Tootfinder

Overcoming logarithmic singularities in the Cahn-Hilliard equation with Flory-Huggins potential: An unconditionally convergent ADMM approach
Ruo Li, Shengtong Liang, Zhonghua Qiao
https://arxiv.org/abs/2506.09361

Overcoming logarithmic singularities in the Cahn-Hilliard equation with Flory-Huggins potential: An unconditionally convergent ADMM approach
The Cahn-Hilliard equation with Flory-Huggins potential serves as a fundamental phase field model for describing phase separation phenomena. Due to the presence of logarithmic singularities at $u=\pm 1$, the solution $u$ is constrained within the interval $(-1,1)$. While convex splitting schemes are commonly employed to preserve this bound and guarantee unconditional unique solvability, their practical implementation requires solving nonlinear systems containing singular logarithmic terms at ea…

ECLIP: Energy-efficient and Practical Co-Location of ML Inference on Spatially Partitioned GPUs
Ryan Quach, Yidi Wang, Ali Jahanshahi, Daniel Wong, Hyoseung Kim
https://arxiv.org/abs/2506.12598

ECLIP: Energy-efficient and Practical Co-Location of ML Inference on Spatially Partitioned GPUs
As AI inference becomes mainstream, research has begun to focus on improving the energy consumption of inference servers. Inference kernels commonly underutilize a GPU's compute resources and waste power from idling components. To improve utilization and energy efficiency, multiple models can co-locate and share the GPU. However, typical GPU spatial partitioning techniques often experience significant overheads when reconfiguring spatial partitions, which can waste additional energy through rep…