jdk
8307513: C2: intrinsify Math.max(long,long) and Math.min(long,long)
#20098
Open

8307513: C2: intrinsify Math.max(long,long) and Math.min(long,long) #20098

galderz wants to merge 36 commits into openjdk:master from galderz:topic.intrinsify-max-min-long
galderz
galderz 8307513: C2: intrinsify Math.max(long,long) and Math.min(long,long)
3dd72b89
bridgekeeper
openjdk
openjdk openjdk added rfr
openjdk
openjdk openjdk added graal
openjdk openjdk added hotspot
openjdk openjdk added core-libs
mlbridge
jaskarth
jaskarth commented on 2024-07-10
iwanowww
jddarcy
jddarcy
jddarcy
jddarcy approved these changes on 2024-07-12
openjdk
galderz
jaskarth
galderz Add IR test
e43b3907
galderz Refactor inline methods to unify their implementations
f9107393
galderz Add math vectorized JMH benchmark
ce71a0e6
galderz
bridgekeeper
galderz Rename benchmark class to MathLoopBench
8d66f7bf
galderz Fix multi long tests to use long arrays
605a78a4
galderz
franz1981
galderz
galderz Implement cmovL as a jump+mov branch
1522e26b
galderz Switch movl to movq
a64fcdab
galderz Fix format of assembly for the movl to movq switch
13ed8729
galderz Distribute values targetting a branch percentage
da720c5a
galderz Fix min case to distribute numbers as per probability
0b71cb5d
galderz Fix compilation error
fe3aff41
galderz Add an intermediate % that is more representative of real life
0047a4b2
galderz Skip single array benchmarks
f622852d
galderz Add min/max benchmark that includes loops and reductions
6fd8805f
galderz Renamed benchmark methods
93799d5d
galderz Multiply array value in reduction for vectorization to kick in
c06e869a
galderz Remove previous benchmark effort
28778c84
galderz
galderz Revert "Fix format of assembly for the movl to movq switch"
bc648aa2
galderz Revert "Switch movl to movq"
7a07aa82
galderz Revert "Implement cmovL as a jump+mov branch"
16ae2a33
galderz
galderz
galderz
theRealAph
minborg
TobiHartmann
franz1981
franz1981 commented on 2024-10-04
galderz
TobiHartmann
TobiHartmann
galderz Merge branch 'master' into topic.intrinsify-max-min-long
3f712e23
galderz Avoid creating result array in benchmark method
6cc5484d
galderz Encapsulate benchmark state within an inner class
c9560127
galderz Add clipping range benchmark that uses min/max
0b197899
galderz Restore previous benchmark iterations and default param size
e6698934
galderz Make state class non-final
dcf6b547
galderz Double/Float tests only when avx enabled
b19fc816
galderz Renamed benchmark class
f6f02449
galderz Use same default size as in other vector reduction benchmarks
0a8718e1
galderz
galderz
rwestrel
rwestrel approved these changes on 2024-11-04
galderz
openjdk
galderz
eme64
eme64 commented on 2024-10-24
galderz
galderz Merge branch 'master' into topic.intrinsify-max-min-long
aca09222
eme64
galderz Add empty line
65e2e48c
galderz Add max reduction test
c964c264
galderz Fix style
cfe0239f
galderz Adjust min/max identity IR test expectations after changes
7353a072
galderz Added comment around the assertions
130b4755
galderz
eme64
eme64 commented on 2024-12-18
eme64
fg1417
fg1417 commented on 2024-12-20

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone