There was a problem with the old pattern, so we were copying some
larger immediates into registers when we could have been encoding
them in the instruction.
llvm-svn: 200932
No functional change. Updated loops from:
for (I = scc_begin(), E = scc_end(); I != E; ++I)
to:
for (I = scc_begin(); !I.isAtEnd(); ++I)
for teh win.
llvm-svn: 200789
This fixes a crashes in the OpenCV test suite and also the scrypt
kernel in bfgminer.
I was unable to come up with a reduced test case for this.
https://bugs.freedesktop.org/show_bug.cgi?id=72785
llvm-svn: 200776
There is no lit test for this, because it would be too big and
complicated, but it does fix a crash in the Arithm/Absdiff.* OpenCV test.
llvm-svn: 200775
The OpenCL specs say: "The vector versions of the math functions operate
component-wise. The description is per-component."
Patch by: Jan Vesely
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 200773
V_ADD_F32 with source modifier does not produce -0.0 for this. Just
manipulate the sign bit directly instead.
Also add a pattern for (fneg (fabs ...)).
Fixes a bunch of bit encoding piglit tests with radeonsi.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 200743
This didn't work for any integer vectors, and didn't
work with some sizes of float vectors. This should now
work with all sizes of float and i32 vectors.
llvm-svn: 200619
There is nothing wrong with printing the disassembly section when printing
text. An hypothetical assembler would then produce a .o just like our
direct object emission produces.
llvm-svn: 200583
The subtarget info is explicitly passed to the EncodeInstruction
method and we should use that subtarget info to influence any
encoding decisions.
llvm-svn: 200350
This has a few advantages:
* Only targets that use a MCTargetStreamer have to worry about it.
* There is never a MCTargetStreamer without a MCStreamer, so we can use a
reference.
* A MCTargetStreamer can talk to the MCStreamer in its constructor.
llvm-svn: 200129
This fixes a crash in the OpenCV OpenCL test suite.
There is no lit test for this, because the test would be very large
and could easily be invalidated by changes to the scheduler
or other parts of the compiler.
Patch by: Vincent Lejeune
llvm-svn: 199919
This pattern uses an SDNodeXForm, which isn't being emitted for some
reason. I can get it to work by attaching the PatLeaf that has the
XForm to the argument in the output pattern, but this results in an
immediate being used in a register operand, which the backend can't
handle yet.
llvm-svn: 199918
The control flow finalizer would sometimes use an ALU_POP_AFTER
instruction before the vetex fetch clause instead of using a POP
instruction after it.
llvm-svn: 199917
Implement the getUnrollingPreferences() function for
AMDGPUTargetTransformInfo so that loops that do address calculations
on pointers derived from alloca are unconditionally unrolled.
Unrolling these loops makes it more likely that SROA will be able to
eliminate the allocas, which is a big win for R600 since memory
allocated by alloca (private memory) is really slow.
llvm-svn: 199916
The unit test is now disabled on non-asserts builds.
The CF stack can be corrupted if you use CF_ALU_PUSH_BEFORE,
CF_ALU_ELSE_AFTER, CF_ALU_BREAK, or CF_ALU_CONTINUE when the number of
sub-entries on the stack is greater than or equal to the stack entry
size and sub-entries modulo 4 is either 0 or 3 (on cedar the bug is
present when number of sub-entries module 8 is either 7 or 0)
We choose to be conservative and always apply the work-around when the
number of sub-enries is greater than or equal to the stack entry size,
so that we can safely over-allocate the stack when we are unsure of the
stack allocation rules.
reviewed-by: Vincent Lejeune <vljn at ovi.com>
llvm-svn: 199905
This reverts commit 35b8331cad6eb512a2506adbc394201181da94ba.
The -debug-only flag for llc doesn't appear to be available in
all build configurations.
llvm-svn: 199845
The CF stack can be corrupted if you use CF_ALU_PUSH_BEFORE,
CF_ALU_ELSE_AFTER, CF_ALU_BREAK, or CF_ALU_CONTINUE when the number of
sub-entries on the stack is greater than or equal to the stack entry
size and sub-entries modulo 4 is either 0 or 3 (on cedar the bug is
present when number of sub-entries module 8 is either 7 or 0)
We choose to be conservative and always apply the work-around when the
number of sub-enries is greater than or equal to the stack entry size,
so that we can safely over-allocate the stack when we are unsure of the
stack allocation rules.
reviewed-by: Vincent Lejeune <vljn at ovi.com>
llvm-svn: 199842
can be used by both the new pass manager and the old.
