GregF [Fri, 6 Oct 2017 02:37:00 +0000 (20:37 -0600)]
DeadBranchElim: Add dead case elimination
Expands dead branch elimination to eliminate dead switch cases. It also
changes dbe to eliminate orphaned merge blocks and recursively eliminate
any blocks thereby orphaned.
Andrey Tuganov [Thu, 5 Oct 2017 23:07:31 +0000 (19:07 -0400)]
Add Android.mk to SPIRV-Tools
Android.mk was migrated from shaderc/third_party.
android_test is a stub, only builds spirv-tools libs.h to test build.
Diego Novillo [Wed, 30 Aug 2017 18:19:22 +0000 (14:19 -0400)]
Add -O, -Os and -Oconfig flags.
These flags are expanded to a series of spirv-opt flags with the
following semantics:
-O: expands to passes that attempt to improve the performance of the
generated code.
-Os: expands to passes that attempt to reduce the size of the generated
code.
-Oconfig=<file> expands to the sequence of passes determined by the
flags specified in the user-provided file.
Tim Diekmann [Tue, 10 Oct 2017 12:23:42 +0000 (14:23 +0200)]
Set cmake-policy CMP0048 to NEW
Pierre Moreau [Thu, 13 Jul 2017 00:16:51 +0000 (02:16 +0200)]
Implement Linker (module combiner)
Add extra iterators for ir::Module's sections
Add extra getters to ir::Function
Add a const version of BasicBlock::GetLabelInst()
Use the max of all inputs' version as version
Split debug in debug1 and debug2
- Debug1 instructions have to be placed before debug2 instructions.
Error out if different addressing or memory models are found
Exit early if no binaries were given
Error out if entry points are redeclared
Implement copy ctors for Function and BasicBlock
- Visual Studio ends up generating copy constructors that call deleted
functions while compiling the linker code, while GCC and clang do not.
So explicitly write those functions to avoid Visual Studio messing up.
Move removing duplicate capabilities to its own pass
Add functions running on all IDs present in an instruction
Remove duplicate SpvOpExtInstImport
Give default options value for link functions
Remove linkage capability if not making a library
Check types before allowing to link
Detect if two types/variables/functions have different decorations
Remove decorations of imported variables/functions and their types
Add a DecorationManager
Add a method for removing all decorations of id
Add methods for removing operands from instructions
Error out if one of the modules has a non-zero schema
Update README.md to talk about the linker
Do not freak out if an imported built-in variable has no export
Andrey Tuganov [Thu, 5 Oct 2017 20:26:09 +0000 (16:26 -0400)]
Remove duplicate dead branch elim pass declaration
The function had two declarations in the same header with somewhat
different comments.
Andrew Woloszyn [Thu, 5 Oct 2017 15:01:34 +0000 (11:01 -0400)]
Hack around bug in gcc-4.8.1 templates.
This keeps the previous behavior for other compilers that will
throw warnings on a negative shift operation, but works around
the internal compiler error in GCC.
GregF [Fri, 29 Sep 2017 22:22:39 +0000 (16:22 -0600)]
AggressiveDCE: Fix to not treat parameter memory refs as local
This fixes a bug that incorrectly deletes stores to parameters, which
can be used to return values from functions.
Pierre Moreau [Sun, 1 Oct 2017 19:47:44 +0000 (21:47 +0200)]
Compact-ids pass should update the header ID bound
David Neto [Sun, 1 Oct 2017 13:27:00 +0000 (09:27 -0400)]
DiagnosticStream move ctor moves output duties to new object
- Take over contents of the expiring message stream
- Prevent the expiring object from emitting anything during destruction
David Neto [Thu, 28 Sep 2017 15:14:00 +0000 (11:14 -0400)]
Cache end iterators for speed
Helps scaling of DefUseManager on modules with many thousands
of instructions.
