tree ef5d69113aec10a17f3b0a2f08b702226f965da4
parent 5b112cbc21046aaa028a05da62da1e3f8d19bc20
author Stanley Winata <68087699+raikonenfnu@users.noreply.github.com> 1721447408 -0700
committer GitHub <noreply@github.com> 1721447408 -0700
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsFcBAABCAAQBQJmmzPwCRC1aQ7uu5UhlAAAYF8QAD90EH+36Ca1qNqXQDPMd8aF
 uU5Kdbx0LRmkWN66H8WLVrCOPwdGQ93fZaj3fKH8RE1makshQoW2+qGfTn+zekYV
 SJ/cbRlyF0ti2jN0whr7BhSgnLPPt76Zez+7qfrDOBRGOQZNUgOwGOJWRLbgCMJQ
 UVzYbYi2t1KxiyfkaJjcpWYGhvsqcvt8MC4KKwBsY25rJTpy5Z9/8q8aV+cb8o58
 X55ipFgr36UA5m2u3MRz6ZuCVEvBOQvsYbK4VTG/zC3NMbOo2IhZlXwmq2uU0B35
 rVWNlvPxGSvOX8jCb5x+NqBiB/CZtpGw+kZCUSuwqoV+j3RaG1jZDCn6hS4aKe45
 bdfBQ02AB102xHqN/ek7zYZvw1J7fpaiP2phQTM4FUDx+iso3aX05kwM2CHv68TS
 agwBofaheHUTK5eAhvdcXrXd3SvcWMnSTK0gsSqo3u8fu+M7LtCxyF2mYy3j73QB
 rS7IOSJXrvs/H5CqgDH4EVU04vG8PCKTz9JfEX4FhJMeg1w2qVr6XklbnAspi5er
 X7pR/IvNch97Gi/C5/cSVozXI2kTvfUaN1OrvYp+RiIJXq/ocK2lHfVzDEDbrbA+
 zF8IyvTb2CDe4R/CjbT1pmwmpISx2SJHxRZXq/J72db2Z11xIz3jN9MDvx3xHOro
 qN1FhKfJscgz+55IRzxx
 =jR5A
 -----END PGP SIGNATURE-----
 

[GPU][VectDist] Refactor multiReducOp lowering to reduce acc at the end. (#17974)

The main motivation behind this commit is to fix the numerics issue we
find in Attention GPU C++ pipeline. Where our output seems to be some
scale off from reference (i.e out = k * ref). Through experiments we
determine that the cause of the issue is the multiReducOp distribution,
which is required for scaling/merge computation.

Through analyzing of IR and experimentation, reducing the srcVector and
non constant accumulator at the same time with another multiDimReduction
seems to output numerically wrong values.

This commit fixes numerical issue by refactoring the distribution of
multireductionOp to 3 steps. First every thread locally reduce the
srcVector it holds with combiningIdentity as localInit. Second, it does
a subgroup/warp reduce amongs other threads. Finally, each thread does a
local reduction of the intermediate reduced data it has with the
accumulator it holds.

---------

Signed-off-by: Stanley Winata <stanley.winata@amd.com>