If we vectorize a e.g. store, we leave around a bunch of getelementptrs for the individual scalar stores which we removed. We can go ahead and delete them as well.
This is purely for test output quality and readability. It should have no effect in any sane pipeline.
Differential Revision: https://reviews.llvm.org/D122493
The original commit exposed several missing dependencies (e.g. latent bugs in SLP scheduling). Most of these were fixed over the weekend and have had several days to bake. The last was fixed this morning after being noticed in manual review of test changes yesterday. See the review thread for links to each change.
Original commit message follows:
SLP currently schedules all instructions within a scheduling window which stretches from the first instruction potentially vectorized to the last. This window can include a very large number of unrelated instructions which are not being considered for vectorization. This change switches the code to only schedule the sub-graph consisting of the instructions being vectorized and their transitive users.
This has the effect of greatly reducing the amount of work performed in large basic blocks, and thus greatly improves compile time on degenerate examples. To understand the effects, I added some statistics (not planned for upstream contribution). Here's an illustration from my motivating example:
Before this patch:
704357 SLP - Number of calcDeps actions
699021 SLP - Number of schedule calls
5598 SLP - Number of ReSchedule actions
59 SLP - Number of ReScheduleOnFail actions
10084 SLP - Number of schedule resets
8523 SLP - Number of vector instructions generated
After this patch:
102895 SLP - Number of calcDeps actions
161916 SLP - Number of schedule calls
5637 SLP - Number of ReSchedule actions
55 SLP - Number of ReScheduleOnFail actions
10083 SLP - Number of schedule resets
8403 SLP - Number of vector instructions generated
I do want to highlight that there is a small difference in number of generated vector instructions. This example is hitting the bailout due to maximum window size, and the change in scheduling is slightly perturbing when and how we hit it. This can be seen in the RescheduleOnFail counter change. Given that, I think we can safely ignore.
The downside of this change can be seen in the large test diff. We group all vectorizable instructions together at the bottom of the scheduling region. This means that vector instructions can move quite far from their original point in code. While maybe undesirable, I don't see this as being a major problem as this pass is not intended to be a general scheduling pass.
For context, it's worth noting that the pre-scheduling that SLP does while building the vector tree is exactly the sub-graph scheduling implemented by this patch.
Differential Revision: https://reviews.llvm.org/D118538
SLP is currently assuming that control dependence in these cases is irrelevant. This is only valid if none of the lib-funcs involved can throw or infinite loop in the scalar forms. This appears to be true (or at least we infer the respective attributes) for the libfuncs I spot checked. This change is mostly for shrunking the diff on an upcoming patch.
No need to schedule entry nodes where all instructions are not memory
read/write instructions and their operands are either constants, or
arguments, or phis, or instructions from others blocks, or their users
are phis or from the other blocks.
The resulting vector instructions can be placed at
the beginning of the basic block without scheduling (if operands does
not need to be scheduled) or at the end of the block (if users are
outside of the block).
It may save some compile time and scheduling resources.
Differential Revision: https://reviews.llvm.org/D121121
No need to schedule entry nodes where all instructions are not memory
read/write instructions and their operands are either constants, or
arguments, or phis, or instructions from others blocks, or their users
are phis or from the other blocks.
The resulting vector instructions can be placed at
the beginning of the basic block without scheduling (if operands does
not need to be scheduled) or at the end of the block (if users are
outside of the block).
It may save some compile time and scheduling resources.
Differential Revision: https://reviews.llvm.org/D121121
The costs of vector shifts was 2 as opposed to 1, as the nodes are
marked custom. Fix this like the others and mark the nodes as cheap.
Differential Revision: https://reviews.llvm.org/D120773
Root issue which triggered the revert was fixed in 689bab. No changes in the reapplied patch.
Original commit message follows:
SLP currently schedules all instructions within a scheduling window which stretches from the first instr
uction potentially vectorized to the last. This window can include a very large number of unrelated instruct
ions which are not being considered for vectorization. This change switches the code to only schedule the su
b-graph consisting of the instructions being vectorized and their transitive users.
