input v8f32 shuffles which are not 128-bit lane crossing but have
different shuffle patterns in the low and high lanes. This removes most
of the extract/insert traffic that was unnecessary and is particularly
good at lowering cases where only one of the two lanes is shuffled at
all.
I've also added a collection of test cases with undef lanes because this
lowering is somewhat more sensitive to undef lanes than others.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218226 91177308-0d34-0410-b5e6-96231b3b80d8
This is purely a plumbing patch. No functional changes intended.
The ultimate goal is to allow targets other than PowerPC (certainly X86 and Aarch64) to turn this:
z = y / sqrt(x)
into:
z = y * rsqrte(x)
using whatever HW magic they can use. See http://llvm.org/bugs/show_bug.cgi?id=20900 .
The first step is to add a target hook for RSQRTE, take the already target-independent code selfishly hoarded by PPC, and put it into DAGCombiner.
Next steps:
The code in DAGCombiner::BuildRSQRTE() should be refactored further; tests that exercise that logic need to be added.
Logic in PPCTargetLowering::BuildRSQRTE() should be hoisted into DAGCombiner.
X86 and AArch64 overrides for TargetLowering.BuildRSQRTE() should be added.
Differential Revision: http://reviews.llvm.org/D5425
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218219 91177308-0d34-0410-b5e6-96231b3b80d8
the new vector shuffle lowering no longer needs to check both symmetric
forms of UNPCK patterns for v4f64.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218217 91177308-0d34-0410-b5e6-96231b3b80d8
lowering when it can use a symmetric SHUFPS across both 128-bit lanes.
This required making the SHUFPS lowering tolerant of other vector types,
and adjusting our canonicalization to canonicalize harder.
This is the last of the clever uses of symmetry I've thought of for
v8f32. The rest of the tricks I'm aware of here are to work around
assymetry in the mask.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218216 91177308-0d34-0410-b5e6-96231b3b80d8
a generic vector shuffle mask into a helper that isn't specific to the
other things that influence which choice is made or the specific types
used with the instruction.
No functionality changed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218215 91177308-0d34-0410-b5e6-96231b3b80d8
of a single element into a zero vector for v4f64 and v4i64 in AVX.
Ironically, there is less to see here because xor+blend is so crazy fast
that we can't really beat that to zero the high 128-bit lane.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218214 91177308-0d34-0410-b5e6-96231b3b80d8
UNPCKHPS with AVX vectors by recognizing those patterns when they are
repeated for both 128-bit lanes.
With this, we now generate the exact same (really nice) code for
Quentin's avx_test_case.ll which was the most significant regression
reported for the new shuffle lowering. In fact, I'm out of specific test
cases for AVX lowering, the rest were AVX2 I think. However, there are
a bunch of pretty obvious remaining things to improve with AVX...
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218213 91177308-0d34-0410-b5e6-96231b3b80d8
important bits of cleverness: to detect and lower repeated shuffle
patterns between the two 128-bit lanes with a single instruction.
This patch just teaches it how to lower single-input shuffles that fit
this model using VPERMILPS. =] There is more that needs to happen here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218211 91177308-0d34-0410-b5e6-96231b3b80d8
v8f32 shuffles in the new vector shuffle lowering code.
This is very cheap to do and makes it much more clear that anything more
expensive but overlapping with this lowering should be selected
afterward (for example using AVX2's VPERMPS). However, no functionality
changed here as without this code we would fall through to create no-op
shuffles of each input and a blend. =]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218209 91177308-0d34-0410-b5e6-96231b3b80d8
VBLENDPD over using VSHUFPD. While the 256-bit variant of VBLENDPD slows
down to the same speed as VSHUFPD on Sandy Bridge CPUs, it has twice the
reciprocal throughput on Ivy Bridge CPUs much like it does everywhere
for 128-bits. There isn't a downside, so just eagerly use this
instruction when it suffices.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218208 91177308-0d34-0410-b5e6-96231b3b80d8
awkward conditions. The readability improvement of this will be even
more important as I generalize it to handle more types.
No functionality changed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218205 91177308-0d34-0410-b5e6-96231b3b80d8
128-bit lane crossings, not 'half' crossings. This came up in code
review ages ago, but I hadn't really addresesd it. Also added some
documentation for the helper.
No functionality changed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218203 91177308-0d34-0410-b5e6-96231b3b80d8
actual support for complex AVX shuffling tricks. We can do independent
blends of the low and high 128-bit lanes of an avx vector, so shuffle
the inputs into place and then do the blend at 256 bits. This will in
many cases remove one blend instruction.
The next step is to permute the low and high halves in-place rather than
extracting them and re-inserting them.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218202 91177308-0d34-0410-b5e6-96231b3b80d8
single-input shuffles with doubles. This allows them to fold memory
operands into the shuffle, etc. This is just the analog to the v4f32
case in my prior commit.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218193 91177308-0d34-0410-b5e6-96231b3b80d8
instruction for single-vector floating point shuffles. This in turn
allows the shuffles to fold a load into the instruction which is one of
the common regressions hit with the new shuffle lowering.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218190 91177308-0d34-0410-b5e6-96231b3b80d8
tricky case of single-element insertion into the zero lane of a zero
vector.
