# SPICE

Published articles for SPICE.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## SPICE: Sparse Parameter Gossip for LLM Pretraining

DevFeed: [SPICE: Sparse Parameter Gossip for LLM Pretraining](<https://devfeed.tech/articles/spice-sparse-parameter-gossip-for-llm-pretraining-140393.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.23053>)

Author: Pietro Cagnasso, Daniele Kopyshevskiy, Edouard Oyallon, Eugene Belilovsky

Published: 2026-11-18T05:00:00Z

Content type: news

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [distributed-training](<https://devfeed.tech/topics/distributed-training.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [bandwidth-constraints](<https://devfeed.tech/tags/bandwidth-constraints.md>), [better](<https://devfeed.tech/tags/better.md>), [combines](<https://devfeed.tech/tags/combines.md>), [communication](<https://devfeed.tech/tags/communication.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [ddp](<https://devfeed.tech/tags/ddp.md>), [decentralized](<https://devfeed.tech/tags/decentralized.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [error](<https://devfeed.tech/tags/error.md>), [feedback](<https://devfeed.tech/tags/feedback.md>), [full](<https://devfeed.tech/tags/full.md>), [global](<https://devfeed.tech/tags/global.md>), [internet](<https://devfeed.tech/tags/internet.md>), [language](<https://devfeed.tech/tags/language.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [less](<https://devfeed.tech/tags/less.md>), [link](<https://devfeed.tech/tags/link.md>), [llm](<https://devfeed.tech/tags/llm.md>), [pretraining](<https://devfeed.tech/tags/pretraining.md>), [spice](<https://devfeed.tech/tags/spice.md>)

### AI overview

SPICE reduces communication costs in decentralized LLM pretraining by exchanging sparse parameter updates and using error feedback with corrected parameter values that remain interpretable as peer neighborhoods change. This lets it contact broader neighborhoods with less communication than the evaluated dense decentralized baselines; as replicas increase, it adds peers while keeping messages sparse, limiting the quality degradation seen with dense gossip. Under bandwidth constraints, the abstract reports higher training throughput than DDP and dense gossip, with throughput sustained during transient link slowdowns.

### Source excerpt

arXiv:2611.23053v1 Announce Type: new Abstract: Pretraining language models across geographically distributed resources requires communication-efficient methods that operate under limited and variable internet bandwidth. Decentralized training replaces global collectives with peer exchanges, but full-model messages make larger peer neighborhoods costly. We introduce SPICE, which combines sparse parameter gossip with error feedback through corrected parameter values that remain directly interpretable as peer neighborhoods change. These sparse exchanges make broader neighborhoods affordable, achieving better training quality with less communication than the evaluated dense decentralized baselines. As the replica count grows, SPICEincreases the number of peers contacted while keeping individual messages sparse, limiting the quality degradation observed with dense gossip. Under bandwidth constraints, SPICE achieves higher training throughput than DDP and dense gossip, while sustaining throughput during transient link slowdowns.

## AnalogScout: Simulation-Efficient PVT-Aware Analog Sizing with LLM Agents

DevFeed: [AnalogScout: Simulation-Efficient PVT-Aware Analog Sizing with LLM Agents](<https://devfeed.tech/articles/analogscout-simulation-efficient-pvt-aware-analog-sizing-with-llm-agents-103257.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.37319>)

Author: Yichen Long, Keren Zhu

Published: 2026-10-28T04:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Chip design](<https://devfeed.tech/topics/chip-design.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [analog](<https://devfeed.tech/tags/analog.md>), [bayesian](<https://devfeed.tech/tags/bayesian.md>), [budget](<https://devfeed.tech/tags/budget.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [evolution](<https://devfeed.tech/tags/evolution.md>), [gm](<https://devfeed.tech/tags/gm.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-agents](<https://devfeed.tech/tags/llm-agents.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [power](<https://devfeed.tech/tags/power.md>), [pvt](<https://devfeed.tech/tags/pvt.md>), [reuse](<https://devfeed.tech/tags/reuse.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [specifications](<https://devfeed.tech/tags/specifications.md>), [spice](<https://devfeed.tech/tags/spice.md>), [time](<https://devfeed.tech/tags/time.md>), [verification](<https://devfeed.tech/tags/verification.md>)

### AI overview

AnalogScout combines reuse of verified elite designs, analytic gm/Id local sizing, and global refinement with full-corner verification to reduce the multi-corner SPICE simulation burden in PVT-aware analog sizing. On three increasingly complex SKY130 OTAs, it reports 43.3% lower power than CMA-ES at an equal 16,000-simulation budget, and reaches CMA-ES's final power using 7.0% of that budget. It also reports lower power than Bayesian optimization on all three OTAs and than two reimplemented LLM-driven flows on the two OTAs where those flows were tested.

