PULP Platform

A joint effort of @ETH_en, University of Bologna @Unibo + partners for Parallel Ultra-Low Power computing. Boldly designing open hardware since '13.

[PULP Platform @pulp_platform]

Cheers to both winners of the Best Paper Award of the Embedded Vision Workshop in conjunction with #CVPR2026, Lorenzo & Jaehyeon https://t.co/WIaSSxf6OG. Check out Lorenzo's TinyDEVO paper and video here: https://t.co/sE44VQ8ZOv & https://t.co/WWaXOkf32T

[PULP Platform @pulp_platform]

#CVPR2026 is underway in Denver & Lorenzo just presented "TinyDEVO: Deep Event-based Visual Odometry on Ultra-low-power Multi-core Microcontrollers" as part of the Embedded Vision Workshop. Find the slides and the poster here: https://t.co/1Fu8PoffYS https://t.co/B9Lx5LaDCm

[PULP Platform @pulp_platform]

A new addition to our team! Here is Paolo who works on MemPool, a scalable RISC-V many-core system, with interests in shared-memory architectures, on-chip communication, and SW for DSP workloads. Welcome to PULP!

[PULP Platform @pulp_platform]

Here is our paper "CHIMERA: A Flexible and Scalable 3.1 TOPS/W AI-MCU with Transformer Accelerator and 563 Gb/s Shared-L2 Memory Subsystem with QoS Guarantees". Fabricated in GF 22 FDX. See https://t.co/9QzljuHin9

[PULP Platform @pulp_platform]

We present "O-POPE: High-Frequency Pipelined Outer Product based GEMM acceleration with minimal buffering overhead" achieved by repurposing FPU pipeline registers as buffers: https://t.co/DCjZD2HtYA @Ang__93

[PULP Platform @pulp_platform]

The 22nd edition of the @hipeac summer school ACACES 2026 will take place in Fiuggi 12-18 July https://t.co/nGQYWHIxpI! Our Angelo @Ang__93 will be teaching "Hardware/Software Co-design of Energy Efficient RISC-V Edge AI Platforms". See more: https://t.co/nGQYWHIxpI

[PULP Platform @pulp_platform]

Today, Charlie The Bear would like to introduce our new team member Marco. Marco works on fault-tolerant GEMM design for satellite RISC-V clusters executing SLMs, and its physical implementation and reliability assessment. Welcome to the team!

[PULP Platform @pulp_platform]

We are getting ready for the #RISCVSummitEurope Bologna! Black or white? What do you say?

[PULP Platform @pulp_platform]

Time to celebrate! Our colleague Tim @fischetim just defended his PhD thesis "Scalable Interconnects for Integrated Accelerators". Warmest congratulations to you from Charlie The Bear and the entire PULP team.

[PULP Platform @pulp_platform]

We were delighted to welcome Martin Poviser from OpenRoad/PrecisionInno in our lab. Below, you can see Frank, Charlie and Yukio helping Martin choose the best PULP sticker for his laptop. Thank you, Alan Mishchenko, for taking the pictures!

[Luca Benini @LucaBeniniZhFe]

Can't wait to show a heterogeneous, all RISC-V "AI factory" machine... stay tuned... working with @tenstorrent @AxeleraAI and more!

[PULP Platform @pulp_platform]

RISC-V compute nodes are closing the gap with their competitors in the HPC segment! Check out our collaborative paper "Monte Cimone v3: Where RISC-V Stands in High-Performance Computing" https://t.co/kedxwcEQkj

[PULP Platform @pulp_platform]

#MLSys2026 just took place in Bellevue & we were there with Luca presenting "A Lightweight High-Throughput Collective-Capable NoC for Large-Scale ML Accelerators". See https://t.co/GedHSbTjfH https://t.co/yCLPeFZIW1 https://t.co/CYiEN7TvD6 https://t.co/gXLA5pKakR @lucacolagrande3

[PULP Platform @pulp_platform]

We are incredibly proud of all our team members who took part in this year's edition of the SOLA Stafette Zurich. We ranked 136th in over 1000 teams. An achievement that we triumphantly celebrated: https://t.co/QE3bl5Vv61
@CristiCioflan

[PULP Platform @pulp_platform]

Amirhossein's talk "Threads or Vectors? Evaluating SPMD and Vector Accelerators for Resource Constrained RISC-V Architectures" presented at #CF2026 in Catania is now online, as well. Find it here: https://t.co/cgTBg6XVjb

[PULP Platform @pulp_platform]

The slides from Yichao's @yichao_zh #CF2026 Catania talk "A 8.4 TFLOPS@16b/4.3W General-Purpose Programmable Accelerated Cluster for AI-Native RAN" are now online, as well. Check them out: https://t.co/sVymEZF13y

[PULP Platform @pulp_platform]

Maoyuan, too, presented her work "Not All Faults Are Equal: Transient-Fault Sensitivity Characterization of an Open-Source RISC-V Vector Cluster" at #CF2026 in Catania. Find her slides on our website: https://t.co/qJNhgt4uLU

[PULP Platform @pulp_platform]

And here is Kai, our new team member. Kai works on developing next-generation processing platforms for space. Welcome to the PULP team and hugs from Charlie The Bear!

[PULP Platform @pulp_platform]

Computing Frontiers are currently underway at the Museo Diocesano di Catania and we are there with Yichao @yichao_zh who will present "A 8.4 TFLOPS@16b/4.3W General Purpose Programmable Accelerated Cluster for AI-Native RAN" at 14:25 local time today. See https://t.co/s7WesJ8MDc

[PULP Platform @pulp_platform]

Today we would like to introduce PULP's new team member Cong who works on DNN deployment flows and simulation platform extensions for heterogeneous RISC-V SoCs, targeting large-scale mesh accelerators via Deeploy and GVSoC. Welcome to the team!