Answers are grounded in this paper and cite their sources.
In this work, we propose a device-circuit-system co-design approach to achieve the best performance from a given variation-prone resistive random access memory (RRAM) compute-in-memory (CIM) hardware for deep neural network (DNN) acceleration and also demonstrate an optimized hardware implementation of block floating point (BFP) acceleration exploiting FP8 format on RRAM CIM hardware using a 1T-1R array of single-level cell (SLC) and multilevel cell (MLC) devices. Further, bit-level accurate network simulations incorporating device-to-device (D2D) variations on multi-bit SLC / MLC weight mantissa / weight are also conducted and benchmarking is performed with integer (INT4, INT8) formats. In particular, a word line read voltage (<tex-math notation="LaTeX">VWLread</tex-math>) tuning scheme is proposed and experimentally analyzed to (a) optimize D2D variation (tested on 144 1T-1R array bit cells) in SLC-based implementation to reduce partial MAC (PMAC) mapping errors for robust inference and b) obtain linearly separated conductance states for linear accumulation in MLC-based implementation (tested on 200 1T-1R bitcells). An optimized write-verify scheme is also proposed to obtain a tight MLC states distribution. Extensive analysis on hardware obtained FP8 inference results is also performed by measuring 16,384<tex-math notation="LaTeX">$<$</tex-math>PMACs<tex-math notation="LaTeX">$<$</tex-math> 1,31,072 from hardware to identify sources of error for different word-line parallelism, SLC / MLC mapping and their trade-off with sparsity and energy consumption. For SLC based hardware implementation, the test accuracy of the FP8 trained networks at <tex-math notation="LaTeX">σ/μ =</tex-math> 0.5 is 1.2x better than INT4 at only <tex-math notation="LaTeX">~</tex-math> 1.3x the area consumption of the INT4 RRAM array. INT8 network performance on the other hand, is comparable to FP8 but at <tex-math notation="LaTeX">~</tex-math> 2x FP8 RRAM array area consumption. Hardware inference using 2-bit MLC mapped weight mantissas is also demonstrated to analyze the effect of individual state variation dependence and identify critical variability states.
No takes yet. Share an insight, caveat, or question.
Shaban et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: