PIM or CXL-PIM? Understanding Architectural Trade-offs Through Large-Scale Benchmarking
Authors:
- I-Ting LeeILI-Ting Lee
- Bao-Kai WangBWBao-Kai Wang
- Liang-Chi ChenLCLiang-Chi Chen
- Wen Sheng LimWLWen Sheng Lim
- Da-Wei ChangDCDa-Wei Chang
- Yu-Ming ChangYCYu-Ming Chang
- Chieng-Chung HoCHChieng-Chung Ho
Preprint
Abstract
Processing-in-memory (PIM) reduces data movement by executing near memory, but our large-scale characterization on real PIM hardware shows that end-to-end performance is often limited by disjoint host and device address spaces that force explicit staging transfers. In contrast, CXL-PIM provides a unified address space and cache-coherent access at the cost of higher access latency. These opposing interface models create workload-dependent tradeoffs that are not captured by small-scale studies. This work presents a side-by-side, large-scale comparison of PIM and CXL-PIM using measurements from real PIM hardware and trace-driven CXL modeling. We identify when unified-address access amortizes link latency enough to overcome transfer bottlenecks, and when tightly coupled PIM remains preferable. Our results reveal phase- and dataset-size regimes in which the relative ranking between the two architectures reverses, offering practical guidance for future near-memory system design.Keywords
All Related Versions
This publication has 4 references indexed in Scilit:
- Compute-Enabled CXL Memory Expansion for Efficient Retrieval Augmented GenerationIEEE Micro, 2025
- An Introduction to the Compute Express Link (CXL) InterconnectACM Computing Surveys, 2024
- PID-Comm: A Fast and Flexible Collective Communication Framework for Commodity Processing-in-DIMM Devices2024
- Analysis of Data Transfer Bottlenecks in Commercial PIM Systems: A Study With UPMEM-PIMIEEE Computer Architecture Letters, 2024