This removes it from any of the virtual mess of the pass interfaces and
lets it derive cleanly from the DominatorTreeBase<> template. In turn,
tons of boilerplate interface can be nuked and it turns into a very
straightforward extension of the base DominatorTree interface.
The old analysis pass is now a simple wrapper. The names and style of
this split should match the split between CallGraph and
CallGraphWrapperPass. All of the users of DominatorTree have been
updated to match using many of the same tricks as with CallGraph. The
goal is that the common type remains the resulting DominatorTree rather
than the pass. This will make subsequent work toward the new pass
manager significantly easier.
Also in numerous places things became cleaner because I switched from
re-running the pass (!!! mid way through some other passes run!!!) to
directly recomputing the domtree.
llvm-svn: 199104
trees into the Support library.
These are all expressed in terms of the generic GraphTraits and CFG,
with no reliance on any concrete IR types. Putting them in support
clarifies that and makes the fact that the static analyzer in Clang uses
them much more sane. When moving the Dominators.h file into the IR
library I claimed that this was the right home for it but not something
I planned to work on. Oops.
So why am I doing this? It happens to be one step toward breaking the
requirement that IR verification can only be performed from inside of
a pass context, which completely blocks the implementation of
verification for the new pass manager infrastructure. Fixing it will
also allow removing the concept of the "preverify" step (WTF???) and
allow the verifier to cleanly flag functions which fail verification in
a way that precludes even computing dominance information. Currently,
that results in a fatal error even when you ask the verifier to not
fatally error. It's awesome like that.
The yak shaving will continue...
llvm-svn: 199095
directory. These passes are already defined in the IR library, and it
doesn't make any sense to have the headers in Analysis.
Long term, I think there is going to be a much better way to divide
these matters. The dominators code should be fully separated into the
abstract graph algorithm and have that put in Support where it becomes
obvious that evn Clang's CFGBlock's can use it. Then the verifier can
manually construct dominance information from the Support-driven
interface while the Analysis library can provide a pass which both
caches, reconstructs, and supports a nice update API.
But those are very long term, and so I don't want to leave the really
confusing structure until that day arrives.
llvm-svn: 199082
subsequent changes are easier to review. About to fix some layering
issues, and wanted to separate out the necessary churn.
Also comment and sink the include of "Windows.h" in three .inc files to
match the usage in Memory.inc.
llvm-svn: 198685
v2: Add ftrunc->TRUNC pattern instead of replacing int_AMDGPU_trunc
v3: move ftrunc pattern next to TRUNC definition, it's available since R600
Patch By: Jan Vesely
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Jan Vesely <jan.vesely@rutgers.edu>
llvm-svn: 197783
Different sized address spaces should theoretically work
most of the time now, and since 64-bit add is currently
disabled, using more 32-bit pointers fixes some cases.
llvm-svn: 197659
Without this, MachineCSE is powerless to handle redundant operations with truncated source operands.
This required fixing the 2-addr pass to handle tied subregisters. It isn't clear what combinations of subregisters can legally be tied, but the simple case of truncated source operands is now safely handled:
%vreg11<def> = COPY %vreg1:sub_32bit; GR32:%vreg11 GR64:%vreg1
%vreg12<def> = COPY %vreg2:sub_32bit; GR32:%vreg12 GR64:%vreg2
%vreg13<def,tied1> = ADD32rr %vreg11<tied0>, %vreg12<kill>, %EFLAGS<imp-def>
Test case: cse-add-with-overflow.ll.
This exposed an existing bug in
PPCInstrInfo::commuteInstruction. Thanks to Rafael for the test case:
PowerPC/crash.ll.
llvm-svn: 197465
I think, in principle, intrinsics_gen may be added explicitly.
That said, it can be added incidentally, since each target already has dependencies to llvm-tblgen.
Almost all source files depend on both CommonTaleGen and intrinsics_gen.