This has the effect of greatly reducing the amount of work performed in large basic blocks, and thus greatly improves compile time on degenerate examples. To understand the effects, I added some statistics (not planned for upstream contribution). Here's an illustration from my motivating example:
Before this patch:
704357 SLP - Number of calcDeps actions
699021 SLP - Number of schedule calls
5598 SLP - Number of ReSchedule actions
59 SLP - Number of ReScheduleOnFail actions
10084 SLP - Number of schedule resets
8523 SLP - Number of vector instructions generated
After this patch:
102895 SLP - Number of calcDeps actions
161916 SLP - Number of schedule calls
5637 SLP - Number of ReSchedule actions
55 SLP - Number of ReScheduleOnFail actions
10083 SLP - Number of schedule resets
8403 SLP - Number of vector instructions generated
I do want to highlight that there is a small difference in number of generated vector instructions. This example is hitting the bailout due to maximum window size, and the change in scheduling is slightly perturbing when and how we hit it. This can be seen in the RescheduleOnFail counter change. Given that, I think we can safely ignore.
The downside of this change can be seen in the large test diff. We group all vectorizable instructions together at the bottom of the scheduling region. This means that vector instructions can move quite far from their original point in code. While maybe undesirable, I don't see this as being a major problem as this pass is not intended to be a general scheduling pass.
For context, it's worth noting that the pre-scheduling that SLP does while building the vector tree is exactly the sub-graph scheduling implemented by this patch.
Differential Revision: https://reviews.llvm.org/D118538
This reverts commit 0539a26d91a1b7c74022fa9cf33bd7faca87544d.
Causes a miscompile, see comments on D118538.
Required updating bottom-to-top-reorder.ll.
This cap was first added in 848c1aa45 (back in 2015). Per the original commit message, the purpose was to avoid a compile time explosion in long basic blocks. The algorithmic problem in scheduling has now been fixed in 0539a26d.
In the meantime, the code has rotten fairly badly. Some intermediate refactoring caused the size to only be incremented if *both* iterators advance in the window search. This causes the size to be badly undercounted when near one end of a basic block. We no longer have any test which exercises the logic in an intentional way; there's one test which differs with this change, but the changes appear fairly orthoganol to the purpose of the test file.
Unfortunately, we no longer have the original motivating example, so it's possible that it also hits some other issue. I tested locally with a large example, but even at it's worst, that one doesn't demonstrate anything too extreme even without the algorithmic fix. It's clearly faster with, but only by ~20% which doesn't seem in line with the original commit message. If regressions with this patch are seen, please file a bug and I'll try to fix any other algorithmic problems which fall out.
SLP currently schedules all instructions within a scheduling window which stretches from the first instruction potentially vectorized to the last. This window can include a very large number of unrelated instructions which are not being considered for vectorization. This change switches the code to only schedule the sub-graph consisting of the instructions being vectorized and their transitive users.
This has the effect of greatly reducing the amount of work performed in large basic blocks, and thus greatly improves compile time on degenerate examples. To understand the effects, I added some statistics (not planned for upstream contribution). Here's an illustration from my motivating example:
Before this patch:
704357 SLP - Number of calcDeps actions
699021 SLP - Number of schedule calls
5598 SLP - Number of ReSchedule actions
59 SLP - Number of ReScheduleOnFail actions
10084 SLP - Number of schedule resets
8523 SLP - Number of vector instructions generated
After this patch:
102895 SLP - Number of calcDeps actions
161916 SLP - Number of schedule calls
5637 SLP - Number of ReSchedule actions
55 SLP - Number of ReScheduleOnFail actions
10083 SLP - Number of schedule resets
8403 SLP - Number of vector instructions generated
I do want to highlight that there is a small difference in number of generated vector instructions. This example is hitting the bailout due to maximum window size, and the change in scheduling is slightly perturbing when and how we hit it. This can be seen in the RescheduleOnFail counter change. Given that, I think we can safely ignore.
The downside of this change can be seen in the large test diff. We group all vectorizable instructions together at the bottom of the scheduling region. This means that vector instructions can move quite far from their original point in code. While maybe undesirable, I don't see this as being a major problem as this pass is not intended to be a general scheduling pass.
For context, it's worth noting that the pre-scheduling that SLP does while building the vector tree is exactly the sub-graph scheduling implemented by this patch.