jcaraban [Fri, 29 Sep 2017 05:55:41 +0000 (08:55 +0300)]
No use to check OpBitCount result width
Lei Zhang [Thu, 28 Sep 2017 01:22:05 +0000 (21:22 -0400)]
Skip checking copyright if SPIRV_SKIP_TESTS is enabled
David Neto [Wed, 27 Sep 2017 14:03:20 +0000 (10:03 -0400)]
Update CHANGES to reflect fix for #827
David Neto [Tue, 26 Sep 2017 22:05:27 +0000 (18:05 -0400)]
Compact-ids pass should update instruction's result_id member
Also update the result type field.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/827
Lei Zhang [Tue, 26 Sep 2017 17:55:23 +0000 (13:55 -0400)]
Remove duplicated declaration of CreateAggressiveDCEPass().
Andrey Tuganov [Mon, 18 Sep 2017 20:49:11 +0000 (16:49 -0400)]
Add bitwise operations validator pass
The pass checks correctness of operand types of all bitwise instructions
(opcode range from SpvOpShiftRightLogical to SpvOpBitCount).
Andrey Tuganov [Mon, 18 Sep 2017 17:06:40 +0000 (13:06 -0400)]
Add remaining opcodes to arithmetics validation
Add validation rules for:
- OpIAddCarry
- OpISubBorrow
- OpUMulExtended
- OpSMulExtended
Includes some refactoring of old code.
David Neto [Tue, 26 Sep 2017 15:30:57 +0000 (11:30 -0400)]
Update CHANGES with recent news
Steven Perron [Tue, 19 Sep 2017 14:12:13 +0000 (10:12 -0400)]
Create the dead function elimination pass
Creates a pass called eliminate dead functions that looks for functions
that could never be called, and deletes them from the module.
To support this change a new function was added to the Pass class to
traverse the call trees from diffent starting points.
Includes a test to ensure that annotations are removed when deleting a
dead function. They were not, so fixed that up as well.
Did some cleanup of the assembly for the test in pass_test.cpp. Trying
to make them smaller and easier to read.
Andrey Tuganov [Mon, 25 Sep 2017 18:55:44 +0000 (14:55 -0400)]
Detach MARK-V from the validator
MARK-V codec was previously dependent on the validation state.
Now it doesn't need the validator to function, but can still optionally
create it and validate every instruction once it's decoded.
Lei Zhang [Thu, 21 Sep 2017 21:24:57 +0000 (17:24 -0400)]
Avoid using global static variables
Previously we have several grammar tables defined as global static
variables and these grammar table entries contains non-POD struct
fields (CapabilitySet/ExtensionSet). The initialization of these
non-POD struct fields may require calling operator new. If used
as a library and the caller defines its own operator new, things
can screw up.
This pull request changes all global static variables into
function static variables, which is lazy evaluated in a thread
safe way as guaranteed by C++11.
Andrey Tuganov [Mon, 25 Sep 2017 16:45:48 +0000 (12:45 -0400)]
Add SPIRV_SPIRV_COMPRESSION option to cmake
The option is off by default.
cmake -DSPIRV_BUILD_COMPRESSION=ON ..
enables the compression lib, executable, and test build.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/834
Ehsan [Thu, 21 Sep 2017 22:50:09 +0000 (18:50 -0400)]
Merge pull request #829 from atgoo/fix_val_logicals
Validator: fix logicals pass for OpSelect pointers
Andrey Tuganov [Thu, 21 Sep 2017 20:12:14 +0000 (16:12 -0400)]
Validator: fix logicals pass for OpSelect pointers
OpSelect works with pointers also when capability
VariablePointersStorageBuffer is declared (before worked only with
capability VariablePointers).
David Neto [Wed, 20 Sep 2017 15:07:55 +0000 (11:07 -0400)]
elim-multi-store: only patch loop header phis that we created
There can already be OpPhi instructions in a loop header that
are unrelated to the optimization. We should not be patching those.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/826
Andrey Tuganov [Thu, 7 Sep 2017 21:27:57 +0000 (17:27 -0400)]
Add validate logicals pass to the validator
New pass checks operands of all instructions listed under
3.32.15. Relational and Logical Instructions
Andrey Tuganov [Mon, 18 Sep 2017 21:07:00 +0000 (17:07 -0400)]
Refactored validate_aritmetics.cpp
Improved error messages and readability.
