BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//cfp.riscv-europe.org//eu-summit-2026//talk//S3LMTB
BEGIN:VTIMEZONE
TZID:CET
BEGIN:STANDARD
DTSTART:20001029T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000326T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-eu-summit-2026-S3LMTB@cfp.riscv-europe.org
DTSTART;TZID=CET:20260611T104000
DTEND;TZID=CET:20260611T105000
DESCRIPTION:Conservative/qualification-sensitive RISC-V ecosystems tend to 
 view large architectural changes as costly due to hardware overhead\, inte
 gration effort\, software/toolchain adaptation\, and assurance scope. This
  is especially relevant for platforms intended for harsh environments and 
 long lifetimes such as space-oriented and radiation-tolerant platforms (e.
 g.\, NOEL-V). At the same time\, there is growing interest in on-board pro
 cessing to support time-critical decisions close to the sensor and reduce 
 reliance on transmitting raw sensor data\, increasing the demand for compu
 te-intensive Edge-AI inference. In such settings\, full vector architectur
 es can deliver high throughput\, but they tend to introduce additional arc
 hitectural state and increase integration complexity across the hardware a
 nd software stack. Therefore\, to introduce data-parallel acceleration wit
 h minimal disruption\, we evaluate packed-SIMD as a small-change alternati
 ve based on packed subword parallelism that remains close to the existing 
 register and memory model.\nWe consider two packed-SIMD options: SWAR and 
 SPARROW. On a NOEL-V softcore\, we implement SWAR operator kernels for the
  most computationally expensive layers and integrate them into the math ba
 ckend of a space prequalified inference engine\, running on a space prequa
 lified RTOS (RTEMS6 SMP). Using a hardware SWAR unit for packed subword op
 erations\, we report full-model results with and without SWAR acceleration
 \, showing improved inference performance without requiring a full vector 
 architecture. Finally\, we outline future work extending the same backend 
 methodology to SPARROW to compare performance across packed-SIMD options.
DTSTAMP:20260522T163537Z
LOCATION:Poster Island C
SUMMARY:RISC-V Packed-SIMD Acceleration for Quantized Edge-AI Inference on 
 Space-Qualified Platforms - Carlos Rafael Tordoya Taquichiri
URL:https://cfp.riscv-europe.org/eu-summit-2026/talk/S3LMTB/
END:VEVENT
END:VCALENDAR