Explicit add_dependencies() have been pruned under lib/Target.
llvm-svn: 195929
add_public_tablegen_target adds *CommonTableGen to LLVM_COMMON_DEPENDS.
LLVM_COMMON_DEPENDS affects add_llvm_library (and other add_target stuff) within its scope.
llvm-svn: 195927
SGPRs are spilled into VGPRs using the {READ,WRITE}LANE_B32 instructions.
v2:
- Fix encoding of Lane Mask
- Use correct register flags, so we don't overwrite the low dword
when restoring multi-dword registers.
v3:
- Register spilling seems to hang the GPU, so replace all shaders
that need spilling with a dummy shader.
v4:
- Fix *LANE definitions
- Change destination reg class for 32-bit SMRD instructions
v5:
- Remove small optimization that was crashing Serious Sam 3.
https://bugs.freedesktop.org/show_bug.cgi?id=68224https://bugs.freedesktop.org/show_bug.cgi?id=71285
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195880
Writing to the M0 register from an SMRD instruction hangs the GPU, so
we need to use the SGPR_32 register class, which does not include M0.
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195879
We were ignoring the ordered/onordered bits and also the signed/unsigned
bits of condition codes when lowering the DAG to MachineInstrs.
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195514
Splitting a basic block will create a new ALU clause, so we need to make
sure we aren't moving uses of registers that are local to their
current clause into a new one.
I had a test case for this, but unfortunately unrelated schedule changes
invalidated it, and I wasn't been able to come up with another one.
NOTE: This is a candidate for the 3.4 branch.
llvm-svn: 195399
This patch removes most of the trivial cases of weak vtables by pinning them to
a single object file. The memory leaks in this version have been fixed. Thanks
Alexey for pointing them out.
Differential Revision: http://llvm-reviews.chandlerc.com/D2068
Reviewed by Andy
llvm-svn: 195064
This change is incorrect. If you delete virtual destructor of both a base class
and a subclass, then the following code:
Base *foo = new Child();
delete foo;
will not cause the destructor for members of Child class. As a result, I observe
plently of memory leaks. Notable examples I investigated are:
ObjectBuffer and ObjectBufferStream, AttributeImpl and StringSAttributeImpl.
llvm-svn: 194997
This patch removes most of the trivial cases of weak vtables by pinning them to
a single object file.
Differential Revision: http://llvm-reviews.chandlerc.com/D2068
Reviewed by Andy
llvm-svn: 194865
This is to avoid this transformation in some cases:
fold (conv (load x)) -> (load (conv*)x)
On architectures that don't natively support some vector
loads efficiently casting the load to a smaller vector of
larger types and loading is more efficient.
Patch by Micah Villmow.
llvm-svn: 194783
The LDS output queue is accessed via the OQAP register. The OQAP
register cannot be live across clauses, so if value is written to the
output queue, it must be retrieved before the end of the clause.
With the machine scheduler, we cannot statisfy this constraint, because
it lacks proper alias analysis and it will mark some LDS accesses as
having a chain dependency on vertex fetches. Since vertex fetches
require a new clauses, the dependency may end up spiltting OQAP uses and
defs so the end up in different clauses. See the lds-output-queue.ll
test for a more detailed explanation.
To work around this issue, we now combine the LDS read and the OQAP
copy into one instruction and expand it after register allocation.
This patch also adds some checks to the EmitClauseMarker pass, so that
it doesn't end a clause with a value still in the output queue and
removes AR.X and OQAP handling from the scheduler (AR.X uses and defs
were already being expanded post-RA, so the scheduler will never see
them).
Reviewed-by: Vincent Lejeune <vljn at ovi.com>
llvm-svn: 194755
All shift operations will be selected as SALU instructions and then
if necessary lowered to VALU instructions in the SIFixSGPRCopies pass.
This allows us to do more operations on the SALU which will improve
performance and is also required for implementing private memory
using indirect addressing, since the private memory pointers must stay
in the scalar registers.
This patch includes some fixes from Matt Arsenault.
llvm-svn: 194625
Accepting quotes is a property of an assembler, not of an object file. For
example, ELF can support any names for sections and symbols, but the gnu
assembler only accepts quotes in some contexts and llvm-mc in a few more.