Andrey Tuganov [Wed, 6 Sep 2017 20:44:15 +0000 (16:44 -0400)]
Updated markv_autogen
- now includes a table of all descriptors with coding scheme
(improves performance by 5% by allowing to avoid creation of
move-to-front sequences which will never be used)
- increased the size of markv_autogen.inc, clang doesn't seem
to have the long compilation time problem now
(probably was inadvertently fixed by using Huffman codec
serialization)
Greg Fischer [Tue, 19 Sep 2017 19:47:28 +0000 (13:47 -0600)]
ElimLocalMultiStore: Reset structured successors for each function
Steven Perron [Fri, 8 Sep 2017 16:08:03 +0000 (12:08 -0400)]
Add strength reduction; for now replace multiply by power of 2
Create a new optimization pass, strength reduction, which will replace
integer multiplication by a constant power of 2 with an equivalent bit
shift. More changes could be added later.
- Does not duplicate constants
- Adds vector |Concat| utility function to a common test header.
GregF [Wed, 13 Sep 2017 00:15:09 +0000 (18:15 -0600)]
ExtractInsert: Handle rudimentary CompositeConstruct and ConstantComposite
This optimizes a single index extract whose composite value terminates with a
CompositeConstruct (or ConstantComposite) by evaluating to the correct
component. This was needed for opaque legalization.
This highlights the need/opportunity to improve this optimization to deal
with more complex composite expressions including currently handled ops
plus Null ops and special vector composition. A TODO has been added.
David Neto [Thu, 14 Sep 2017 14:35:22 +0000 (10:35 -0400)]
Recognize SPV_AMD_shader_fragment_mask
Andrey Tuganov [Wed, 6 Sep 2017 18:30:27 +0000 (14:30 -0400)]
Add new checks to validate arithmetics pass
New operations:
- OpDot
- OpVectorTimesScalar
- OpMatrixTimesScalar
- OpVectorTimesMatrix
- OpMatrixTimesVector
- OpMatrixTimesMatrix
- OpOuterProduct
David Neto [Fri, 8 Sep 2017 13:16:15 +0000 (09:16 -0400)]
Update CHANGES for OpModuleProcessed validation rule
David Neto [Thu, 7 Sep 2017 21:38:31 +0000 (17:38 -0400)]
validator: OpModuleProcessed allowed in layout section 7c
Recent spec fix from SPIR Working group:
Allow OpModuleProcessed after debug names, but before any
annotation instructions.
Andrey Tuganov [Wed, 9 Aug 2017 18:01:12 +0000 (14:01 -0400)]
Update MARK-V to version 1.01
Includes:
- Multi-sequence move-to-front
- Coding by id descriptor
- Statistical coding of non-id words
- Joint coding of opcode and num_operands
Removed explicit form Huffman codec constructor
- The standard use case for it is to be constructed from initializer list.
Using serialization for Huffman codecs
David Neto [Fri, 1 Sep 2017 22:12:15 +0000 (18:12 -0400)]
spirv-as: Fail for unrecognized long option
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/798
David Neto [Sat, 2 Sep 2017 23:01:03 +0000 (19:01 -0400)]
Inliner: Fix LoopMerge when inline into loop header of multi block loop
This adapts the fix for the single-block loop. Split the loop like
before. But when we move the OpLoopMerge back to the loop header,
redirect the continue target only when the original loop was a single
block loop.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/800
Andrey Tuganov [Wed, 30 Aug 2017 14:13:10 +0000 (10:13 -0400)]
Add validation pass for arithmetic operations
The pass checks if arithmetic operations (such as OpFMul) receive
correct operands.
David Neto [Thu, 17 Aug 2017 17:53:14 +0000 (13:53 -0400)]
Make enums for all currently published extensions
Use the list from the SPIR-V registry page. Also, capture it as
a string so it's much easier to update via copy-paste.