Differential Revision: https://reviews.llvm.org/D118538
Compiler adds the estimation for the external uses during operands
reordering analysis, which makes it tend to prefer duplicates in the
lanes rather than diamond/shuffled match in the graph. It changes the sizes of
the vector operands and may prevent some vectorization. We don't need
this kind of estimation for the analysis phase, because we just need to
choose the most compatible instruction and it does not matter if it has
external user or used in the non-matching lane. Instead, we count the number
of unique instruction in the lane and see if the reassociation changes
the number of unique scalars to be power of 2 or not. If we have power
of 2 unique scalars in the lane, it is considered more profitable rather
than having non-power-of-2 number of unique scalars.
Metric: SLP.NumVectorInstructions
test-suite :: MultiSource/Benchmarks/FreeBench/distray/distray.test 70.00 86.00 22.9%
test-suite :: External/SPEC/CFP2017rate/544.nab_r/544.nab_r.test 346.00 353.00 2.0%
test-suite :: External/SPEC/CFP2017speed/644.nab_s/644.nab_s.test 346.00 353.00 2.0%
test-suite :: MultiSource/Benchmarks/mediabench/gsm/toast/toast.test 235.00 239.00 1.7%
test-suite :: MultiSource/Benchmarks/MiBench/telecomm-gsm/telecomm-gsm.test 235.00 239.00 1.7%
test-suite :: External/SPEC/CFP2017rate/526.blender_r/526.blender_r.test 8723.00 8834.00 1.3%
test-suite :: MultiSource/Applications/JM/ldecod/ldecod.test 1051.00 1064.00 1.2%
test-suite :: External/SPEC/CINT2017speed/625.x264_s/625.x264_s.test 1628.00 1646.00 1.1%
test-suite :: External/SPEC/CINT2017rate/525.x264_r/525.x264_r.test 1628.00 1646.00 1.1%
test-suite :: External/SPEC/CFP2017rate/510.parest_r/510.parest_r.test 9100.00 9184.00 0.9%
test-suite :: External/SPEC/CFP2017rate/538.imagick_r/538.imagick_r.test 3565.00 3577.00 0.3%
test-suite :: External/SPEC/CFP2017speed/638.imagick_s/638.imagick_s.test 3565.00 3577.00 0.3%
test-suite :: External/SPEC/CFP2017rate/511.povray_r/511.povray_r.test 4235.00 4245.00 0.2%
test-suite :: MultiSource/Benchmarks/tramp3d-v4/tramp3d-v4.test 1996.00 1998.00 0.1%
test-suite :: MultiSource/Applications/JM/lencod/lencod.test 1671.00 1672.00 0.1%
test-suite :: MultiSource/Benchmarks/Prolangs-C/TimberWolfMC/timberwolfmc.test 783.00 782.00 -0.1%
test-suite :: SingleSource/Benchmarks/Misc/oourafft.test 69.00 68.00 -1.4%
test-suite :: External/SPEC/CINT2017speed/641.leela_s/641.leela_s.test 207.00 192.00 -7.2%
test-suite :: External/SPEC/CINT2017rate/541.leela_r/541.leela_r.test 207.00 192.00 -7.2%
test-suite :: External/SPEC/CINT2017rate/531.deepsjeng_r/531.deepsjeng_r.test 89.00 80.00 -10.1%
test-suite :: External/SPEC/CINT2017speed/631.deepsjeng_s/631.deepsjeng_s.test 89.00 80.00 -10.1%
test-suite :: MultiSource/Benchmarks/mediabench/jpeg/jpeg-6a/cjpeg.test 260.00 215.00 -17.3%
test-suite :: MultiSource/Benchmarks/MiBench/consumer-jpeg/consumer-jpeg.test 256.00 211.00 -17.6%
MultiSource/Benchmarks/Prolangs-C/TimberWolfMC - pretty the same.
SingleSource/Benchmarks/Misc/oourafft.test - 2 <2 x > loads replaced by
one <4 x> load.
External/SPEC/CINT2017speed/641.leela_s - function gets vectorized and
not inlined anymore.