LLVM should not produce different symbols based on a guess about which assembler
will be reading the code it is printing.
llvm-svn: 194575
Print the range of registers used with a single letter prefix.
This better matches what the shader compiler produces and
is overall less obnoxious than concatenating all of the
subregister names together.
Instead of SGPR0, it will print s0. Instead of SGPR0_SGPR1,
it will print s[0:1] and so on.
There doesn't appear to be a straightforward way
to get the actual register info in the InstPrinter,
so this parses the generated name to print with the
new syntax.
The required test changes are pretty nasty, and register
matching regexes are now worse. Since there isn't a way to
add to a variable in FileCheck, some of the tests now don't
check the exact number of registers used, but I don't think that
will be a real problem.
llvm-svn: 194443
The SelectionDAGBuilder was promoting vector kernel arguments to legal
types, but this won't work for R600 and SI since kernel arguments are
stored in memory and can't be promoted. In order to handle vector
arguments correctly we need to look at the original types from the LLVM IR
function.
llvm-svn: 193215
The AMDGPUIndirectAddressing pass was previously responsible for
lowering private loads and stores to indirect addressing instructions.
However, this pass was buggy and way too complicated. The only
advantage it had over the new simplified code was that it saved one
instruction per direct write to private memory. This optimization
likely has a minimal impact on performance, and we may be able
to duplicate it using some other transformation.
For the private address space, we now:
1. Lower private loads/store to Register(Load|Store) instructions
2. Reserve part of the register file as 'private memory'
3. After regalloc lower the Register(Load|Store) instructions to
MOV instructions that use indirect addressing.
llvm-svn: 193179
We were calling llvm_unreachable() when failing to optimize the
branch into if case. However, it is still possible for us
to structurize the CFG by duplicating blocks even if this optimization
fails.
Reviewed-by: Vincent Lejeune<vljn at ovi.com>
llvm-svn: 192813
During instruction selection, we rewrite the destination register
class for MIMG instructions based on their writemasks. This creates
machine verifier errors since the new register class does not match
the register class in the MIMG instruction definition.
We can avoid this by defining different MIMG instructions for each
possible destination type and then switching to the correct instruction
when we change the register class.
llvm-svn: 192365
This prevents the machine verifier from complaining about uses of
an undefined physical register.
Reviewed-by: Vincent Lejeune<vljn at ovi.com>
llvm-svn: 192364
StructurizeCFG pass allows to make complex cfg reducible ; it allows a lot of
shader from shadertoy (which exhibits complex control flow constructs) to works
correctly with respect to CFG handling (and allow us to detect potential bug in
other part of the backend).
We provide a cmd line argument to disable the pass for debug purpose.
Patch by: Vincent Lejeune
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 192363
This patch fixes an old FIXME by creating a MCTargetStreamer interface
and moving the target specific functions for ARM, Mips and PPC to it.
The ARM streamer is still declared in a common place because it is
used from lib/CodeGen/ARMException.cpp, but the Mips and PPC are
completely hidden in the corresponding Target directories.
I will send an email to llvmdev with instructions on how to use this.
llvm-svn: 192181
For targets that have instruction itineraries this means no change. Targets
that move over to the new schedule model will use be able the new schedule
module for instruction latencies in the if-converter (the logic is such that if
there is no itineary we will use the new sched model for the latencies).
Before, we queried "TTI->getInstructionLatency()" for the instruction latency
and the extra prediction cost. Now, we query the TargetSchedule abstraction for
the instruction latency and TargetInstrInfo for the extra predictation cost. The
TargetSchedule abstraction will internally call "TTI->getInstructionLatency" if
an itinerary exists, otherwise it will use the new schedule model.
ATTENTION: Out of tree targets!
(I will also send out an email later to LLVMDev)
This means, if your target implements
unsigned getInstrLatency(const InstrItineraryData *ItinData,
const MachineInstr *MI,
unsigned *PredCost);
and returns a value for "PredCost", you now also need to implement
unsigned getPredictationCost(const MachineInstr *MI);
(if your target uses the IfConversion.cpp pass)
radar://15077010
llvm-svn: 191671
We were completely ignoring the unorder/ordered attributes of condition
codes and also incorrectly lowering seto and setuo.