The validator will accept modules that declare these known
extensions. However, we might not know about new tokens or
instructions declared in them. For that we need grammar updates
applied to SPIRV-Headers.
David Neto [Fri, 1 Sep 2017 20:28:22 +0000 (16:28 -0400)]
Starge v2017.1-dev
David Neto [Fri, 1 Sep 2017 20:24:27 +0000 (16:24 -0400)]
Create v2017.0
Update README to describe that we understand SPIR-V syntax based on the
grammar files included from the SPIRV-Headers repo.
(Also, it's high time we issue a v2017 release!
David Neto [Fri, 1 Sep 2017 20:14:25 +0000 (16:14 -0400)]
Finalize v2016.7
David Neto [Fri, 1 Sep 2017 20:13:12 +0000 (16:13 -0400)]
Update CHANGES for recent inliner fixes
Andrey Tuganov [Mon, 28 Aug 2017 22:36:52 +0000 (18:36 -0400)]
Fix mingw build (source/print.cpp)
source/print.cpp doesn't compile due to integer conversion.
Tested by @dneto0 on a Windows machine.
David Neto [Thu, 31 Aug 2017 21:33:44 +0000 (17:33 -0400)]
Inliner: Remap callee entry block id to single-trip loop header
Otherwise cloned phis can be invalid.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/790
David Neto [Thu, 31 Aug 2017 19:47:31 +0000 (15:47 -0400)]
Inline: Fix single-block loop caller cases
If the caller block is a single-block loop and inlining will
replace the caller block by several blocks, then:
- The original OpLoopMerge instruction will end up in the *last*
such block. That's the wrong place to put it.
- Move it back to the end of the first block.
- Update its Continue Target ID to point to the last block
We also have to take care of cases where the inlined code
begins with a structured header block. In this case
we need to ensure the restored OpLoopMerge does not appear
in the same block as the merge instruction from the callee's
first block.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/787
David Neto [Thu, 31 Aug 2017 15:37:17 +0000 (11:37 -0400)]
BasicBlock: add ctail, GetMergeInst, GetLoopMergeInst
David Neto [Fri, 1 Sep 2017 14:26:41 +0000 (10:26 -0400)]
Update tests for new preferred name as ShaderViewportIndexLayerEXT
This reacts to a recent update to SPIRV-Headers
David Neto [Fri, 1 Sep 2017 14:15:42 +0000 (10:15 -0400)]
Update CHANGES for fix to 781
Andrey Tuganov [Tue, 29 Aug 2017 20:25:23 +0000 (16:25 -0400)]
Extension allows multiple same OpTypePointer types
SPV_KHR_variable_pointers allows OpTypePointer to declare multiple
pointer identical types.
https://github.com/KhronosGroup/SPIRV-Tools/issues/781
David Neto [Wed, 30 Aug 2017 18:25:23 +0000 (14:25 -0400)]
Update CHANGES for fix to 776
GregF [Wed, 23 Aug 2017 23:05:38 +0000 (17:05 -0600)]
DeadBranchElim: Fix dead block detection to ignore backedges
- DeadBranchElim: Make sure to mark orphan'd merge blocks and continue
targets as live.
- Add test with loop in dead branch
- Add test that orphan'd merge block is handled.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/776
GregF [Tue, 29 Aug 2017 00:02:14 +0000 (18:02 -0600)]
InlineOpaque: Remove from usage until complete Opaque policy is designed.
David Neto [Wed, 30 Aug 2017 03:22:07 +0000 (23:22 -0400)]
Update changes for fix to issue 782
GregF [Wed, 30 Aug 2017 00:35:05 +0000 (18:35 -0600)]
Inline: Fix remapping of non-label forward references in callee phi
David Neto [Mon, 28 Aug 2017 14:30:58 +0000 (10:30 -0400)]
Disassembler: Print colour codes only when writing to a terminal
Andrey Tuganov [Mon, 28 Aug 2017 17:18:45 +0000 (13:18 -0400)]
Fix encode zero bits on word boundary bug
Bit stream writer was manifesting incorrect behaviour when the following
two conditions were met:
- writer was on 64-bit word boundary
- WriteBits was invoked with num_bits=0 (can happen when a Huffman codec has only one
value)
The bug was causing very rare sporadic corruption which was detected by
tests after a random experimental change in MARK-V model.