We can't just use the same pattern here as we do in every other vector
type because the general insertion logic can handle insertion into the
non-zero lane of the vector. However, in SSE4.1 with v4f32 vectors we
have INSERTPS that is a much better choice than the generic one for such
lowerings. But INSERTPS can do lots of other lowerings as well so
factoring its logic into the general insertion logic doesn't work very
well. We also can't just extract the core common part of the general
insertion logic that is faster (forming VZEXT_MOVL synthetic nodes that
lower to MOVSS when they can) because VZEXT_MOVL is often *faster* than
a blend while INSERTPS is slower! So instead we do a restrictive
condition on attempting to use the generic insertion logic to narrow it
to those cases where VZEXT_MOVL won't need a shuffle afterward and thus
will do better than INSERTPS. Then we try blending. Then we go back to
INSERTPS.
This still doesn't generate perfect code for some silly reasons that can
be fixed by tweaking the td files for lowering VZEXT_MOVL to use
XORPS+BLENDPS when available rather than XORPS+MOVSS when the input ends
up in a register rather than a load from memory -- BLENDPSrr has twice
the reciprocal throughput of MOVSSrr. Don't you love this ISA?
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218177 91177308-0d34-0410-b5e6-96231b3b80d8
analysis used elsewhere. This removes the last duplicate of this logic.
Also simplify the code here quite a bit. No functionality changed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218176 91177308-0d34-0410-b5e6-96231b3b80d8
floating point types and use it for both v2f64 and v2i64 single-element
insertion lowering.
This fixes the last non-AVX performance regression test case I've gotten
of for the new vector shuffle lowering. There is obvious analogous
lowering for v4f32 that I'll add in a follow-up patch (because with
INSERTPS, v4f32 requires special treatment). After that, its AVX stuff.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218175 91177308-0d34-0410-b5e6-96231b3b80d8
vector lanes can be modeled as zero with a call to the new function that
computes a bit-vector representing that information.
No functionality changed here, but will allow doing more clever things
with the zero-test.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218174 91177308-0d34-0410-b5e6-96231b3b80d8
I just tried reproducing some of the optimization failures in README.txt in the
X86 backend, and many of them could not be reproduced. In general the entire
file appears quite bit-rotted, whatever interesting parts remain should be
moved to bugzilla, and the rest deleted. I did not spend the time to do that,
so I just deleted the few I tried reproducing which are obsolete, to save some
time to whoever will find the courage to do it.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218170 91177308-0d34-0410-b5e6-96231b3b80d8
When looking through sign/zero-extensions the code would always assume there is
such an extension instruction and use the wrong operand for the address.
There was also a minor issue in the handling of 'AND' instructions. I
accidentially used a 'cast' instead of a 'dyn_cast'.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218161 91177308-0d34-0410-b5e6-96231b3b80d8
it from the shuffle pattern matching logic.
Also cleaned up variable names, comments, etc. No functionality changed.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218152 91177308-0d34-0410-b5e6-96231b3b80d8
In r217636, the value stored in KernelInfo.Num[VS]GPRSs was changed from
the highest GPR index used to the number of gprs in order to be
consistent with the name of the variable.
The code writing the config values still assumed that the value in this
variable was the highest GPR index used, which caused the compiler to
over report the number of GPRs being used.
https://bugs.freedesktop.org/show_bug.cgi?id=84089
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218150 91177308-0d34-0410-b5e6-96231b3b80d8
lowering to support both anyext and zext and to custom lower for many
different microarchitectures.
Using this allows us to get *exactly* the right code for zext and anyext
shuffles in all the vector sizes. For v16i8, the improvement is *huge*.
The new SSE2 test case added I refused to add before this because it was
sooooo muny instructions.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218143 91177308-0d34-0410-b5e6-96231b3b80d8
The heuristic used by DAGCombine to form FMAs checks that the FMUL has only one
use, but this is overly-conservative on some systems. Specifically, if the FMA
and the FADD have the same latency (and the FMA does not compete for resources
with the FMUL any more than the FADD does), there is no need for the
restriction, and furthermore, forming the FMA leaving the FMUL can still allow
for higher overall throughput and decreased critical-path length.
Here we add a new TLI callback, enableAggressiveFMAFusion, false by default, to
elide the hasOneUse check. This is enabled for PowerPC by default, as most
PowerPC systems will benefit.
Patch by Olivier Sallenave, thanks!
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218120 91177308-0d34-0410-b5e6-96231b3b80d8
to undef lanes as well as defined widenable lanes. This dramatically
improves the lowering we use for undef-shuffles in a zext-ish pattern
for SSE2.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218115 91177308-0d34-0410-b5e6-96231b3b80d8
shuffles that are zext-ing.
Not a lot to see here; the undef lane variant is better handled with
pshufd, but this improves the actual zext pattern.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218112 91177308-0d34-0410-b5e6-96231b3b80d8
to the new vector shuffle lowering code.