Reviewed-by: Vincent Lejeune<vljn at ovi.com>
llvm-svn: 191603
SelectionDAG will now attempt to inverse an illegal conditon in order to
find a legal one and if that doesn't work, it will attempt to swap the
operands using the inverted condition.
There are no new test cases for this, but a nubmer of the existing R600
tests hit this path.
llvm-svn: 191602
This is useful for targets like R600, which only support GT, GE, NE, and EQ
condition codes as it removes the need to handle unsupported condition
codes in target specific code.
There are no tests with this commit, but R600 has been updated to take
advantage of this new feature, so its existing selectcc tests are now
testing the swapped operands path.
llvm-svn: 191601
Previously, the DAGISel function WalkChainUsers was spotting that it
had entered already-selected territory by whether a node was a
MachineNode (amongst other things). Since it's fairly common practice
to insert MachineNodes during ISelLowering, this was not the correct
check.
Looking around, it seems that other nodes get their NodeId set to -1
upon selection, so this makes sure the same thing happens to all
MachineNodes and uses that characteristic to determine whether we
should stop looking for a loop during selection.
This should fix PR15840.
llvm-svn: 191165
The global registry is used to allow command line override of the
scheduler selection, but does not work well as the normal selection
API. For example, the same LLVM process should be able to target
multiple targets or subtargets.
llvm-svn: 191071
For _XYZ, the type of VDATA is v4i32, because v3i32 doesn't exist.
The ADDR64 bit is not exposed. A simpler intrinsic that doesn't take
a resource descriptor might be nicer.
The maximum number of input SGPRs is bumped to 17.
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 190575
This fixes some regressions in the piglit local memory store tests
introduced by recent commits which made the scheduler aware of the trans
slot.
It's not possible to test this using lit, because there is no way to
determine from the assembly dumps whether or not an instruction is in
the trans slot.
Even if this were possible, the test would be highly sensitive to
changes in the scheduler and might generate confusing false negatives.
Reviewed-by: Vincent Lejeune<vljn at ovi.com>
llvm-svn: 190574
We used to generate the compact unwind encoding from the machine
instructions. However, this had the problem that if the user used `-save-temps'
or compiled their hand-written `.s' file (with CFI directives), we wouldn't
generate the compact unwind encoding.
Move the algorithm that generates the compact unwind encoding into the
MCAsmBackend. This way we can generate the encoding whether the code is from a
`.ll' or `.s' file.
<rdar://problem/13623355>
llvm-svn: 190290
This pass was segfaulting when it ran into a non-intrinsic function
call. Function calls are not supported, so now instead of segfaulting,
we will get an assertion failure with a nice error message.
I'm not sure how to test this using lit.
llvm-svn: 190076
Iterator of std::vector may be implemented as a raw pointer. In
this case begin iterators are rvalues and cannot be incremented.
For example, this is the case with STDCXX implementation of vector.
Patch by Konstantin Tokarev <annulen@yandex.ru>.
llvm-svn: 189911
This adds minimal support to the SelectionDAG for handling address spaces
with different pointer sizes. The SelectionDAG should now correctly
lower pointer function arguments to the correct size as well as generate
the correct code when lowering getelementptr.
This patch also updates the R600 DataLayout to use 32-bit pointers for
the local address space.
v2:
- Add more helper functions to TargetLoweringBase
- Use CHECK-LABEL for tests
llvm-svn: 189221
The logic in SIInsertWaits::getHwCounts() only really made sense for SMRD
instructions, and trying to shoehorn it into handling DS_WRITE_B32 caused
it to corrupt the encoding of that by clobbering the first operand with
the second one.
Undo that damage and only apply the SMRD logic to that.
Fixes some derivates related piglit regressions with radeonsi.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 188558
The SIInsertWaits pass was overwriting the first operand (gds bit) of
DS_WRITE_B32 with the second operand (value to write). This meant that
any time the value to write was stored in an odd number VGPR, the gds
bit would be set causing the instruction to write to GDS instead of LDS.
llvm-svn: 188522
Now that compute support is better on SI, we can't continue using v16i8
for descriptors since this is also a legal type in OpenCL.
