[AMDGPU] Occupancy w.r.t. workgroup size range is also a range (#123748)

Occupancy (i.e., the number of waves per EU) depends, in addition to
register usage, on per-workgroup LDS usage as well as on

[AMDGPU] Occupancy w.r.t. workgroup size range is also a range (#123748)

Occupancy (i.e., the number of waves per EU) depends, in addition to
register usage, on per-workgroup LDS usage as well as on the range of
possible workgroup sizes. Mirroring the latter, occupancy should
therefore be expressed as a range since different group sizes generally
yield different achievable occupancies.

`getOccupancyWithLocalMemSize` currently returns a scalar occupancy
based on the maximum workgroup size and LDS usage. With respect to the
workgroup size range, this scalar can be the minimum, the maximum, or
neither of the two of the range of achievable occupancies. This commit
fixes the function by making it compute and return the range of
achievable occupancies w.r.t. workgroup size and LDS usage; it also
renames it to `getOccupancyWithWorkGroupSizes` since it is the range of
workgroup sizes that produces the range of achievable occupancies.

Computing the achievable occupancy range is surprisingly involved.
Minimum/maximum workgroup sizes do not necessarily yield maximum/minimum
occupancies i.e., sometimes workgroup sizes inside the range yield the
occupancy bounds. The implementation finds these sizes in constant time;
heavy documentation explains the rationale behind the sometimes
relatively obscure calculations.

As a justifying example, consider a target with 10 waves / EU, 4 EUs/CU,
64-wide waves. Also consider a function with no LDS usage and a flat
workgroup size range of [513,1024].

- A group of 513 items requires 9 waves per group. Only 4 groups made up
of 9 waves each can fit fully on a CU at any given time, for a total of
36 waves on the CU, or 9 per EU. However, filling as much as possible
the remaining 40-36=4 wave slots without decreasing the number of groups
reveals that a larger group of 640 items yields 40 waves on the CU, or
10 per EU.
- Similarly, a group of 1024 items requires 16 waves per group. Only 2
groups made up of 16 waves each can fit fully on a CU ay any given time,
for a total of 32 waves on the CU, or 8 per EU. However, removing as
many waves as possible from the groups without being able to fit another
equal-sized group on the CU reveals that a smaller group of 896 items
yields 28 waves on the CU, or 7 per EU.

Therefore the achievable occupancy range for this function is not [8,9]
as the group size bounds directly yield, but [7,10].

Naturally this change causes a lot of test churn as instruction
scheduling is driven by achievable occupancy estimates. In most unit
tests the flat workgroup size range is the default [1,1024] which,
ignoring potential LDS limitations, would previously produce a scalar
occupancy of 8 (derived from 1024) on a lot of targets, whereas we now
consider the maximum occupancy to be 10 in such cases. Most tests are
updated automatically and checked manually for sanity. I also manually
changed some non-automatically generated assertions when necessary.

Fixes #118220.

show more ...

Reapply "[AMDGPU] Still set up the two SGPRs for queue ptr even it is COV5 (#112403)"

This reverts commit ca33649abe5fad93c57afef54e43ed9b3249cd86.

Revert "[AMDGPU] Still set up the two SGPRs for queue ptr even it is COV5 (#112403)"

This reverts commit e215a1e27d84adad2635a52393621eb4fa439dc9 as it broke both
hip and openmp buildbots.

[AMDGPU] Still set up the two SGPRs for queue ptr even it is COV5 (#112403)

Reapply "AMDGPU: Move attributor into optimization pipeline (#83131)" and follow up commit "clang/AMDGPU: Defeat attribute optimization in attribute test" (#98851)

This reverts commit adaff46d087799

Reapply "AMDGPU: Move attributor into optimization pipeline (#83131)" and follow up commit "clang/AMDGPU: Defeat attribute optimization in attribute test" (#98851)

This reverts commit adaff46d087799072438dd744b038e6fd50a2d78.

Drop the -O3 checks from default-attributes.hip. I don't know why they
are different on some bots but reverting this is far too disruptive.

show more ...

Revert "AMDGPU: Move attributor into optimization pipeline (#83131)" and follow up commit "clang/AMDGPU: Defeat attribute optimization in attribute test" (#98851)

This reverts commits 677cc15e0ff2e0

Revert "AMDGPU: Move attributor into optimization pipeline (#83131)" and follow up commit "clang/AMDGPU: Defeat attribute optimization in attribute test" (#98851)

This reverts commits 677cc15e0ff2e0e6aa30538eb187990a6a8f53c0 and
78bc1b64a6dc3fb6191355a5e1b502be8b3668e7.

