Pith. sign in

REVIEW 9 cited by

Dissecting the Graphcore IPU Architecture via Microbenchmarking

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

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 1912.03413 v1 pith:NKLQ7RPB submitted 2019-12-07 cs.DC cs.ARcs.PF

Zhe Jia , Blake Tillman , Marco Maggioni , Daniele Paolo Scarpazza This is my paper
classification cs.DCcs.ARcs.PF
keywords performancearchitecturefocusesgraphcoreintelligencemassivelynoveloffer
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

This report focuses on the architecture and performance of the Intelligence Processing Unit (IPU), a novel, massively parallel platform recently introduced by Graphcore and aimed at Artificial Intelligence/Machine Learning (AI/ML) workloads. We dissect the IPU's performance behavior using microbenchmarks that we crafted for the purpose. We study the IPU's memory organization and performance. We study the latency and bandwidth that the on-chip and off-chip interconnects offer, both in point-to-point transfers and in a spectrum of collective operations, under diverse loads. We evaluate the IPU's compute power over matrix multiplication, convolution, and AI/ML primitives. We discuss actual performance in comparison with its theoretical limits. Our findings reveal how the IPU's architectural design affects its performance. Moreover, they offer simple mental models to predict an application's performance on the IPU, on the basis of the computation and communication steps it involves. This report is the natural extension to a novel architecture of a continuing effort of ours that focuses on the microbenchmark-based discovery of massively parallel architectures.

Discussion (0). Sign in to comment.

Forward citations

Cited by 9 Pith papers

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

  1. AccelSync: Verifying Synchronization Coverage in Accelerator Pipeline Programs

    cs.AR 2026-05 conditional novelty 7.0 of 10

    AccelSync reduces synchronization correctness in accelerator pipelines to a decidable barrier-sufficiency check with O(|E|^2) complexity, proves soundness and completeness, and detects real hazards in thousands of pro...

  2. FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness

    cs.LG 2022-05 accept novelty 7.0 of 10

    FlashAttention reduces GPU high-bandwidth memory accesses in self-attention via tiling, delivering exact attention with lower IO complexity, 2-3x wall-clock speedups on models like GPT-2, and the ability to train on s...

  3. KV-RM: Regularizing KV-Cache Movement for Static-Graph LLM Serving

    cs.AR 2026-05 unverdicted novelty 6.0 of 10

    KV-RM regularizes KV-cache movement in static-graph LLM serving via block paging and merge-staged transport to improve throughput, tail latency, and memory use for variable-length decoding.

  4. TCL: Enabling Fast and Efficient Cross-Hardware Tensor Program Optimization via Continual Learning

    cs.LG 2026-04 conditional novelty 6.0 of 10

    TCL delivers 16.8x faster tuning on CPU and 12.48x on GPU with modestly lower inference latency by combining RDU active sampling, a lightweight Mamba cost model, and cross-platform continual knowledge distillation.

  5. TileLoom: Automatic Dataflow Planning for Tile-Based Languages on Spatial Dataflow Accelerators

    cs.DC 2025-12 unverdicted novelty 6.0 of 10

    TileLoom compiles tile-based languages to spatial dataflow hardware by distributing tiles across cores and optimizing data reuse via on-chip networks, delivering vendor-comparable performance on Tenstorrent systems.

  6. ClusterFusion: Expanding Operator Fusion Scope for LLM Inference via Cluster-Level Collective Primitive

    cs.DC 2025-08 conditional novelty 6.0 of 10

    ClusterFusion fuses QKV projection, attention, and output projection into one kernel using Hopper cluster collectives, cutting average decoding latency by about 1.6x on an H100.

  7. ELK: Exploring the Efficiency of Inter-core Connected AI Chips with Deep Learning Compiler Techniques

    cs.AR 2025-07 conditional novelty 6.0 of 10

    A compiler framework, Elk, jointly schedules compute, inter-core data exchange, and HBM preloading on inter-core connected AI chips, reaching 94.84 percent of an ideal roofline on LLM workloads.

  8. KV-RM: Regularizing KV-Cache Movement for Static-Graph LLM Serving

    cs.AR 2026-05 unverdicted novelty 5.0 of 10

    KV-RM regularizes KV-cache movement via block paging and coalesced transfers to improve throughput, tail latency, and memory efficiency in static-graph LLM serving without changing the decoder interface.

  9. Hardware Acceleration in Portable MRIs: State of the Art and Future Prospects

    cs.AR 2025-09 conditional novelty 4.0 of 10

    A survey arguing that hardware acceleration, especially FPGAs and edge GPUs, is central to practical portable MRI, plus a proposed Low-Field MRI Consortium.

Pith tools