This patch fixes numerous hangs with the piglit OpenCL test and since
we now use a target specific DAG node for LOAD_CONSTANT with the
correct MemOperandFlags, this should also fix:
https://bugs.freedesktop.org/show_bug.cgi?id=66805
llvm-svn: 188429
Using REG_SEQUENCE for BUILD_VECTOR rather than a series of INSERT_SUBREG
instructions should make it easier for the register allocator to coalasce
unnecessary copies.
v2:
- Use an SGPR register class if all the operands of BUILD_VECTOR are
SGPRs.
llvm-svn: 188427
The instruction selector will now try to infer the destination register
so it can decided whether to use V_MOV_B32 or S_MOV_B32 when copying
immediates.
llvm-svn: 188426
The previous code declared the operand as unknown:$vaddr, which made
it possible for scalar registers to be used instead of vector registers.
llvm-svn: 188425
R600 doesn't need to do any scheduling on the SelectionDAG now that it
has a very good MachineScheduler. Also, using the VLIW SelectionDAG
scheduler was having a major impact on compile times. For example with
the phatk kernel here are the LLVM IR to machine code compile times:
With Sched::VLIW
Total Compile Time: 1.4890 Seconds (User + System)
SelectionDAG Instruction Scheduling: 1.1670 Seconds (User + System)
With Sched::Source
Total Compile Time: 0.3330 Seconds (User + System)
SelectionDAG Instruction Scheduling: 0.0070 Seconds (User + System)
The code ouput was identical with both schedulers. This may not be true
for all programs, but it gives me confidence that there won't be much
reduction, if any, in code quality by using Sched::Source.
llvm-svn: 188215
Since the VSrc_* register classes contain both VGPRs and SGPRs, copies
that used be emitted by isel like this:
SGPR = COPY VGPR
Will now be emitted like this:
VSrC = COPY VGPR
This patch also adds a pass that tries to identify and fix situations where
a VGPR to SGPR copy may occur. Hopefully, these changes will make it
impossible for the compiler to generate illegal VGPR to SGPR copies.
llvm-svn: 187831
Without explicit dependencies, both per-file action and in-CommonTableGen action could run in parallel.
It races to emit *.inc files simultaneously.
llvm-svn: 187780
* Added R600_Reg64 class
* Added T#Index#.XY registers definition
* Added v2i32 register reads from parameter and global space
* Added f32 and i32 elements extraction from v2f32 and v2i32
* Added v2i32 -> v2f32 conversions
Tom Stellard:
- Mark vec2 operations as expand. The addition of a vec2 register
class made them all legal.
Patch by: Dmitry Cherkassov
Signed-off-by: Dmitry Cherkassov <dcherkassov@gmail.com>
llvm-svn: 187582
If we merge vector when a vector is used, it will generate an artificial
antidependency that can prevent 2 tex/vtx instructions to use the same
clause and thus generate extra clauses that reduce performance.
There is no test case as such situation is really hard to predict.
llvm-svn: 187516
There are a lot of restrictions on instruction groups that contain
LDS instructions, so for now we will be conservative and not packetize
anything else with them.
llvm-svn: 187513
We were using two instructions for similar purpose : break and
predicated break. Only predicated_break was emitted and it was
lowered at R600ControlFlowFinalizer to JUMP;CF_BREAK;POP.
This commit simplify the situation by making AMDILCFGStructurizer
emit IF_PREDICATE;BREAK;ENDIF; instead of predicated_break (which
is now removed).
There is no functionality change.
llvm-svn: 187510
Merge consecutive if-regions if they contain identical statements.
Both transformations reduce number of branches. The transformation
is guarded by a target-hook, and is currently enabled only for +R600,
but the correctness has been tested on X86 target using a variety of
CPU benchmarks.
Patch by: Mei Ye
llvm-svn: 187278
These are really the same address space in hardware. The only
difference is that CONSTANT_ADDRESS uses a special cache for faster
access. When we are unable to use the constant kcache for some reason
(e.g. smaller types or lack of indirect addressing) then the instruction
selector must use GLOBAL_ADDRESS loads instead.
llvm-svn: 187006
This increases the number of opportunites we have for folding. With the
previous implementation we were unable to fold into any instructions
other than the first when multiple instructions were selected from a
single SDNode.