The test CodeGenHIP/default-attributes.hip is failing on multiple bots
even after the attempted fix including the following:
- https://lab.llvm.org/buildbot/#/builders/3/builds/1473
- https://lab.llvm.org/buildbot/#/builders/65/builds/1380
- https://lab.llvm.org/buildbot/#/builders/161/builds/595
- https://lab.llvm.org/buildbot/#/builders/154/builds/1372
- https://lab.llvm.org/buildbot/#/builders/133/builds/1547
- https://lab.llvm.org/buildbot/#/builders/81/builds/755
- https://lab.llvm.org/buildbot/#/builders/40/builds/570
- https://lab.llvm.org/buildbot/#/builders/13/builds/748
- https://lab.llvm.org/buildbot/#/builders/12/builds/1845
- https://lab.llvm.org/buildbot/#/builders/11/builds/1695
- https://lab.llvm.org/buildbot/#/builders/190/builds/1829
- https://lab.llvm.org/buildbot/#/builders/193/builds/962
- https://lab.llvm.org/buildbot/#/builders/23/builds/991
- https://lab.llvm.org/buildbot/#/builders/144/builds/2256
- https://lab.llvm.org/buildbot/#/builders/46/builds/1614

These bots have been broken for a day, so reverting to get everything
back to green.

show more ...

AMDGPU: Move attributor into optimization pipeline (#83131)

Removing it from the codegen pipeline induces a lot of test churn
because llc is no longer optimizing out implicit arguments to kernels.

AMDGPU: Move attributor into optimization pipeline (#83131)

Removing it from the codegen pipeline induces a lot of test churn
because llc is no longer optimizing out implicit arguments to kernels.

Mostly mechanical, but there are some creative test updates. I preferred
to take the changes as-is in tests where the ABI isn't relevant. In
cases where it's more relevant, or the optimize out logic was too
ingrained in the test, I pre-run the optimization. Some cases manually
add attributes to disable inputs.

show more ...

[AMDGPU,test] Change llc -march= to -mtriple= (#75982)

Similar to 806761a7629df268c8aed49657aeccffa6bca449.

For IR files without a target triple, -mtriple= specifies the full
target triple while

[AMDGPU,test] Change llc -march= to -mtriple= (#75982)

Similar to 806761a7629df268c8aed49657aeccffa6bca449.

For IR files without a target triple, -mtriple= specifies the full
target triple while -march= merely sets the architecture part of the
default target triple, leaving a target triple which may not make sense,
e.g. amdgpu-apple-darwin.

Therefore, -march= is error-prone and not recommended for tests without
a target triple. The issue has been benign as we recognize
$unknown-apple-darwin as ELF instead of rejecting it outrightly.

This patch changes AMDGPU tests to not rely on the default
OS/environment components. Tests that need fixes are not changed:

```
LLVM :: CodeGen/AMDGPU/fabs.f64.ll
LLVM :: CodeGen/AMDGPU/fabs.ll
LLVM :: CodeGen/AMDGPU/floor.ll
LLVM :: CodeGen/AMDGPU/fneg-fabs.f64.ll
LLVM :: CodeGen/AMDGPU/fneg-fabs.ll
LLVM :: CodeGen/AMDGPU/r600-infinite-loop-bug-while-reorganizing-vector.ll
LLVM :: CodeGen/AMDGPU/schedule-if-2.ll
```

show more ...

Revert "[CodeGen] Really renumber slot indexes before register allocation (#67038)"

This reverts commit 2501ae58e3bb9a70d279a56d7b3a0ed70a8a852c.

Reverted due to various buildbot failures.

[CodeGen] Really renumber slot indexes before register allocation (#67038)

PR #66334 tried to renumber slot indexes before register allocation, but
the numbering was still affected by list entries

[CodeGen] Really renumber slot indexes before register allocation (#67038)

PR #66334 tried to renumber slot indexes before register allocation, but
the numbering was still affected by list entries for instructions which
had been erased. Fix this to make the register allocator's live range
length heuristics even less dependent on the history of how instructions
have been added to and removed from SlotIndexes's maps.

show more ...

[AMDGPU] Introduce SIInstrWorklist to process instructions in moveToVALU

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D147168

AMDGPU: Clean up LDS-related occupancy calculations

Occupancy is expressed as waves per SIMD. This means that we need to
take into account the number of SIMDs per "CU" or, to be more precise,
the nu

AMDGPU: Clean up LDS-related occupancy calculations

Occupancy is expressed as waves per SIMD. This means that we need to
take into account the number of SIMDs per "CU" or, to be more precise,
the number of SIMDs over which a workgroup may be distributed.

getOccupancyWithLocalMemSize was wrong because it didn't take SIMDs
into account at all.