External/SPEC/CINT2017rate/541.leela_r - same
xternal/SPEC/CINT2017rate/531.deepsjeng_r - changed the order in
multi-block tree, the result is pretty the same.
External/SPEC/CINT2017speed/631.deepsjeng_s - same.
MultiSource/Benchmarks/mediabench/jpeg/jpeg-6a - the result is the same
as before.
MultiSource/Benchmarks/MiBench/consumer-jpeg - same.
Differential Revision: https://reviews.llvm.org/D116688
No need to include the order of the scalars beeing used as part of the
alternate vectorization into account when trying to reorder the whole
graph. Such elements better to reorder in the following phase because
the subtree still ends up in shuffle.
Part of D116688, fixes the regression in D116690.
Differential Revision: https://reviews.llvm.org/D116740
There is a bug in the reordering analysis stage. If the element with the
given hash is not added to the map but has the same number of APOs and
instructions with same parent, but different instruction opcode, it will
be initalized with default values and then the counter is increased by
1. But the lane is not updated and default to 0 instead of the actual
`Lane` value. It leads to the fact that the analysis is useless in
many cases and default to lane 0 instead of actual lane with the
minimum amount of APO operands.
Differential Revision: https://reviews.llvm.org/D116690
Changes the preliminary multinode analysis:
1. Introduced scores for reversed loads/extractelements.
2. Improved shallow score calculation.
3. Lowered the cost of external uses (no need to consider it several times, just ones).
4. The initial lane for analysis is the one with the minimal possible
reorderings.
These changes in general shall reduce compile time and improve the
reordering in many cases.
Part of D57059.
Differential Revision: https://reviews.llvm.org/D101109
The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order.
I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions.
We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause.
Differential Revision: https://reviews.llvm.org/D115387
Compiler has an analysis for perfect diamond matching but it does not
support nodes with main/alternate opcodes. The problem is that the
scalars themselves are different and might not match directly with other
nodes, but operands and main/alternate opcodes might match and compiler
might reuse some previously emitted vector instructions. Need to include
this analysis in the cost model and actual vector instructions emission
process.
Differential Revision: https://reviews.llvm.org/D114101
Compiler has an analysis for perfect diamond matching but it does not
support nodes with main/alternate opcodes. The problem is that the
scalars themselves are different and might not match directly with other
nodes, but operands and main/alternate opcodes might match and compiler
might reuse some previously emitted vector instructions. Need to include
this analysis in the cost model and actual vector instructions emission
process.
Differential Revision: https://reviews.llvm.org/D114101
No need to count the final shuffle cost for the constants, gathering of
the constants is just a constant vector + extra inserts, if required.
Differential Revision: https://reviews.llvm.org/D113770
Need to fix ther cost estimation for split loads, since we look at the
subregs already, no need to permute them, need just to estimate
subregister insert, if it is smaller than the real register. Also, using
split loads, it might be profitable already to vectorize smaller trees
with gathering of the loads.
Differential Revision: https://reviews.llvm.org/D107188
Gathered loads/extractelements/extractvalue instructions should be
checked if they can represent a vector reordering node too and their
order should ve taken into account for better graph reordering analysis/
Also, if the gather node has reused scalars, they must be reordered
instead of the scalars themselves.
Differential Revision: https://reviews.llvm.org/D112454
Gathered loads/extractelements/extractvalue instructions should be
checked if they can represent a vector reordering node too and their
order should ve taken into account for better graph reordering analysis/
Also, if the gather node has reused scalars, they must be reordered
instead of the scalars themselves.
Differential Revision: https://reviews.llvm.org/D112454
Gathered loads/extractelements/extractvalue instructions should be
checked if they can represent a vector reordering node too and their
order should ve taken into account for better graph reordering analysis/
Also, if the gather node has reused scalars, they must be reordered
instead of the scalars themselves.
Differential Revision: https://reviews.llvm.org/D112454
Need to change the order of the reduction/binops args pair vectorization
attempts. Need to try to find the reduction at first and postpone
vectorization of binops args. This may help to find more reduction
patterns and vectorize them.
Part of D111574.
Differential Revision: https://reviews.llvm.org/D112224
D104809 changed `buildTree_rec` to check for extract element instructions
with scalable types. However, if the extract is extended or truncated,
these changes do not apply and we assert later on in isShuffle(), which
attempts to cast the type of the extract to FixedVectorType.
