Electronics and Telecommunications Trends (전자통신동향분석)

Electronics and Telecommunications Research Institute (한국전자통신연구원)
권호 표지
DOI QR코드

DOI QR Code

Trends in Artificial Intelligence Semiconductor Memory Technology

인공지능 반도체 메모리 기술 동향

  • K.D. Hwang ;
  • K.I. Oh ;
  • J.J. Lee ;
  • B.T. Koo
  • 황규동 (지능형엣지반도체연구실 ) ;
  • 오광일 (지능형엣지반도체연구실) ;
  • 이재진 (지능형엣지반도체연구실 ) ;
  • 구본태 (지능형반도체연구본부 )
  • Published : 2024.10.01
Download PDF
⟨ Previous Next ⟩

Abstract

Memory can refer to a storage device that collects data, and it has evolved to increase the reading/writing speed and reduce the power consumption. As large amounts of data are processed by artificial intelligence services, the memory data capacity requires expansion. Dynamic random-access memory (DRAM) is the most widely used type of memory. In particular, graphics double date rate and high-bandwidth memory allow to quickly transfer large amounts of data and are used as memory solutions for artificial intelligence semiconductors. We analyze development trends in DRAM from the perspectives of processing speed and power consumption. We summarize the characteristics required for next-generation memory by comparing DRAM and other types of memory implementations. Moreover, we examine the shortcomings of DRAM and infer a next-generation memory for their compensation. We also describe the operating principles of spin-torque transfer magnetic random access memory, which may replace DRAM in next-generation devices, and explain its characteristics and advantages.

Keywords

Acknowledgement

This work was supported by Electronics and Telecommunications Research Institute (ETRI) grant funded by the Korea government[24ZS1230, Memory-computation convergence neuromorphic computing technology].

References

  1. R.H. Dennard, "Field-effect transistor memory," US3387286A, Jun. 4, 1968. 
  2. https://www.jedec.org/about-jedec/jedec-history/2000s/ 
  3. H.Y. Yoon et al., "A 2.5-V, 333-Mb/s/pin, 1-Gbit, double-date-rate synchronous DRAM," IEEE J. Solid-State Circuits, vol. 34, no. 11, Nov. 1999, pp. 1589- 1599. 
  4. K.-D. Hwang et al., "A 16 Gb/s/pin 8 Gb GDDR6 DRAM with bandwidth extension techniques for high-speed applications," in IEEE Int. Solid-State Circuits Conf., (San Francisco, CA, USA), Feb. 2018, pp. 210-212. 
  5. T.-Y. Oh, "A 3.2 Gbps/pin 8 Gbit 1.0 V LPDDR4 SDRAM with integrated ECC engine for sub-1 V DRAM core operation," IEEE J. Solid-State Circuits, vol. 50, no. 1, Jan. 2015, pp. 178-190. 
  6. D.U. Lee et al., "A 128 Gb 8-high 512 GB/s HBM2E DRAM with a pseudo quarter bank structure, power dispersion and an instruction-based At-speed PMBIST," in IEEE Int. Solid-State Circuits Conf. (San Francisco, CA, USA), Feb. 2020, pp. 334-336. 
  7. D.-H. Kim et al., "A 16 Gb 9.5 Gb/S/pin LPDDR5X SDRAM with low-power schemes exploiting dynamic voltage-frequency scaling and offset-calibrated readout sense amplifiers in a fourth generation 10nm DRAM process," in IEEE Int. Solid-State Circuits Conf. (San Francisco, CA, USA), Feb. 2022, pp. 448-449. 
  8. J.-H. Yang et al., "13.1 A 35.4Gb/s/pin 16Gb GDDR7 with a Low-Power Clocking Architecture and PAM3 IO Circuitry," in IEEE Int. Solid-State Circuits Conf. (San Francisco, CA, USA), Feb. 2024, pp. 232-233. 
  9. K. Ishibashi et al., "Low Power and Reliable SRAM Memory Cell and Array Design," Berlin, Germany, Springer, 2011. 
  10. H.-J. Kim et al., "1 GB/s 2Tb NAND flash multi-chip package with frequency-boosting interface chip," in IEEE Int. Solid-State Circuits Conf. (San Francisco, CA, USA), Feb. 2015, pp. 1-3. 
  11. S. Jung et al., "A crossbar array of magnetoresistive memory devices for in-memory computing," Nature, vol. 601, 2022, pp. 211-216. 
  12. S. Bhatti et al., "Spintronics based random access memory: a review," Materials Today, vol. 20, no. 9, 2017, pp. 530-548.