BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//scipy-2026//speaker//3ZSL8Z
BEGIN:VTIMEZONE
TZID:CST
BEGIN:STANDARD
DTSTART:20001029T030000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10;UNTIL=20061029T080000Z
TZNAME:CST
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
END:STANDARD
BEGIN:STANDARD
DTSTART:20071104T030000
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:CST
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20000402T030000
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=4;UNTIL=20060402T090000Z
TZNAME:CDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
END:DAYLIGHT
BEGIN:DAYLIGHT
DTSTART:20070311T030000
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:CDT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-scipy-2026-EJSCLM@pretalx.com
DTSTART;TZID=CST:20260715T160500
DTEND;TZID=CST:20260715T163500
DESCRIPTION:Awkward Array is a Python library widely used in high-energy ph
 ysics for representing and manipulating nested\, variable-length data. As 
 analysis workloads increasingly rely on GPU acceleration\, there is a need
  for solutions that deliver high performance while remaining accessible to
  Python developers. In this joint talk between the Awkward and the NVIDIA 
 teams\, we present recent developments in GPU execution for Awkward Array 
 using the new native Python CUDA support for CCCL called `cuda.compute`. T
 his novel Python interface for CCCL that enables users to achieve state-of
 -the-art GPU performance without dropping down to C++ when building new GP
 U algorithms.\n\nOur approach fuses sequences of Awkward operations into a
  minimal set of CUDA kernels\, reducing kernel launch overhead and improvi
 ng memory efficiency. Lazy execution allows entire expression graphs to be
  optimized before kernel generation\, which benefits workflows involving j
 agged arrays\, combinatorial operations\, and reductions. In addition\, th
 e design enables user-defined Python code to be incorporated into GPU exec
 ution paths with minimal boilerplate\, lowering the barrier for extending 
 Awkward with custom GPU-accelerated logic.\n\n`cuda.compute`  is a new com
 ponent in the CUDA Python ecosystem that provides native access to optimiz
 ed algorithms\, such as transforms\, reductions\, and scans. It also provi
 des a collection of iterators that defer execution of operations and enabl
 e fusing multiple operations. We demonstrate performance improvements usin
 g this approach over an eager GPU execution strategy for representative an
 alysis patterns and show how it integrates into the existing Python workfl
 ows. These developments provide a practical\, user-friendly path toward hi
 gh-performance GPU-accelerated data analysis in Python.\n\nWe thank NVIDIA
  for support and collaboration in developing the CUDA kernels and providin
 g guidance on GPU optimization strategies. Their contributions are gratefu
 lly acknowledged.
DTSTAMP:20260715T021019Z
LOCATION:Johnson Great Room
SUMMARY:GPU-Accelerated Awkward Arrays with CUDA Python - Ianna Osborne\, A
 shwin Srinath
URL:https://pretalx.com/scipy-2026/talk/EJSCLM/
END:VEVENT
END:VCALENDAR