This allows us to emit PMOVZX variants consistently for patterns where
it is a viable lowering. This instruction is both fast and allows us to
fold loads into it. This only hooks the new lowering up for i16 and i8
element widths, mostly so I could manage the change to the tests. I'll
add the i32 one next, although it is significantly less interesting.
One thing to note is that we already had some tests for these patterns
but those tests had far less horrible instructions. The problem is that
those tests weren't checking the strict start and end of the instruction
sequence. =[ As a consequence something changed in the lowering making
us generate *TERRIBLE* code for these patterns in SSE2 through SSSE3.
I've consolidated all of the tests and spelled out the madness that we
currently emit for these shuffles. I'm going to try to figure out what
has gone wrong here.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218102 91177308-0d34-0410-b5e6-96231b3b80d8
The fix is slightly different then x86 (see r216117) because the number of values
attached to a return can vary even for a single returned value (e.g., f64 yields
two returned values).
<rdar://problem/18352998>
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218076 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
This patch was originally in D5304 (I could not find a way to reopen that revision).
It was accepted, commited and broke the build bots because the overloading of
the constructor of ArrayRef for braced initializer lists is not supported by all
toolchains. I then reverted it, and propose this fixed version that uses a plain
C array instead in makeDMB (that array is then converted implicitly to an
ArrayRef, but that is not behind an ifdef). Could someone confirm me whether
initialization lists for plain C arrays are supported by every toolchain used
to build llvm ? Otherwise I can just initialize the array in the old way:
args[0] = ...; .. ; args[5] = ...;
Below is the description of the original patch:
```
I had only tested this code for ARMv7 and ARMv8. This patch adds several
fallback paths if the processor does not support dmb ish:
- dmb sy if a cortex-M with support for dmb
- mcr p15, #0, r0, c7, c10, #5 for ARMv6 (special instruction equivalent to a DMB)
These fallback paths were chosen based on the code for fence seq_cst.
Thanks to luqmana for having noticed this bug.
```
Test Plan: Added more cases to atomic-load-store.ll + make check-all
Reviewers: jfb, t.p.northover, luqmana
Subscribers: llvm-commits, aemerson
Differential Revision: http://reviews.llvm.org/D5386
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218066 91177308-0d34-0410-b5e6-96231b3b80d8
There is no purpose in using it for single-input shuffles as
pshufd is just as fast and doesn't tie the two operands. This removes
a substantial amount of wrong-domain blend operations in SSSE3 mode. It
also completes the usage of PALIGNR for integer shuffles and addresses
one of the test cases Quentin hit with the new vector shuffle lowering.
There is still the question of whether and when to use this for floating
point shuffles. It is faster than shufps or shufpd but in the integer
domain. I don't yet really have a good heuristic here for when to use
this instruction for floating point vectors.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218038 91177308-0d34-0410-b5e6-96231b3b80d8
Summary:
The N32/N64 ABI's return f128 values in $f0 and $f2 for hard-float and $v0 and
$a0 for soft-float. The registers used in the soft-float case differ from the
usual $v0, and $v1 specified for return values.
Both cases were previously handled by duplicating the CCState::AnalyzeReturn()
and CCState::AnalyzeCallReturn() functions and modifying them to delegate to
a different assignment function for f128 and further replace the register type
for the hard-float case. There is a simpler way to do both of these.
We now use the common functions and select an initial assignment function based
on whether the original type is f128 or not. We then handle the hard-float case
using CCBitConvertToType<>.
No functional change.
Reviewers: vmedic
Reviewed By: vmedic
Subscribers: llvm-commits
Differential Revision: http://reviews.llvm.org/D5269
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218036 91177308-0d34-0410-b5e6-96231b3b80d8
When folding the intrinsic flag into the branch or select we also have to
consider the fact if the intrinsic got simplified, because it changes the
flag we have to check for.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218034 91177308-0d34-0410-b5e6-96231b3b80d8
Small optimization in 'simplifyAddress'. When the offset cannot be encoded in
the load/store instruction, then we need to materialize the address manually.
The add instruction can encode a wider range of immediates than the load/store
instructions. This change tries to fold the offset into the add instruction
first before materializing the offset in a register.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218031 91177308-0d34-0410-b5e6-96231b3b80d8
The 'AND' instruction could be used to mask out the lower 32 bits of a register.
If this is done inside an address computation we might be able to fold the
instruction into the memory instruction itself.
and x1, x1, #0xffffffff ---> ldrb x0, [x0, w1, uxtw]
ldrb x0, [x0, x1]
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218030 91177308-0d34-0410-b5e6-96231b3b80d8
Certain directives are unsupported on Windows (some of which could/should be
supported). We would not diagnose the use but rather crash during the emission
as we try to access the Target Streamer. Add an assertion to prevent creating a
NULL reference (which is not permitted under C++) as well as a test to ensure
that we can diagnose the disabled directives.
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218014 91177308-0d34-0410-b5e6-96231b3b80d8
