CROSS

Collaborative Research: PPoSS: LARGE: �Cross-layer Coordination and Optimization for Scalable and Sparse Tensor Networks

– 2nd Annual Meeting

PIs: Jiajia Li, Frank Mueller, Dong Li, Lizhong Chen

Sep 23, 2025 @ Zoom

Future Plans

• Future tasks
• Future collaboration
• PI <-> PI
• PI <-> Advisory Committee
• PI <-> Others

​

2

Project Timeline

3

Future collaboration: PI-PI

• Jiajia Li <-> Frank Mueller
• Block sparsity exploration for tensor network approach in quantum circuit simulation
• Jiajia Li <-> Dong Li
• Accelerating tensor network on CXL machines
• Jiajia Li <-> Lizhong Chen
• Customized architecture for tensor networks
• Frank Mueller <-> Lizhong Chen
• Customized architecture for diagonal sparse format.
• Frank Mueller <-> Dong Li
• Runtime optimization for CXL machines
• All PIs
• Development AI-driven auto-tuning for tensor networks

4

Future collaboration: PI-Advisory Committee

• Jiajia Li <-> Murali Emani
• Tensor operations on Tenstorrent
• Others under exploration

5

Future collaboration: PI-Others

• PI <-> Others
• More conference attendance and advertisement
• Plan to held a workshop at ISCA’26

6

Appreciate your attention and feedback!

Future Tasks

8

T2.4

SpTC

performance

modeling

T1.1

Memory

hierarchy-aware

rep. 

T4.1

Improving data reuse

T3.2

Intelligent

data allocation 

T1.2

Memory

heterogeneity-aware

rep.

T1.3

Effective

yet efficient

(re-)arranging

Sparse Tensors

SpTC

Tensor Network

Arrange non-zeros

Data representation

+ NVRAM

Applications

T2.3

Balanced

SpTC algo.

T2.1

Cache-friendly

SpTC algo.

T2.2

Smart page

arrangement

T3.1

Memoization

Data

Methods

Computational Kernel

Memory hierarchy

Workload

Heterogenous memory

T4.2

Addressing

load imbalance

Distributed memory

GPUs

AI Testbeds

Specialized Accelerator

T4.3

Accelerating

the accelerator 

T4.5

Exploring

immerse design space

T4.4

Customized

on-chip network

Platforms

Dependency

Feedback

Thrusts