BEGIN:VCALENDAR
PRODID:-//eluceo/ical//2.0/EN
VERSION:2.0
CALSCALE:GREGORIAN
BEGIN:VEVENT
UID:f100ab94af4a9af7ea2beea30cbe8481
DTSTAMP:20260607T173656Z
SUMMARY:Exjobbspresentation: Beamforming algorithm development for a Neural
  Processing Unit (NPU)
DESCRIPTION:Kontakt: susanna.lonnqvist@eit.lth.se\n\nExjobbet genomfördes 
 hos BeammWave AB med Samir Kumar Mishra som industrihandledare\, Victor Å
 berg som akademisk handledare och Erik Larsson examinator.This thesis inve
 stigates the development of a digital beamforming algorithm targeting effi
 cient execution on Neural Processing Unit (NPU)-oriented architectures for
  millimeter-wave (mmWave) Fifth Generation New Radio (5G NR) systems. Whil
 e mmWave frequencies enable high data rates\, they also introduce signific
 ant propagation losses that require advanced beamforming techniques that c
 an overcome this limitation reliably. &nbsp\;Our work focuses on the desig
 n and implementation of a precoding algorithm in C\, with emphasis on adap
 tations relevant to NPU deployment\, including reduced numerical precision
  and data-parallel computation. To explore performance characteristics\, t
 he algorithm was optimized using Intel's intrinsic instruction sets to lev
 erage Single Instruction Multiple Data (SIMD) execution on a CPU-based pla
 tform\, serving as a proxy for parallel hardware acceleration.Results show
  that reliable and consistent beamforming performance can be maintained un
 der reduced precision\, supporting the suitability of such approaches for 
 resource-constrained and parallel architectures. However\, the applied SIM
 D optimizations led to increased execution times\, with the Advanced Vecto
 r Extension (AVX) based implementation failing to deliver the expected per
 formance improvements.These findings highlight important tradeoffs between
  numerical efficiency and computational performance\, and emphasize the ne
 ed for hardware-specific optimization when targeting NPU-like architecture
 s. The work contributes practical insights into the challenges of mapping 
 beamforming algorithms to emerging parallel processing platforms.\n\nMer i
 nformation om händelsen: https://www.eit.lth.se/evenemang/exjobbspresenta
 tion-beamforming-algorithm-development-neural-processing-unit-npu
DTSTART;TZID=GMT:20260603T091500
DTEND;TZID=GMT:20260603T100000
LOCATION:E:3139
END:VEVENT
END:VCALENDAR