Reviewed-by: Vincent Lejeune <vljn at ovi.com>
llvm-svn: 186919
A side-effect of this is that now the compiler expects kernel arguments
to be 4-byte aligned.
Reviewed-by: Vincent Lejeune <vljn at ovi.com>
llvm-svn: 186916
Test is not included as it is several 1000 lines long.
To test this functionnality, a test case must generate at least 2 ALU clauses,
where an ALU clause is ~110 instructions long.
NOTE: This is a candidate for the stable branch.
llvm-svn: 185943
This is dead code since PIC16 was removed in 2010. The result was an odd mix,
where some parts would carefully pass it along and others would assert it was
zero (most of the object streamer for example).
llvm-svn: 185436
By default, we expand these operations for both EG and SI. Move the
duplicated code into a common space for now. If the targets ever actually
implement these operations as instructions, we can override that in the relevant
target.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 184848
Also add lit test for both cases on SI, and v2i32 for evergreen.
Note: I followed the guidance of the v4i32 EG check... UREM produces really
complex code, so let's just check that the instruction was lowered
successfully.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 184844
Also add lit test for both cases on SI, and v2i32 for evergreen.
Note: I followed the guidance of the v4i32 EG check... UDIV produces really
complex code, so let's just check that the instruction was lowered
successfully.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
llvm-svn: 184843
In reality, some unaligned memory accesses are legal for 32-bit types and
smaller too, but it all depends on the address space. Allowing
unaligned loads/stores for > 32-bit types is mainly to prevent the
legalizer from splitting one load into multiple loads of smaller types.
https://bugs.freedesktop.org/show_bug.cgi?id=65873
llvm-svn: 184822
This should only make a difference in programs that use a lot of the
vector ALU instructions like BFI_INT and BIT_ALIGN. There is a slight
improvement in the phatk bitcoin mining kernel with this patch on
Evergreen (vector size == 1):
Before:
1173 Instruction Groups / 9520 dwords
After:
1167 Instruction Groups / 9510 dwords
Reviewed-by: Reviewed-by: Vincent Lejeune<vljn at ovi.com>
llvm-svn: 184819
Also add a v2i32 test to the existing v4i32 test.
Patch by: Aaron Watry
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Aaron Watry<awatry@gmail.com>
llvm-svn: 184482
Also add SI tests to existing file and a v2i32 test for both
R600 and SI.
Patch by: Aaron Watry
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 184481
The custom lowering causes llc to crash with a segfault.
Ideally, the custom lowering can be fixed, but this allows
programs which load/store v2i32 to work without crashing.
Patch by: Aaron Watry
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Aaron Watry<awatry@gmail.com>
llvm-svn: 184480
Also add a seperate vector lit test file, since r600 doesn't seem to handle
v2i32 load/store yet, but we can test both for SI.
Patch by: Aaron Watry
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Aaron Watry <awatry@gmail.com>
llvm-svn: 184021
We were using RAT_INST_STORE_RAW, which seemed to work, but the docs
say this instruction doesn't exist for Cayman, so it's probably safer
to use a documented instruction instead.
Reviewed-by: Vincent Lejeune<vljn at ovi.com>
llvm-svn: 184015
This is using a hint from AMD APP OpenCL Programming Guide with
empirically tweaked parameters.
I used Unigine Heaven 3.0 to determine best parameters on my system
(i7 2600/Radeon 6950/Kernel 3.9.4) the benchmark :
it went from 38.8 average fps to 39.6, which is ~3% gain.
(Lightmark 2008.2 gain is much more marginal: from 537 to 539)
There is no lit test provided as the parameter were determined
empirically and it it would be nearly impossiblet to find a test
program that check for optimal behavior.
llvm-svn: 183593
We weren't computing structure size correctly and we were relying on
the original alloca instruction to compute the offset, which isn't
always reliable.
Reviewed-by: Vincent Lejeune <vljn@ovi.com>
llvm-svn: 183568
The CopyToReg nodes will sometimes try to copy a value from a VGPR to an
SGPR. This kind of copy is not possible, so we need to detect
VGPR->SGPR copies and do something else. The current strategy is to
replace these copies with VGPR->VGPR copies and hope that all the users
of CopyToReg can accept VGPRs as arguments.
llvm-svn: 183132
Tom Stellard