Reviewed By: ABataev
Differential Revision: https://reviews.llvm.org/D110640
This patch is for fixing potential shufflevector-related bugs like D93818.
As D93818, this patch change shufflevector's default placeholder to poison.
To reduce risk, it was divided into several patches, and this patch is for InstCombineCompares and InstructionCombining.
Reviewed By: spatel
Differential Revision: https://reviews.llvm.org/D110227
Reworked reordering algorithm. Originally, the compiler just tried to
detect the most common order in the reordarable nodes (loads, stores,
extractelements,extractvalues) and then fully rebuilding the graph in
the best order. This was not effecient, since it required an extra
memory and time for building/rebuilding tree, double the use of the
scheduling budget, which could lead to missing vectorization due to
exausted scheduling resources.
Patch provide 2-way approach for graph reodering problem. At first, all
reordering is done in-place, it doe not required tree
deleting/rebuilding, it just rotates the scalars/orders/reuses masks in
the graph node.
The first step (top-to bottom) rotates the whole graph, similarly to the previous
implementation. Compiler counts the number of the most used orders of
the graph nodes with the same vectorization factor and then rotates the
subgraph with the given vectorization factor to the most used order, if
it is not empty. Then repeats the same procedure for the subgraphs with
the smaller vectorization factor. We can do this because we still need
to reshuffle smaller subgraph when buildiong operands for the graph
nodes with lasrger vectorization factor, we can rotate just subgraph,
not the whole graph.
The second step (bottom-to-top) scans through the leaves and tries to
detect the users of the leaves which can be reordered. If the leaves can
be reorder in the best fashion, they are reordered and their user too.
It allows to remove double shuffles to the same ordering of the operands in
many cases and just reorder the user operations instead. Plus, it moves
the final shuffles closer to the top of the graph and in many cases
allows to remove extra shuffle because the same procedure is repeated
again and we can again merge some reordering masks and reorder user nodes
instead of the operands.
Also, patch improves cost model for gathering of loads, which improves
x264 benchmark in some cases.
Gives about +2% on AVX512 + LTO (more expected for AVX/AVX2) for {625,525}x264,
+3% for 508.namd, improves most of other benchmarks.
The compile and link time are almost the same, though in some cases it
should be better (we're not doing an extra instruction scheduling
anymore) + we may vectorize more code for the large basic blocks again
because of saving scheduling budget.
Differential Revision: https://reviews.llvm.org/D105020
This reverts commit 84cbd71c95923f9912512f3051c6ab548a99e016.
This commit breaks one of the internal tests. As agreed with Alexey I
will provide the reproducer later.
Reworked reordering algorithm. Originally, the compiler just tried to
detect the most common order in the reordarable nodes (loads, stores,
extractelements,extractvalues) and then fully rebuilding the graph in
the best order. This was not effecient, since it required an extra
memory and time for building/rebuilding tree, double the use of the
scheduling budget, which could lead to missing vectorization due to
exausted scheduling resources.
Patch provide 2-way approach for graph reodering problem. At first, all
reordering is done in-place, it doe not required tree
deleting/rebuilding, it just rotates the scalars/orders/reuses masks in
the graph node.
The first step (top-to bottom) rotates the whole graph, similarly to the previous
implementation. Compiler counts the number of the most used orders of
the graph nodes with the same vectorization factor and then rotates the
subgraph with the given vectorization factor to the most used order, if
it is not empty. Then repeats the same procedure for the subgraphs with
the smaller vectorization factor. We can do this because we still need
to reshuffle smaller subgraph when buildiong operands for the graph
nodes with lasrger vectorization factor, we can rotate just subgraph,
not the whole graph.
The second step (bottom-to-top) scans through the leaves and tries to
detect the users of the leaves which can be reordered. If the leaves can
be reorder in the best fashion, they are reordered and their user too.
It allows to remove double shuffles to the same ordering of the operands in
many cases and just reorder the user operations instead. Plus, it moves
the final shuffles closer to the top of the graph and in many cases
allows to remove extra shuffle because the same procedure is repeated
again and we can again merge some reordering masks and reorder user nodes
instead of the operands.
