Harnessing Photonic Computing for Next-Generation CPUs and GPUs in High-Performance Computing
DOI:
https://doi.org/10.63282/3050-9246.IJETCSIT-V5I4P104Keywords:
Photonic Computing, High-Performance Computing (HPC), Photonic Processing Unit (PPU), Optical Interconnects, Hybrid Photonic-Electronic ArchitectureAbstract
Advancements in HPC for AI, scientific modeling, and simulation, as well as big data, have also brought out the shortcomings of conventional silicon-based processors. The use of optical signals as inputs rather than electrical signals provides an innovative approach to solving these issues through photonic computing. This paper, therefore takes a look at the massively enhanced architectures like the Photonic Processing Units (PPUs), photonic electronic compounding architecture, and the optical memory that is proving superior in terms of speed, energy efficiency, and bandwidth. Photonic computing facilitates optical communication through wire and integration of optical interconnects and waveguide-based computation for low energy consumption and heat dissipation. When combined with CPU and GPUs, photonic processors allow for faster computations required in such tasks as deep learning and modeling and cryptographic computations. This paper briefly describes recent experimental advances in stalked graphical models such as VCSEL-based spiking neural networks, Diffractive Optical Neural Networks (DONNs) and Orbital Angular Momentum (OAM) optical vector processing that are reported to be more efficient and computationally precise. Nevertheless, photonic computing still has problems like complexity in fabrication, integration with the current silicon models and thermal concern. However, it is easy to note that hybrid computing models and quantum-enhanced photonics offer certain ways of getting past these issues. In more ways, photonic computing is set to transform data centers, clouds and AI-based computing workloads to a new era of ultra-fast, ultra-energy efficient and scalable computation
Downloads
References
[1] Li, Q., Zhang, Y., Wang, J., & Liu, H. (2023). An Overview of Photonic Neuromorphic Computing Techniques Based on Phase-Change Materials. Laser & Optoelectronics Progress, 60(21), 2100007.
[2] Saleh, B. E., & Teich, M. C. (2019). Fundamentals of photonics, 2-volume set. john Wiley & sons.
[3] Tsakyridis, A., Moralis-Pegios, M., Giamougiannis, G., Kirtas, M., Passalis, N., Tefas, A., & Pleros, N. (2024). Photonic neural networks and optics-informed deep learning fundamentals. APL Photonics, 9(1).
[4] Cheng, J., Zhou, H., & Dong, J. (2021). Photonic matrix computing: from fundamentals to applications. nanomaterials, 11(7), 1683.
[5] Huang, C., Shastri, B., & Pruncal, P. (2024). Photonic computing: An introduction. In Phase Change Materials-Based Photonic Computing (pp. 37-65). Elsevier.
[6] Xu, X. Y., & Jin, X. M. (2023). Integrated photonic computing beyond the von Neumann architecture. ACS Photonics, 10(4), 1027-1036.
[7] Lian, C., Vagionas, C., Alexoudi, T., Pleros, N., Youngblood, N., & Ríos, C. (2022). Photonic (computational) memories: tunable nanophotonics for data storage and computing. Nanophotonics, 11(17), 3823-3854.
[8] Salmani, M., Eshaghi, A., Luan, E., & Saha, S. (2021). Photonic computing to accelerate data processing in wireless communications. Optics Express, 29(14), 22299-22314.
[9] Brückerhoff-Plückelmann, F., Borras, H., Klein, B., Varri, A., Becker, M., Dijkstra, J., ... & Pernice, W. (2024). Probabilistic photonic computing with chaotic light. Nature Communications, 15(1), 10445.
[10] Bai, B., Yang, Q., Shu, H., Chang, L., Yang, F., Shen, B., ... & Wang, X. (2023). Microcomb-based integrated photonic processing unit. Nature Communications, 14(1), 66.
[11] De Marinis, L., Catania, A., Castoldi, P., Contestabile, G., Bruschi, P., Piotto, M., & Andriolli, N. (2022). A codesigned integrated photonic electronic neuron. IEEE Journal of Quantum Electronics, 58(5), 1-10.
[12] Miscuglio, M., Adam, G. C., Kuzum, D., & Sorger, V. J. (2019). Roadmap on material-function mapping for photonic-electronic hybrid neural networks. APL Materials, 7(10).
[13] Design: GPU vs. CPU, Cornell University, online. https://cvw.cac.cornell.edu/gpu-architecture/gpu-characteristics/design
[14] Maniotis, P., Gitzenis, S., Tassiulas, L., & Pleros, N. (2016). An optically-enabled chip–chip-multiprocessor architecture using a single-level shared optical cache memory. Optical Switching and Networking, 22, 54-68.
[15] Thraskias, C. A., Lallas, E. N., Neumann, N., Schares, L., Offrein, B. J., Henker, R., ... & Tomkos, I. (2018). Survey of photonic and plasmonic interconnect technologies for intra-datacenter and high-performance computing communications. IEEE Communications Surveys & Tutorials, 20(4), 2758-2783.
[16] Ning, S., Zhu, H., Feng, C., Gu, J., Jiang, Z., Ying, Z., ... & Chen, R. T. (2024). Photonic-electronic integrated circuits for high-performance computing and AI accelerators. Journal of Lightwave Technology.
[17] Nikdast, M., Pasricha, S., Nicolescu, G., Seyedi, A., & Liang, D. (Eds.). (2021). Silicon photonics for high-performance computing and beyond. CRC Press.
[18] Photonic Computing, Intelligent Computing, online. https://spj.science.org/page/icomputing/si/Photonic-Computing
[19] Fu, T., Zang, Y., Huang, Y., Du, Z., Huang, H., Hu, C., ... & Chen, H. (2023). Photonic machine learning with on-chip diffractive optics. Nature Communications, 14(1), 70.
[20] Zhao, G., Shu, X., & Zhou, R. (2024). High-performance real-world optical computing trained by in situ gradient-based model-free optimization. IEEE Transactions on Pattern Analysis and Machine Intelligence.
[21] Hong, L., Guo, H., Qiu, X., Lin, F., Zhang, W., & Chen, L. (2023). Experimental optical computing of complex vector convolution with twisted light. Advanced Photonics Nexus, 2(4), 046008-046008.