### Source excerpt

arXiv:2610.37319v1 Announce Type: cross Abstract: Analog device sizing must satisfy all specifications across process, voltage, and temperature (PVT) corners to ensure tape-out success, but simulating every candidate at all corners dominates the sizing time and slows design iteration. Conventional surrogate-based optimizers search the sizing space directly, while recent large language model (LLM) agents add circuit knowledge to designs but still delegate the sizes to global optimizers. Both learn the circuit response from repeated multi-corner SPICE simulations and optimize on top of it, which becomes expensive under a limited simulation budget. Providing an agent with reusable verified designs and sizing sub-blocks analytically could save most of this budget. We propose AnalogScout, a simulation-efficient PVT-aware sizing framework driven by an LLM agent. Combining elite-design reuse, analytic gm/Id local sizing and global refinement under full-corner verification, AnalogScout reduces power by 43.3% relative to the evolution strategy CMA-ES at an equal budget of 16k simulations on three SKY130 OTAs of increasing complexity, and reaches the final CMA-ES power after 7.0% of that budget. It also attains lower power than Bayesian optimization on all three OTAs and than two re-implemented LLM-driven flows on the two OTAs where they were tested.

## Neural Networks Speed 3nm Nanosheet SPICE Simulation (TU Munich)

DevFeed: [Neural Networks Speed 3nm Nanosheet SPICE Simulation (TU Munich)](<https://devfeed.tech/articles/neural-networks-speed-3nm-nanosheet-spice-simulation-tu-munich-99310.md>)

Original publisher: [Read original article](<https://semiengineering.com/neural-networks-speed-3nm-nanosheet-spice-simulation-tu-munich/>)

Author: Technical Paper Link

Published: 2026-10-25T20:14:26Z

Content type: news

Language: en

Sources: [Semiconductor Engineering](<https://devfeed.tech/sources/semiconductor-engineering.md>)

Topics: [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Simulation](<https://devfeed.tech/topics/simulation.md>), [Chip design](<https://devfeed.tech/topics/chip-design.md>)

Tags: [3nm](<https://devfeed.tech/tags/3nm.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [accelerate](<https://devfeed.tech/tags/accelerate.md>), [ai-ml-dl](<https://devfeed.tech/tags/ai-ml-dl.md>), [based](<https://devfeed.tech/tags/based.md>), [bsim-cmg](<https://devfeed.tech/tags/bsim-cmg.md>), [characterization](<https://devfeed.tech/tags/characterization.md>), [circuit](<https://devfeed.tech/tags/circuit.md>), [compact-models](<https://devfeed.tech/tags/compact-models.md>), [complex](<https://devfeed.tech/tags/complex.md>), [design-verification](<https://devfeed.tech/tags/design-verification.md>), [device](<https://devfeed.tech/tags/device.md>), [device-modeling](<https://devfeed.tech/tags/device-modeling.md>), [eda](<https://devfeed.tech/tags/eda.md>), [fast](<https://devfeed.tech/tags/fast.md>), [library](<https://devfeed.tech/tags/library.md>), [library-characterization](<https://devfeed.tech/tags/library-characterization.md>), [nanosheet-transistors](<https://devfeed.tech/tags/nanosheet-transistors.md>), [networks](<https://devfeed.tech/tags/networks.md>), [neural](<https://devfeed.tech/tags/neural.md>), [neural-networks](<https://devfeed.tech/tags/neural-networks.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [speed](<https://devfeed.tech/tags/speed.md>), [spice](<https://devfeed.tech/tags/spice.md>), [spice-simulation](<https://devfeed.tech/tags/spice-simulation.md>), [standard-cells](<https://devfeed.tech/tags/standard-cells.md>), [static-timing-analysis](<https://devfeed.tech/tags/static-timing-analysis.md>), [technical-papers](<https://devfeed.tech/tags/technical-papers.md>), [technical-university-of-munich](<https://devfeed.tech/tags/technical-university-of-munich.md>), [transistor-modeling](<https://devfeed.tech/tags/transistor-modeling.md>), [transistors](<https://devfeed.tech/tags/transistors.md>), [tu-munich](<https://devfeed.tech/tags/tu-munich.md>)

### AI overview

The researchers developed neural-network transistor models for a 3 nm nanosheet that predict drain current and charges, then tested them in standard cell library characterization and complex analog- and digital-level circuit simulations. Against the conventional Verilog-A BSIM-CMG model, they report accurate delay and power estimates, with errors below 0.11% for delay and 0.76% for total power across a wide range of circuits. Library characterization took up to eight times less time. The work also examines how network size affects simulation precision. The abstract presents the models as a potential alternative, with evaluation focused on the reported circuits and characterization tasks.

### Source excerpt

Researchers at Technical University of Munich published "NN-Based Transistor Modeling for Ultra-Fast SPICE Simulation and Standard Cell Library Characterization." Abstract: "Neural network (NN)-based transistor models have proven to be a promising solution to accelerate device modeling. Although these models demonstrated remarkable speedup in circuit simulations, they were not rigorously tested in complex tasks within standard... " read more The post Neural Networks Speed 3nm Nanosheet SPICE Simulation (TU Munich) appeared first on Semiconductor Engineering.