David Neto [Thu, 24 Aug 2017 22:11:23 +0000 (18:11 -0400)]
Show result id for CompositeInsert validation failure
GregF [Thu, 24 Aug 2017 22:36:53 +0000 (16:36 -0600)]
README: Add reference to SPIR-V size reduction white paper
David Neto [Thu, 24 Aug 2017 14:34:00 +0000 (10:34 -0400)]
Windows: Increase intensity of blue text
Lukas Hermanns [Sun, 20 Aug 2017 12:11:50 +0000 (14:11 +0200)]
Fix: background color was erroneously reset on Win32 platform.
Fix: background color was erroneously reset on Win32 platform.
David Neto [Wed, 23 Aug 2017 17:28:48 +0000 (13:28 -0400)]
Updated capabilites for SampleMask
SPIRV-Headers recently fixed the capability dependency
for SampleMask. It depends on Shader, not SampleRateShading
GregF [Tue, 15 Aug 2017 23:58:28 +0000 (17:58 -0600)]
Opt: Create InlineOpaquePass
Only inline calls to functions with opaque params or return
TODO: Handle parameter type or return type where the opqaue
type is buried within an array.
GregF [Thu, 10 Aug 2017 22:42:16 +0000 (16:42 -0600)]
Opt: Have "size" passes process full entry point call tree.
Includes code to deal correctly with OpFunctionParameter. This
is needed by opaque propagation which may not exhaustively inline
entry point functions.
Adds ProcessEntryPointCallTree: a method to do work on the
functions in the entry point call trees in a deterministic order.
Lei Zhang [Wed, 16 Aug 2017 20:42:50 +0000 (16:42 -0400)]
Support supplying extra definitions via CMAKE variable
Andrey Tuganov [Fri, 11 Aug 2017 20:51:24 +0000 (16:51 -0400)]
Huffman codec can serialize to text
Refactored the Huffman codec implementation and added ability to
serialize to C++-like text format. This would reduce the time-complexity
if loading hard-coded codecs.
GregF [Tue, 15 Aug 2017 21:54:41 +0000 (15:54 -0600)]
Opt: Add opaque tests
Andrey Tuganov [Wed, 2 Aug 2017 20:47:25 +0000 (16:47 -0400)]
Add id descriptor feature to SPIR-V
Id descriptors are computed as a recursive hash of all instructions used
to define an id. Descriptors are invarint of actual id values and
the similar code in different files would produce the same descriptors.
Multiple ids can have the same descriptor. For example
%1 = OpConstant %u32 1
%2 = OpConstant %u32 1
would produce two ids with the same descriptor. But
%3 = OpConstant %s32 1
%4 = OpConstant %u32 2
would have descriptors different from %1 and %2.
Descriptors will be used as handles of move-to-front sequences in SPIR-V
compression.
GregF [Fri, 4 Aug 2017 21:04:37 +0000 (15:04 -0600)]
ADCE: Add support for function calls
ADCE will now generate correct code in the presence of function calls.
This is needed for opaque type optimization needed by glslang. Currently
all function calls are marked as live. TODO: mark calls live only if they
write a non-local.
David Neto [Thu, 10 Aug 2017 17:13:33 +0000 (13:13 -0400)]
Update CHANGE to list fix for issue 755
David Neto [Wed, 9 Aug 2017 18:59:04 +0000 (14:59 -0400)]
Inliner: callee can have early return that isn't multi-return
Avoid generating an invalid OpLabel.