At the same time, we need to take into account that WGP mode offers
access to a larger total amount of LDS, since this can affect how
non-power-of-two LDS allocations are rounded. To make this work
consistently, we distinguish between (available) local memory size and
addressable local memory size (which is always limited by 64kB on
gfx10+, even with WGP mode).

This change results in a massive amount of test churn. A lot of it is
caused by the fact that the default work group size is 1024, which means
that (due to rounding effects) the default occupancy on older hardware
is 8 instead of 10, which affects scheduling via register pressure
estimates. I've adjusted most tests by just running the UTC tools, but
in some cases I manually changed the work group size to 32 or 64 to make
sure that work group size chunkiness has no effect.

Differential Revision: https://reviews.llvm.org/D139468

show more ...

AMDGPU: Convert some bit operation tests to opaque pointers

[AMDGPU] Lowering VGPR to SGPR copies to v_readfirstlane_b32 if profitable.

Since the divergence-driven instruction selection has been enabled for AMDGPU,
all the uniform instructions are expected

[AMDGPU] Lowering VGPR to SGPR copies to v_readfirstlane_b32 if profitable.

Since the divergence-driven instruction selection has been enabled for AMDGPU,
all the uniform instructions are expected to be selected to SALU form, except those not having one.
VGPR to SGPR copies appear in MIR to connect values producers and consumers. This change implements an algorithm
that evolves a reasonable tradeoff between the profit achieved from keeping the uniform instructions in SALU form
and overhead introduced by the data transfer between the VGPRs and SGPRs.

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D128252

show more ...

[LSV] Vectorize loads of vectors by turning it into a larger vector

Use shufflevector to do the subvector extracts. This allows a lot more
load merging on AMDGPU and also on NVPTX when <2 x half> is

[LSV] Vectorize loads of vectors by turning it into a larger vector

Use shufflevector to do the subvector extracts. This allows a lot more
load merging on AMDGPU and also on NVPTX when <2 x half> is involved.

Differential Revision: https://reviews.llvm.org/D117219

show more ...

[AMDGPU] Set most sched model resource's BufferSize to one

Using a BufferSize of one for memory ProcResources will result in better
ILP since it more accurately models the dependencies between memor

[AMDGPU] Set most sched model resource's BufferSize to one

Using a BufferSize of one for memory ProcResources will result in better
ILP since it more accurately models the dependencies between memory ops
and their consumers on an in-order processor. After this change, the
scheduler will treat the data edges from loads as blocking so that
stalls are guaranteed when waiting for data to be retreaved from memory.
Since we don't actually track waitcnt here, this should do a better job
at modeling their behavior.

Practically, this means that the scheduler will trigger the 'STALL'
heuristic more often.

This type of change needs to be evaluated experimentally. Preliminary
results are positive.

Fixes: SWDEV-282962

Reviewed By: rampitec

Differential Revision: https://reviews.llvm.org/D114777

show more ...

RegAllocGreedy: Account for reserved registers in num regs heuristic

This simple heuristic uses the estimated live range length combined
with the number of registers in the class to switch which heu

RegAllocGreedy: Account for reserved registers in num regs heuristic

This simple heuristic uses the estimated live range length combined
with the number of registers in the class to switch which heuristic to
use. This was taking the raw number of registers in the class, even
though not all of them may be available. AMDGPU heavily relies on
dynamically reserved numbers of registers based on user attributes to
satisfy occupancy constraints, so the raw number is highly misleading.

There are still a few problems here. In the original testcase that
made me notice this, the live range size is incorrect after the
scheduler rearranges instructions, since the instructions don't have
the original InstrDist offsets. Additionally, I think it would be more
appropriate to use the number of disjointly allocatable registers in
the class. For the AMDGPU register tuples, there are a large number of
registers in each tuple class, but only a small fraction can actually
be allocated at the same time since they all overlap with each
other. It seems we do not have a query that corresponds to the number
of independently allocatable registers. Relatedly, I'm still debugging
some allocation failures where overlapping tuples seem to not be
handled correctly.

The test changes are mostly noise. There are a handful of x86 tests
that look like regressions with an additional spill, and a handful
that now avoid a spill. The worst looking regression is likely
test/Thumb2/mve-vld4.ll which introduces a few additional
spills. test/CodeGen/AMDGPU/soft-clause-exceeds-register-budget.ll
shows a massive improvement by completely eliminating a large number
of spills inside a loop.

show more ...

[AMDGPU] Switch PostRA sched to MachineSched

Use GCNHazardRecognizer in postra sched.
Updated tests for the new schedules.

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D1095

[AMDGPU] Switch PostRA sched to MachineSched

Use GCNHazardRecognizer in postra sched.
Updated tests for the new schedules.