Also, patch improves cost model for gathering of loads, which improves
x264 benchmark in some cases.
Gives about +2% on AVX512 + LTO (more expected for AVX/AVX2) for {625,525}x264,
+3% for 508.namd, improves most of other benchmarks.
The compile and link time are almost the same, though in some cases it
should be better (we're not doing an extra instruction scheduling
anymore) + we may vectorize more code for the large basic blocks again
because of saving scheduling budget.
Differential Revision: https://reviews.llvm.org/D105020
Reworked reordering algorithm. Originally, the compiler just tried to
detect the most common order in the reordarable nodes (loads, stores,
extractelements,extractvalues) and then fully rebuilding the graph in
the best order. This was not effecient, since it required an extra
memory and time for building/rebuilding tree, double the use of the
scheduling budget, which could lead to missing vectorization due to
exausted scheduling resources.
Patch provide 2-way approach for graph reodering problem. At first, all
reordering is done in-place, it doe not required tree
deleting/rebuilding, it just rotates the scalars/orders/reuses masks in
the graph node.
The first step (top-to bottom) rotates the whole graph, similarly to the previous
implementation. Compiler counts the number of the most used orders of
the graph nodes with the same vectorization factor and then rotates the
subgraph with the given vectorization factor to the most used order, if
it is not empty. Then repeats the same procedure for the subgraphs with
the smaller vectorization factor. We can do this because we still need
to reshuffle smaller subgraph when buildiong operands for the graph
nodes with lasrger vectorization factor, we can rotate just subgraph,
not the whole graph.
The second step (bottom-to-top) scans through the leaves and tries to
detect the users of the leaves which can be reordered. If the leaves can
be reorder in the best fashion, they are reordered and their user too.
It allows to remove double shuffles to the same ordering of the operands in
many cases and just reorder the user operations instead. Plus, it moves
the final shuffles closer to the top of the graph and in many cases
allows to remove extra shuffle because the same procedure is repeated
again and we can again merge some reordering masks and reorder user nodes
instead of the operands.
Also, patch improves cost model for gathering of loads, which improves
x264 benchmark in some cases.
Gives about +2% on AVX512 + LTO (more expected for AVX/AVX2) for {625,525}x264,
+3% for 508.namd, improves most of other benchmarks.
The compile and link time are almost the same, though in some cases it
should be better (we're not doing an extra instruction scheduling
anymore) + we may vectorize more code for the large basic blocks again
because of saving scheduling budget.
Differential Revision: https://reviews.llvm.org/D105020
If the vectorized insertelements instructions form indentity subvector
(the subvector at the beginning of the long vector), it is just enough
to extend the vector itself, no need to generate inserting subvector
shuffle.
Differential Revision: https://reviews.llvm.org/D107494
If the vectorized insertelements instructions form indentity subvector
(the subvector at the beginning of the long vector), it is just enough
to extend the vector itself, no need to generate inserting subvector
shuffle.
Differential Revision: https://reviews.llvm.org/D107344
This reverts commit e408d1dfab42b27d0aa51b221e50fa6390fb5ed1 and
2 other (4b25c113210e579a5346ca0abc0717ab1ce5d9df and
c2deb2afafee991c06cc96dc5beecb6de448b9fc) related to fix the problem with the
reordering shuffles.
Replace insertelement instructions for splats with just single
insertelement + broadcast shuffle. Also, try to merge these instructions
if they come from the same/shuffled gather node.
Differential Revision: https://reviews.llvm.org/D107104
For the nodes with reused scalars the user may be not only of the size
of the final shuffle but also of the size of the scalars themselves,
need to check for this. It is safe to just modify the check here, since
the order of the scalars themselves is preserved, only indeces of the
reused scalars are changed. So, the users with the same size as the
number of scalars in the node, will not be affected, they still will get
the operands in the required order.
Reported by @mstorsjo in D105020.
Differential Revision: https://reviews.llvm.org/D107080
If the instruction was previously deleted, it should not be treated as
an external user. This fixes cost estimation and removes dead
extractelement instructions.
Differential Revision: https://reviews.llvm.org/D107106
