This was just emitting loads with the ABI alignment
for the raw type. The true alignment is often better,
especially when an illegal vector type was scalarized.
The better alignment allows using a scalar load
more often.
llvm-svn: 333558
In terms of waitcnt insertion/if necessary, the waitcnt pass forces convergence
for a loop. Previously, that kicked if greater than 2 passes over a loop, which
doesn't account for loop with many bottom blocks. So, increase the threshold to
(n+1), where n is the number of bottom blocks. This gives the pass an
opportunity to consider the contribution of each bottom block, to the overall
loop, before the forced convergence potentially kicks in.
Differential Revision: https://reviews.llvm.org/D47488
llvm-svn: 333556
Support for Clang lowering of fused intrinsics. This patch:
1. Removes bindings to clang fma intrinsics.
2. Introduces new LLVM unmasked intrinsics with rounding mode:
int_x86_avx512_vfmadd_pd_512
int_x86_avx512_vfmadd_ps_512
int_x86_avx512_vfmaddsub_pd_512
int_x86_avx512_vfmaddsub_ps_512
supported with a new intrinsic type (INTR_TYPE_3OP_RM).
3. Introduces new x86 fmaddsub/fmsubadd folding.
4. Introduces new tests for code emitted by sequentions introduced in Clang part.
Patch by tkrupa
Reviewers: craig.topper, sroland, spatel, RKSimon
Reviewed By: craig.topper, RKSimon
Differential Revision: https://reviews.llvm.org/D47443
llvm-svn: 333554
It was noticed on D47377 that these tests were being unnecessarily affected by scheduler changes.
This adds vzeroupper at the end of some tests as we lose the 'FeatureFastPartialYMMorZMMWrite' feature from KNL, since Skylake+ don't support this its probably better.
llvm-svn: 333549
Summary: The fX version of floating-point registers only supports
single precision. We need to map the name to dX for doubles and qX
for long doubles if we want getRegForInlineAsmConstraint() to be
able to pick the correct register class.
Reviewers: jyknight, venkatra
Reviewed By: jyknight
Subscribers: eraman, fedor.sergeev, jrtc27, llvm-commits
Differential Revision: https://reviews.llvm.org/D47258
llvm-svn: 333512
AFAIK the driver's allocation will actually have to round this
up anyway. It is useful to track the rounded up size, so that
the end of the kernel segment is known to be dereferencable so
a wider s_load_dword can be used for a short argument at the end
of the segment.
llvm-svn: 333456
Summary:
Base and offset are always separated when a GlobalAddress node is lowered
(rL332641) as an optimization to reduce instruction count. However, this
optimization is not profitable if the Global Address ends up being used in only
instruction.
This patch adds peephole optimizations that merge an offset of
an address calculation into the LUI %%hi and ADD %lo of the lowering sequence.
The peephole handles three patterns:
1) ADDI (ADDI (LUI %hi(global)) %lo(global)), offset
--->
ADDI (LUI %hi(global + offset)) %lo(global + offset).
This generates:
lui a0, hi (global + offset)
add a0, a0, lo (global + offset)
Instead of
lui a0, hi (global)
addi a0, hi (global)
addi a0, offset
This pattern is for cases when the offset is small enough to fit in the
immediate filed of ADDI (less than 12 bits).
2) ADD ((ADDI (LUI %hi(global)) %lo(global)), (LUI hi_offset))
--->
offset = hi_offset << 12
ADDI (LUI %hi(global + offset)) %lo(global + offset)
Which generates the ASM:
lui a0, hi(global + offset)
addi a0, lo(global + offset)
Instead of:
lui a0, hi(global)
addi a0, lo(global)
lui a1, (offset)
add a0, a0, a1
This pattern is for cases when the offset doesn't fit in an immediate field
of ADDI but the lower 12 bits are all zeros.
