BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260731T133319Z
LOCATION:Bldg. 8 - B 102
DTSTART;TZID=Europe/Stockholm:20260701T115200
DTEND;TZID=Europe/Stockholm:20260701T121500
UID:submissions.pasc-conference.org_PASC26_sess185_msa271@linklings.com
SUMMARY:Exploring Tenstorrent RISC-V Accelerators for High-Performance Sci
 entific Computing: An N-Body Case Study
DESCRIPTION:Elisabetta Boella (E4 COMPUTER ENGINEERING S.p.A.)\n\nRISC‑V–b
 ased accelerators have recently seen rapid growth, primarily driven by art
 ificial intelligence workloads. However, some architectures also show stro
 ng potential for high‑performance scientific computing. One example is the
  Tenstorrent n300 (“Wormhole”) accelerator.\nTo evaluate its capabilities,
  we developed an N‑body application that computes gravitational interactio
 ns among particles, offloading the most computationally intensive componen
 t, the force calculation kernel, to the accelerator. The implementation us
 es the Tenstorrent TT‑Metalium API and leverages overlap between computati
 on and data movement.\nPerformance was compared against a highly optimized
  CPU version employing AVX‑512 vectorization and OpenMP parallelism across
  32 cores. For the same simulation, the Tenstorrent n300 achieved a 2× spe
 edup while also reducing energy consumption by a similar factor.\nWe furth
 er extended the application to support multiple accelerator cards, enablin
 g larger simulations beyond a single device’s capacity. To our knowledge, 
 this is the first scientific computing application demonstrating parallel 
 execution across multiple Tenstorrent n300 accelerators.\nThese results hi
 ghlight the potential of emerging RISC‑V–based accelerators as efficient a
 nd scalable alternatives for scientific workloads.\n\nSession Chair: Prash
 anth Kanduri (ETH Zurich / CSCS)\n\n
END:VEVENT
END:VCALENDAR