Pith. sign in

REVIEW 4 cited by

Benchmarking a New Paradigm: An Experimental Analysis of a Real Processing-in-Memory Architecture

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2105.03814 v7 pith:WXAYFJVC submitted 2021-05-09 cs.AR cs.DCcs.PF

Benchmarking a New Paradigm: An Experimental Analysis of a Real Processing-in-Memory Architecture

classification cs.AR cs.DCcs.PF
keywords memoryarchitecturefirstprocessingworkloadsdataparadigmanalysis
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Many modern workloads, such as neural networks, databases, and graph processing, are fundamentally memory-bound. For such workloads, the data movement between main memory and CPU cores imposes a significant overhead in terms of both latency and energy. A major reason is that this communication happens through a narrow bus with high latency and limited bandwidth, and the low data reuse in memory-bound workloads is insufficient to amortize the cost of main memory access. Fundamentally addressing this data movement bottleneck requires a paradigm where the memory system assumes an active role in computing by integrating processing capabilities. This paradigm is known as processing-in-memory (PIM). Recent research explores different forms of PIM architectures, motivated by the emergence of new 3D-stacked memory technologies that integrate memory with a logic layer where processing elements can be easily placed. Past works evaluate these architectures in simulation or, at best, with simplified hardware prototypes. In contrast, the UPMEM company has designed and manufactured the first publicly-available real-world PIM architecture. This paper provides the first comprehensive analysis of the first publicly-available real-world PIM architecture. We make two key contributions. First, we conduct an experimental characterization of the UPMEM-based PIM system using microbenchmarks to assess various architecture limits such as compute throughput and memory bandwidth, yielding new insights. Second, we present PrIM, a benchmark suite of 16 workloads from different application domains (e.g., linear algebra, databases, graph processing, neural networks, bioinformatics).

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Clutch: High Performance Vector-Scalar Comparison using DRAM via Chunked Temporal Coding

    cs.AR 2026-06 unverdicted novelty 7.0

    Clutch accelerates vector-scalar comparisons in PuD systems via chunked temporal coding, delivering 2.9x throughput and 3.0x energy gains over prior bit-serial PuD while also mapping decision tree inference to PuD for...

  2. CODA: Algorithm-Hardware Co-design for Edge Video Diffusion via NMP-Enabled Compute-Cache Operator Disaggregation

    cs.AR 2026-07 conditional novelty 6.0

    Disaggregating cache operators from compute and overlapping them across the two classifier-free-guidance branches turns cross-timestep caching into up to 1.80x real end-to-end speedup on edge GPUs when the cache overf...

  3. A comparative study on power delivery aspects of compute-in/near-memory approaches using DRAM

    cs.AR 2026-04 unverdicted novelty 5.0

    The survey proposes a taxonomy for PIM-induced current behaviors in DRAM and analyzes how representative techniques create voltage droop and thermal issues, along with mitigation strategies using existing DRAM mechanisms.

  4. A comparative study on power delivery aspects of compute-in/near-memory approaches using DRAM

    cs.AR 2026-04 unverdicted novelty 4.0

    DRAM PIM creates bursty and concurrent current draws that stress the PDN; a temporal–spatial taxonomy organizes those stresses and points to controller and placement mitigations.