3) ADD ((ADDI (LUI %hi(global)) %lo(global)), (ADDI lo_offset, (LUI hi_offset)))
--->
offset = global + offhi20<<12 + offlo12
ADDI (LUI %hi(global + offset)) %lo(global + offset)
Which generates the ASM:
lui a1, %hi(global + offset)
addi a1, %lo(global + offset)
Instead of:
lui a0, hi(global)
addi a0, lo(global)
lui a1, (offhi20)
addi a1, (offlo12)
add a0, a0, a1
This pattern is for cases when the offset doesn't fit in an immediate field
of ADDI and both the lower 1 bits and high 20 bits are non zero.
Reviewers: asb
Reviewed By: asb
Subscribers: rbar, johnrusso, simoncook, jordy.potman.lists, apazos,
niosHD, kito-cheng, shiva0217, zzheng, edward-jones, mgrang
llvm-svn: 333455
We've had Thumb1 support for ARMISD::SUBE for a while now, so this just
works. Reduces codesize a bit for 64-bit integer comparisons.
Differential Revision: https://reviews.llvm.org/D47387
llvm-svn: 333445
As suggested in https://bugs.llvm.org/show_bug.cgi?id=32384#c1, this change
makes the inlining of `memset()` and `memcpy()` more aggressive when
compiling for speed. The tuning remains the same when optimizing for size.
Patch by: Sebastian Pop <s.pop@samsung.com>
Evandro Menezes <e.menezes@samsung.com>
Differential revision: https://reviews.llvm.org/D45098
llvm-svn: 333429
1. Introduction of mask scalar TableGen patterns.
2. Introduction of new scalar move TableGen patterns
and refactoring of existing ones.
3. Folding of pattern created by introducing scalar
masking in Clang header files.
Patch by tkrupa
Differential Revision: https://reviews.llvm.org/D47012
llvm-svn: 333419
Summary:
Avoid assert/crash during liveness calculation in situations
where the incoming machine function has statically unreachable BBs.
Second attempt at submitting; this version of the change includes
a revised testcase.
Fixes PR37130.
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D47372
llvm-svn: 333416
Instruction selection can insert nodes into the underlying list after the root
node so iterating will thereby miss it. We should NOT assume that, the root node
is the last element in the DAG nodelist.
Patch by: steven.zhang (Qing Shan Zhang)
Differential Revision: https://reviews.llvm.org/D47437
llvm-svn: 333415
Previously, their listed predicates were overridden at the scope level.
Reviewers: atanasyan, abeserminji, smaksimovic
Differential Revision: https://reviews.llvm.org/D46947
llvm-svn: 333405
Summary: We already get this right if the i64 didn't come from a load.
Reviewers: RKSimon
Reviewed By: RKSimon
Subscribers: llvm-commits
Differential Revision: https://reviews.llvm.org/D47439
llvm-svn: 333393
We have unmasked intrinsics now and wrap them with a select. This is a net reduction of 36 intrinsics from before the unmasked intrinsics were added.
llvm-svn: 333388
This will allow us to remove the 3 different flavors of masked intrinsics. I'm leaving the actual intrinsic removal for another patch.
llvm-svn: 333386
These do the same thing with the first and second sources swapped. They previously came from separate intrinsics that specified different masking behavior. But we can cover that with isel patterns and a single node.
This is a step towards reducing the number of intrinsics needed.
A bunch of tests change because we are now biased to choosing VPERMT over VPERMI when there is nothing to signal that commuting is beneficial.
llvm-svn: 333383
Implemente patterns to extract HWord and Byte vector elements and convert to
quad-precision.
Differential Revision: https://reviews.llvm.org/D46774
llvm-svn: 333377
Summary:
For a block with WQM on entry and exit and containing no exact mode
code, but containing some WWM code, the WQM pass forgot to process the
block at all and so did not insert code to enter and leave WWM.
This commit fixes that.
Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D47027
Change-Id: I044792eead1293bed4203fb26ce75f47878afeb6
llvm-svn: 333362
With the removal of the old waitcnt pass, the '-enable-si-insert-waitcnts' option is obsolete. Remove it.
Differential Revision: https://reviews.llvm.org/D47378
llvm-svn: 333303
This is adoption of HSAIL perfhint pass. Two types of hints are produced:
1. Function is memory bound.
2. Kernel can use wave limiter.
Currently these hints are used in the scheduler. If a function is suspected
to be memory bound we allow occupancy to decrease to 4 waves in the course
of scheduling.
Differential Revision: https://reviews.llvm.org/D46992
llvm-svn: 333289
When the shuffle mask selected a subvector of the second input vector,
and aligning of the source was performed, the shuffle mask was updated
incorrectly, resulting in an ICE further in the selection process.
llvm-svn: 333279
As confirmed by llvm-exegesis, there is no scheduler difference between MOVDQA/MOVDQU and VMOVDQA/VMOVDQU xmm reg-reg moves
Another chapter in the never ending crusade to remove useless InstRW overrides from the x86 scheduler models......
llvm-svn: 333271
Previously, this pass would look at the (static) set returned by
getCallPreservedMask() and add those back as preserved in the case when
isSafeForNoCSROpt() returns false.
A problem is that a target may have to save some registers even when NoCSROpt
takes place. For instance, on SystemZ, the return register is needed upon
return from a function.
Furthermore, getCallPreservedMask() only includes the registers that the
target actually wishes to emit save/restore instructions for. This means that
subregs and (fully saved) superregs are missing.
This patch instead takes the (dynamic) set returned by target for the
function from determineCalleeSaves() and then adds sub/super regs to build
the set to be used when building the RegMask for the function.
Review: Quentin Colombet, Ulrich Weigand
https://reviews.llvm.org/D46315
llvm-svn: 333261
Summary:
Lower control flow did not correctly handle the case that a loop break
in if/else was on a condition that was not guaranteed to be masked by
exec. The first test kernel shows an example of this going wrong; after
exiting the loop, exec is all ones, even if it was not before the loop.
The fix is for lowering of if-break and else-break to insert an
S_AND_B64 to mask the break condition with exec. This commit also
includes the optimization of not inserting that S_AND_B64 if it is
obviously not needed because the break condition is the result of a
V_CMP in the same basic block.
V2: Addressed some review comments.
V3: Test fixes.
Subscribers: arsenm, kzhuravl, wdng, nhaehnle, yaxunl, dstuttard, t-tye, llvm-commits
Differential Revision: https://reviews.llvm.org/D44046
Change-Id: I0fc56a01209a9e99d1d5c9b0ffd16f111caf200c
llvm-svn: 333258
Re-add the feature flag for invpcid, which was removed in r294561.
Add an intrinsic, which always uses a 32 bit integer as first argument,
while the instruction actually uses a 64 bit register in 64 bit mode
for the INVPCID_TYPE argument.
Reviewers: craig.topper
Reviewed By: craig.topper
Differential Revision: https://reviews.llvm.org/D47141
llvm-svn: 333255
The existing code has three different ways to try to lower a 64-bit
immediate to the sequence ORR+MOVK. The result is messy: it misses
some possible sequences, and the order of the checks means we sometimes
emit two MOVKs when we only need one.
Instead, just use a simple loop to try all possible two-instruction
ORR+MOVK sequences.
Differential Revision: https://reviews.llvm.org/D47176
llvm-svn: 333218
The match pattern in the definition of LXSDX is xoaddr, so the Pseudo
instruction XFLOADf64 never gets selected. XFLOADf64 expands to LXSDX/LFDX post
RA based on the register pressure. To avoid ambiguity, we need to remove the
select pattern for LXSDX, same as what was done for LXSD. STXSDX also have
the same issue.
Patch by Qing Shan Zhang (steven.zhang).
Differential Revision: https://reviews.llvm.org/D47178
llvm-svn: 333150