Reviewed By: arsenm

Differential Revision: https://reviews.llvm.org/D109536

Change-Id: Ia86ba2ae168f12fb34b4d8efdab491f84d936cde

show more ...

[DAG] Add a generic expansion for SHIFT_PARTS opcodes using funnel shifts

Based off a discussion on D89281 - where the AARCH64 implementations were being replaced to use funnel shifts.

Any target t

[DAG] Add a generic expansion for SHIFT_PARTS opcodes using funnel shifts

Based off a discussion on D89281 - where the AARCH64 implementations were being replaced to use funnel shifts.

Any target that has efficient funnel shift lowering can handle the shift parts expansion using the same expansion, avoiding a lot of duplication.

I've generalized the X86 implementation and moved it to TargetLowering - so far I've found that AARCH64 and AMDGPU benefit, but many other targets (ARM, PowerPC + RISCV in particular) could easily use this with a few minor improvements to their funnel shift lowering (or the folding of their target ops that funnel shifts lower to).

NOTE: I'm trying to avoid adding full SHIFT_PARTS legalizer handling as I think it might actually be possible to remove these opcodes in the medium-term and use funnel shift / libcall expansion directly.

Differential Revision: https://reviews.llvm.org/D101987

show more ...

[AMDGPU] Regenerate shift tests. NFCI.

AMDGPU: Move R600 test compatability hack

Instead of handling the r600 intrinsics on amdgcn, handle the amdgcn
intrinsics on r600.

[FileCheck] Add -allow-deprecated-dag-overlap to failing llvm tests

See https://reviews.llvm.org/D47106 for details.

Reviewed By: probinson

Differential Revision: https://reviews.llvm.org/D47171

[FileCheck] Add -allow-deprecated-dag-overlap to failing llvm tests

See https://reviews.llvm.org/D47106 for details.

Reviewed By: probinson

Differential Revision: https://reviews.llvm.org/D47171

This commit drops that patch's changes to:

llvm/test/CodeGen/NVPTX/f16x2-instructions.ll
llvm/test/CodeGen/NVPTX/param-load-store.ll

For some reason, the dos line endings there prevent me from commiting
via the monorepo. A follow-up commit (not via the monorepo) will
finish the patch.

llvm-svn: 336843

show more ...

AMDGPU: Add pass to lower kernel arguments to loads

This replaces most argument uses with loads, but for
now not all.

The code in SelectionDAG for calling convention lowering
is actively harmful fo

AMDGPU: Add pass to lower kernel arguments to loads

This replaces most argument uses with loads, but for
now not all.

The code in SelectionDAG for calling convention lowering
is actively harmful for amdgpu_kernel. It attempts to
split the argument types into register legal types, which
results in low quality code for arbitary types. Since
all kernel arguments are passed in memory, we just want the
raw types.

I've tried a couple of methods of mitigating this in SelectionDAG,
but it's easier to just bypass this problem alltogether. It's
possible to hack around the problem in the initial lowering,
but the real problem is the DAG then expects to be able to use
CopyToReg/CopyFromReg for uses of the arguments outside the block.

Exposing the argument loads in the IR also has the advantage
that the LoadStoreVectorizer can merge them.

I'm not sure the best approach to dealing with the IR
argument list is. The patch as-is just leaves the IR arguments
in place, so all the existing code will still compute the same
kernarg size and pointlessly lowers the arguments.

Arguably the frontend should emit kernels with an empty argument
list in the first place. Alternatively a dummy array could be
inserted as a single argument just to reserve space.

This does have some disadvantages. Local pointer kernel arguments can
no longer have AssertZext placed on them as the equivalent !range
metadata is not valid on pointer typed loads. This is mostly bad
for SI which needs to know about the known bits in order to use the
DS instruction offset, so in this case this is not done.

More importantly, this skips noalias arguments since this pass
does not yet convert this to the equivalent !alias.scope and !noalias
metadata. Producing this metadata correctly seems to be tricky,
although this logically is the same as inlining into a function which
doesn't exist. Additionally, exposing these loads to the vectorizer
may result in degraded aliasing information if a pointer load is
merged with another argument load.

I'm also not entirely sure this is preserving the current clover
ABI, although I would greatly prefer if it would stop widening
arguments and match the HSA ABI. As-is I think it is extending
< 4-byte arguments to 4-bytes but doesn't align them to 4-bytes.

llvm-svn: 335650

show more ...

[AMDGPU] Switch scalarize global loads ON by default
Differential revision: https://reviews.llvm.org/D34407

llvm-svn: 307097

Revert r307026, "[AMDGPU] Switch scalarize global loads ON by default"

It broke a testcase.

Failing Tests (1):
LLVM :: CodeGen/AMDGPU/alignbit-pat.ll

llvm-svn: 307054