BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.com//euroscipy-2026//speaker//VKG8RE
BEGIN:VTIMEZONE
TZID:Poland
BEGIN:DAYLIGHT
DTSTART:20250720T000000
TZNAME:CEST
TZOFFSETFROM:+0200
TZOFFSETTO:+0200
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20251026T030000
RDATE:20261025T030000
TZNAME:CET
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20260329T030000
RDATE:20270328T030000
TZNAME:CEST
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
SUMMARY:Python Tile Programming for GPUs - Bryce Adelstein Lelbach
DTSTART;TZID=Poland:20260720T113000
DTEND;TZID=Poland:20260720T120000
DTSTAMP:20260907T124817Z
UID:pretalx-euroscipy-2026-G3SRPL@pretalx.com
DESCRIPTION:Parallel programming can be intimidating\, but doesn’t need 
 to be! Tile-based programming models make GPU parallelism more newcomer-fr
 iendly\, highly productive\, and still fast by letting you write sequentia
 l\, array-centric code while the framework handles parallelization\, synch
 ronization\, and data movement.\n\nIn this example-driven talk\, we’ll i
 ntroduce tile-based programming in Python using NVIDIA’s new stack: [cuT
 ile](https://github.com/NVIDIA/cutile-python) and its compiler foundation\
 , [Tile IR](https://github.com/NVIDIA/cuda-tile). You’ll see recently an
 nounced CUDA Tile capabilities in action\, including multi-GPU communicati
 on\, interoperability with traditional CUDA SIMT\, and support for more di
 verse kernels such as convolutions and stencils. We’ll compare tile and 
 SIMT approaches\, build intuition for performance and execution\, and demo
 nstrate practical debugging and reasoning techniques. Along the way\, you
 ’ll see real workloads: HPC stencils\, an SPMV plus CG solver\, and ML m
 odels from [TileGym](https://github.com/NVIDIA/TileGym). You’ll leave wi
 th a clear sense of when tile programming helps\, and how it enables more 
 portable high-performance Python as hardware trends evolve.
LOCATION:Room 1.38 (Ground Floor\, Turing)
URL:https://pretalx.com/euroscipy-2026/talk/G3SRPL/
END:VEVENT
BEGIN:VEVENT
SUMMARY:Profiling Python GPU Code - Bryce Adelstein Lelbach
DTSTART;TZID=Poland:20260720T152000
DTEND;TZID=Poland:20260720T154000
DTSTAMP:20260907T124817Z
UID:pretalx-euroscipy-2026-VPYLDF@pretalx.com
DESCRIPTION:Your GPU is fast\, so why does your Python code still feel slo
 w? This talk shows a practical\, Python-first profiling workflow with Nsig
 ht Systems\, Nsight Compute\, and NVTX for CuPy\, Numba\, PyTorch\, JAX\, 
 and CUDA extensions. We will use timelines to find launch overhead\, hidde
 n synchronizations\, and host-device copies\, then drill into kernel bottl
 enecks like memory throughput and occupancy. You will leave with a repeata
 ble loop for turning profiles into measurable speedups.
LOCATION:Room 1.38 (Ground Floor\, Turing)
URL:https://pretalx.com/euroscipy-2026/talk/VPYLDF/
END:VEVENT
END:VCALENDAR