~/lightning

Lightning Talk: Enhancements Made to MPS Backend in PyTorch for Applications Running...

on YouTube 08m
Lightningat PyTorch Conference 2023 · 2023-10-16..10-17· [machine-learning]

speaker

name:Kulin Seth
links:none on file

resources

video:https://youtu.be/Np8YEW011dg
slides:none on file

summary

MPS backend qualifies as beta in PyTorch 2.0 with support for 300+ operators and expanded network coverage; profiling is integrated via OS signpost and Metal System Trace (MPSProfiler & Instruments) so developers can see blit calls, MPS execution and CPU fallbacks. Custom kernels compile as C++/Objective‑C extensions using APIs like get_command_buffer, get_dispatch_q, synchronize and commit_and_continue to eliminate CPU fallbacks (softshrink replaced) and reduce copies. MacOS 14 results show up to 5x speedup across TorchBench workloads.

AI-written summary · about

chapters (4)

  1. 0:00MPS backend overview
  2. 2:01Profiler and monitoring
  3. 4:35Custom kernel development
  4. 7:27APIs and performance results

related talks

~/search
type to search 6,249 conferences, 138,987 speakers and 185,172 talks
~/help
cmdctrl k
search every conference, speaker and series
/
focus the filter search
h
home
c
conferences
s
speakers
e
series
t
light / dark — the status bar button does the same
?
this help
esc
close this, then clear every filter — or just the search, while you are typing in it

Ignored while you are typing. Pages without a filter rail have no search to focus.