Fix underflow in Triton's highestPowOf2Divisor function when the input is INT_MIN#11

Closed

Moerafaat wants to merge 27 commits intomainfrom export_cl666302296

+888-858

Commits on Aug 23, 2024

[TEST] Use device fixture for test_math_extern (triton-lang#4558 )
int3
authored
[FRONTEND] interleave does not need to check shape (triton-lang#4535 )

Mwsxy
and
谢双镱
authored
[Proton] Add a better description when possibly importing incorrect hatchet in Proton (triton-lang#4565 )
CRobeck
authored
[BACKEND] Add a knob to fall back to the legacy mma layout conversion code (triton-lang#4561 )
Jokeren
authored
[BE][PIPELINE] Handle the case when values from the peeled prologue may escape out of the loop (triton-lang#4567 )
pawelszczerbuk
authored
[FRONTEND] Print full file name when overriding kernel (triton-lang#4566 )
htyu
authored
[Proton] Move additional hatchet import into try/except check (triton-lang#4568 )
CRobeck
authored
[AMD] Get libamdhip64.so loaded in process first in Python (triton-lang#4255 )

xinyazhang
and
antiagainst
authored

Commits on Aug 24, 2024

[AMD] Support FP8E5M2 with MFMA FP16 instructions (triton-lang#4259 )

binarman
and
antiagainst
authored

Commits on Aug 27, 2024

Commits on Aug 28, 2024