Create the continue target for the single-trip loop only if
you actually created the header for the single-trip loop.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/755
GregF [Fri, 4 Aug 2017 18:00:30 +0000 (12:00 -0600)]
AccessChainConvert: Add HasOnlySupportedRefs()
This avoids conversion on variables which will not ultimately be optimized.
Also removed an obsolete restriction from FindTargetVars(). Also added
decorates to supported refs (eg. RelaxedPrecision). Also fixed name to
IsNonTypeDecorate().
GregF [Tue, 1 Aug 2017 23:20:13 +0000 (17:20 -0600)]
Inline: Split out InlineExhaustivePass from InlinePass
GregF [Thu, 3 Aug 2017 23:27:02 +0000 (17:27 -0600)]
Mem2Reg: Allow Image and Sampler types as base target types.
David Neto [Fri, 4 Aug 2017 21:48:22 +0000 (17:48 -0400)]
Update CHANGES with common uniform loads pass
GregF [Thu, 3 Aug 2017 16:37:26 +0000 (10:37 -0600)]
Update README.md with new code-reduction passes
GregF [Mon, 3 Jul 2017 23:23:04 +0000 (17:23 -0600)]
Add CommonUniformElim pass
- UniformElim: Only process reachable blocks
- UniformElim: Don't reuse loads of samplers and images across blocks.
Added a second phase which only reuses loads within a block for samplers
and images.
- UniformElim: Upgrade CopyObject skipping in GetPtr
- UniformElim: Add extensions whitelist
Currently disallowing SPV_KHR_variable_pointers because it doesn't
handle extended pointer forms.
- UniformElim: Do not process shaders with GroupDecorate
- UniformElim: Bail on shaders with non-32-bit ints.
- UniformElim: Document support for only single index and add TODO.
GregF [Thu, 27 Jul 2017 19:30:12 +0000 (13:30 -0600)]
Add MemPass, move all shared functions to it.
Andrey Tuganov [Mon, 24 Jul 2017 18:29:35 +0000 (14:29 -0400)]
Add multi-sequence move-to-front implementation
Add MultiMoveToFront class which supports multiple move-to-front
sequences and allows to promote value in all sequences at once.
Added caching for last accessed sequence handle and last accessed value
in each sequence.
David Neto [Tue, 1 Aug 2017 19:33:15 +0000 (15:33 -0400)]
Update CHANGES with recently added optimizations
GregF [Tue, 1 Aug 2017 18:55:29 +0000 (12:55 -0600)]
Opt: Add new size-reduction passes to usage message.
Andrey Tuganov [Mon, 31 Jul 2017 17:08:38 +0000 (13:08 -0400)]
Added C++ code generation to spirv-stats
The tool can now generate C++ code returning some of the historgrams and
Huffman codecs generated from those historgrams.
David Neto [Tue, 1 Aug 2017 15:42:28 +0000 (11:42 -0400)]
Add Appveyor config that uses VS 2017
GregF [Wed, 26 Jul 2017 22:34:41 +0000 (18:34 -0400)]
Opt: Delete names and decorations of dead instructions
Lei Zhang [Mon, 24 Jul 2017 15:43:47 +0000 (11:43 -0400)]
Opt: HasOnlySupportedRefs should consider OpCopyObject
This fixes test failure after merging the previous pull request.
Lei Zhang [Mon, 24 Jul 2017 14:37:38 +0000 (10:37 -0400)]
Revert "Revert "Opt: LocalBlockElim: Add HasOnlySupportedRefs""
This reverts commit
df96e243c633b939690a42d89904e04f50126e86.
GregF [Wed, 19 Jul 2017 00:57:26 +0000 (18:57 -0600)]
Add extension whitelists to size-reduction passes.
Currently only SPV_KHR_variable_pointers is disallowed in passes which
do pointer analysis. Positive and negative tests of the general extensions
mechanism were added to aggressive_dce but cover all passes.
