersemi.com/empowersemiconductor-to-present-at-chiplet-summit-2024-oneliminating-external-regulators-in-chiplet-architectures/
[21] Kim, J., Chekuri, V. C. K., Rahman, N. M., Dolatsara, M. A., Torun, H. M., Swaminathan, M., ... & Lim, S. K. (2021). Chiplet/interposer co-design for power delivery network optimization in heterogeneous 2.5-D ICs. IEEE Transactions on Components, Packaging and Manufacturing Technology, 11(12), 2148-2157.
[22] Udkar S. Emerging Advanced Packaging Technologies and Their Impact on Modern Computer Architecture. IJETCSIT:6(1):56-63.Availablefrom: https://ijetcsit.org/index.php/ijetcsit/article/view/97.
[23] Udkar S. Harnessing Photonic Computing for NextGeneration CPUs and GPUs in High-Performance Computing. IJETCSIT; 5(4):23-36. Available from: https://ijetcsit.org/index.php/ijetcsit/article/view/94
