Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2506.23405

CMOS+X: Stacking Persistent Embedded Memories based on Oxide Transistors upon GPGPU Platforms

Faaiq Waqar·Ming-Yen Lee·Seongwon Yoon·Seongkwang Lim·Shimeng Yu
View original source ↗Read source document ↗Export metadata ↓

Abstract

In contemporary general-purpose graphics processing units (GPGPUs), the continued increase in raw arithmetic throughput is constrained by the capabilities of the register file (single-cycle) and last-level cache (high bandwidth), which require the delivery of operands at a cadence demanded by wide single-instruction multiple-data (SIMD) lanes. Enhancing the capacity, density, or bandwidth of these memories can unlock substantial performance gains; however, the recent stagnation of SRAM bit-cell scaling leads to inequivalent losses in compute density. To address the challenges posed by SRAM's scaling and leakage power consumption, this paper explores the potential CMOS+X integration of amorphous oxide semiconductor (AOS) transistors in capacitive, persistent memory topologies (e.g., 1T1C eDRAM, 2T0C/3T0C Gain Cell) as alternative cells in multi-ported and high-bandwidth banked GPGPU memories. A detailed study of the density and energy tradeoffs of back-end-of-line (BEOL) integrated memories utilizing monolithic 3D (M3D)-integrated multiplexed arrays is conducted, while accounting for the macro-level limitations of integrating AOS candidate structures proposed by the device community (an aspect often overlooked in prior work). By exploiting the short lifetime of register operands, we propose a multi-ported AOS gain-cell capable of delivering 3x the read ports in ~76% of the footprint of SRAM with over 70% lower standby power, enabling enhancements to compute capacity, such as larger warp sizes or processor counts. Benchmarks run on a validated NVIDIA Ampere-class GPU model, using a modified version of Accel-Sim, demonstrate improvements of up to 5.2x the performance per watt and an average 8% higher geometric mean instruction per cycle (IPC) on various compute- and memory-bound tasks.

Explore related subjects

cs.ETcs.AR

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Faaiq Waqar, Ming-Yen Lee, Seongwon Yoon, Seongkwang Lim, Shimeng Yu. 2025-06-29. CMOS+X: Stacking Persistent Embedded Memories based on Oxide Transistors upon GPGPU Platforms. https://arxiv.org/abs/2506.23405

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Beyond HBM-on-GPU: Thermal Design Envelope for 3D Volumetric DRAM-on-GPU Integration

The scaling of GPUs for AI and HPC workloads is increasingly constrained by the capacity, bandwidth, and thermal limits of both 2.5D HBM-GPU and direct-stacked 3D HBM-on-GPU integration. This work establishes the thermal design envelope for 3D volumetric DRAM-on-GPU integration, in which vertically oriented DRAM dies and interleaved cooling cavities reshape heat flow and memory interfacing above the GPU. Using a package-level thermal model anchored to a consistent HBM-on-GPU baseline and driven by a realistic reticle-scale non-uniform GPU power map, we quantify the key parameters governing thermal feasibility. Stack height is the dominant limiter of peak temperature, while cooling-cavity conductivity shifts the feasible region, and mold insertion and stack orientation further modulate thermal behavior. A distributed memory-controller and network-on-chip tier introduces only a moderate thermal penalty. Although die-level parallelism increases bandwidth, the reduction in simulated training time saturates once execution becomes compute-bound. These results define a bounded co-design space across bandwidth, capacity, and thermal constraints for 3D volumetric DRAM-on-GPU integration.

cs.ET↗
arXiv2026

Droop-Aware Foundation Model Power Flow

This paper develops a droop-aware extension of the GridFM power systems foundation model, embedding droop gains and frequency/voltage deadband parameters as per-bus node features to enable control-aware AC power-flow analysis. Existing power-flow datasets encode only static electrical features, conflating operating points from qualitatively different control regimes; this work resolves that gap by exposing droop and deadband parameters as structured node features, with deadband discontinuities handled through a smooth tanh approximation that preserves solver differentiability. A transformer-based graph neural network is pre-trained on masked reconstruction and fine-tuned on the resulting control-aware datasets. The framework is validated against PSCAD electromagnetic-transient simulations on a two-bus system (0.11% maximum steady-state error) and cross-validated against an independent PyPower droop solver on the IEEE 24-bus RTS. On the 24-bus system the surrogate attains R2 = 0.9996 for active generation and 0.0015 p.u. voltage-magnitude RMSE; scalability is confirmed on the IEEE 300-bus system (0.0036 p.u. RMSE, R2 = 0.9841 for voltage magnitude across 299,700 predictions). A three-mode control study further shows that the deadband widens the control-error distribution while leaving total droop compensation unchanged, establishing deadband width as an actionable node-level design feature.

cs.ET↗
arXiv2026

GNN-based Path-aware multi-view Circuit Learning for Technology Mapping

Traditional technology mapping suffers from systemic inaccuracies in delay estimation due to its reliance on abstract, technology-agnostic delay models that fail to capture the nuanced timing behavior behavior of real post-mapping circuits. To address this fundamental limitation, we introduce GPA(graph neural network (GNN)-based Path-Aware multi-view circuit learning), a novel GNN framework that learns precise, data-driven delay predictions by synergistically fusing three complementary views of circuit structure: And-Inverter Graphs (AIGs)-based functional encoding, post-mapping technology emphasizes critical timing paths. Trained exclusively on real cell delays extracted from critical paths of industrial-grade post-mapping netlists, GPA learns to classify cut delays with unprecedented accuracy, directly informing smarter mapping decisions. Evaluated on the 19 EPFL combinational benchmarks, GPA achieves 19.9%, 2.1% and 4.1% average delay reduction over the conventional heuristics methods (techmap, MCH) and the prior state-of-the-art ML-based approach SLAP, respectively-without compromising area efficiency.

cs.ET↗