Lei Zhang [Sat, 22 Jul 2017 14:48:28 +0000 (10:48 -0400)]
Revert "Opt: LocalBlockElim: Add HasOnlySupportedRefs"
This reverts commit
2d0f7fbc1151004c4cf111b947f1070a03dfe092.
greg-lunarg [Sat, 22 Jul 2017 14:32:19 +0000 (08:32 -0600)]
Opt: LocalBlockElim: Add HasOnlySupportedRefs
Verifies that targeted variables have only access chain and direct
loads and stores as references.
GregF [Tue, 18 Jul 2017 20:42:51 +0000 (14:42 -0600)]
Fix handling of CopyObject in GetPtr and its call sites
Lenny Komow [Thu, 13 Jul 2017 17:13:14 +0000 (11:13 -0600)]
Fix Visual Studio size_t cast compiler warning
Visual Studio was complaining about possible loss of data on 64-bit
builds, due to an implicit cast from size_t to int. This changes the
data to use an int with no cast.
Greg Fischer [Tue, 11 Jul 2017 22:52:19 +0000 (16:52 -0600)]
LocalMultiStore: Always put varId for backedge on loop phi function.
And always patch the backedge operand when patching phi functions. This
approach is more correct and cleaner. The previous code was generating
incorrect phis when the backedge block had no predecessors.
GregF [Mon, 10 Jul 2017 23:20:35 +0000 (17:20 -0600)]
DeadBranchElim: Improve algorithm to only remove blocks with no predecessors
Must be careful not to remove blocks pointed at by unreachable blocks
David Neto [Mon, 10 Jul 2017 15:45:59 +0000 (11:45 -0400)]
Minor code review feedback on AggressiveDCE
GregF [Thu, 8 Jun 2017 16:37:21 +0000 (10:37 -0600)]
Add AggressiveDCEPass
Create aggressive dead code elimination pass
This pass eliminates unused code from functions. In addition,
it detects and eliminates code which may have spurious uses but which do
not contribute to the output of the function. The most common cause of
such code sequences is summations in loops whose result is no longer used
due to dead code elimination. This optimization has additional compile
time cost over standard dead code elimination.
This pass only processes entry point functions. It also only processes
shaders with logical addressing. It currently will not process functions
with function calls. It currently only supports the GLSL.std.450 extended
instruction set. It currently does not support any extensions.
This pass will be made more effective by first running passes that remove
dead control flow and inlines function calls.
This pass can be especially useful after running Local Access Chain
Conversion, which tends to cause cycles of dead code to be left after
Store/Load elimination passes are completed. These cycles cannot be
eliminated with standard dead code elimination.
Additionally: This transform uses a whitelist of instructions that it
knows do have side effects, (a.k.a. combinators). It assumes other
instructions have side effects: it will not remove them, and assumes
they have side effects via their ID operands.
GregF [Fri, 16 Jun 2017 21:37:31 +0000 (15:37 -0600)]
Add LocalMultiStoreElim pass
A SSA local variable load/store elimination pass.
For every entry point function, eliminate all loads and stores of function
scope variables only referenced with non-access-chain loads and stores.
Eliminate the variables as well.
The presence of access chain references and function calls can inhibit
the above optimization.
Only shader modules with logical addressing are currently processed.
Currently modules with any extensions enabled are not processed. This
is left for future work.
This pass is most effective if preceeded by Inlining and
LocalAccessChainConvert. LocalSingleStoreElim and LocalSingleBlockElim
will reduce the work that this pass has to do.
GregF [Fri, 2 Jun 2017 19:23:20 +0000 (13:23 -0600)]
DeadBranchElim: Add DeadBranchElimPass
David Neto [Tue, 4 Jul 2017 20:24:46 +0000 (16:24 -0400)]
Include memory and semantics IDs when iterating over inbound IDs
Fixes Instruction::ForEachInId so it covers
SPV_OPERAND_TYPE_MEMORY_SEMANTICS_ID and SPV_OPERAND_TYPE_SCOPE_ID.
Future proof a bit by using the common spvIsIdType routine.
Fixes https://github.com/KhronosGroup/SPIRV-Tools/issues/697