Instead of always materializing a new sparse tensor after reshape, this patch tries to fuses the reshape (currently only on COO) with GenericOp and coiterates with the reshaped tensors without allocating a new sparse tensor.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D145016
Rewrite a NewOp into a NewOp of a sorted COO tensor and a ConvertOp for
converting the sorted COO tensor to the destination tensor type.
Codegen a NewOp of a sorted COO tensor to use the new bulk reader API and sort
the elements only when the input is not sorted.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D144504
While dense tensors support random accesses, it is critical to visit them in a row-major order for better cache locality. However, we previously consider dense inputs and outputs together when computing constraints for building iteration graph, it could lead us to less efficient iteration graphs.
This patch adds a new `SortMask::kIncludeDenseInput` to treat dense inputs/outputs separately when building iteration graph, thus increasing the chance for use to construct a better iteration graph.
A more fine-grained approach is to treat each input separately.
Note, related to:
https://github.com/llvm/llvm-project/issues/51651
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D144932
Eliminates the sort seems make the whole conversion slower (probably because loop rotation leads to bad locality).
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D144517
No need for a temp COO and sort even when converting dense -> CSC, we can instead rotate the loop to yield a ordered coordinates at beginning.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D144213
We will support symmetric MTX without expanding the data in the sparse tensor
storage.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D144059
* Flattening/simplifying some nested conditionals
* const-ifying some local variables
Depends On D143800
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D143949
This change adds a new `SparseTensorType` class for making the "dim" vs "lvl" distinction more overt, and for abstracting over the differences between sparse-tensors and dense-tensors. In addition, this change also adds new type aliases `Dimension`, `Level`, and `FieldIndex` to make code more self-documenting.
Although the diff is very large, the majority of the changes are mechanical in nature (e.g., changing types to use the new aliases, updating variable names to match, etc). Along the way I also made many variables `const` when they could be; the majority of which required only adding the keyword. A few places had conditional definitions of these variables, requiring actual code changes; however, that was only done when the overall change was extremely local and easy to extract. All these changes are included in the current patch only because it would be too onerous to split them off into a separate patch.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D143800
UnpackOp Converter used to create reallocOp unconditionally, but it might cause issue when the requested memory size is smaller than the actually storage.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D144065
Previously, when performing a reduction on a sparse tensor, the result
would be different depending on iteration order. For expanded access pattern,
an empty row would contribute no entry in the output. For lex ordering, the
identity would end up in the output.
This code changes that behavior and keeps track of whether any entries were
actually reduced in lex ordering, making the output consistent between the
two iteration styles.
Differential Revision: https://reviews.llvm.org/D142050
`getAliasingOpOperands`/`getAliasingOpResults` now encodes OpOperand/OpResult, buffer relation and a degree of certainty. E.g.:
```
// aliasingOpOperands(%r) = {(%t, EQUIV, DEFINITE)}
// aliasingOpResults(%t) = {(%r, EQUIV, DEFINITE)}
%r = tensor.insert %f into %t[%idx] : tensor<?xf32>
// aliasingOpOperands(%r) = {(%t0, EQUIV, MAYBE), (%t1, EQUIV, MAYBE)}
// aliasingOpResults(%t0) = {(%r, EQUIV, MAYBE)}
// aliasingOpResults(%t1) = {(%r, EQUIV, MAYBE)}
%r = arith.select %c, %t0, %t1 : tensor<?xf32>
```
`BufferizableOpInterface::bufferRelation` is removed, as it is now part of `getAliasingOpOperands`/`getAliasingOpResults`.
This change allows for better analysis, in particular wrt. equivalence. This allows additional optimizations and better error checking (which is sometimes overly conservative). Examples:
* EmptyTensorElimination can eliminate `tensor.empty` inside `scf.if` blocks. This requires a modeling of equivalence: It is not a per-OpResult property anymore. Instead, it can be specified for each OpOperand and OpResult. This is important because `tensor.empty` may be eliminated only if all values on the SSA use-def chain to the final consumer (`tensor.insert_slice`) are equivalent.
* The detection of "returning allocs from a block" can be improved. (Addresses a TODO in `assertNoAllocsReturned`.) This allows us to bufferize IR such as "yielding a `tensor.extract_slice` result from an `scf.if` branch", which currently fails to bufferize because the alloc detection is too conservative.
* Better bufferization of loops. Aliases of the iter_arg can be yielded (even if they are not equivalent) without having to realloc and copy the entire buffer on each iteration.
The above-mentioned examples are not yet implemented with this change. This change just improves the BufferizableOpInterface, its implementations and related helper functions, so that better aliasing information is available for each op.
Differential Revision: https://reviews.llvm.org/D142129
This adds the hint to a number of tensor allocations in codegens,
shaving off quite some time from e.g. reading in sparse matrices
due to zero-reallocation scheme. Note that we can probably provide
hints on all allocations, and refine the heuristics that use them
for general tensors.
Reviewed By: bixia
Differential Revision: https://reviews.llvm.org/D143309
Even though we introduced the size_hint, we never used it.
This is a very first step, using the hint during the codegen path.
Note that we can refine the heuristics. Also, we need to start
adding the hint on all allocation generated for reading tensors,
converting tensors, etc.
Reviewed By: Peiming, bixia
Differential Revision: https://reviews.llvm.org/D143292
in particular, the trailing COO optimization was not
desribed in the general layout description
Reviewed By: Peiming
Differential Revision: https://reviews.llvm.org/D143284
Currently `TypedValue` can be constructed directly from `Value`, hiding
errors that could be caught at compile time. For example the following
will compile, but crash/assert at runtime:
```
void foo(TypedValue<IntegerType>);
void bar(TypedValue<FloatType> v) {
foo(v);
}
```
This change removes the constructors and replaces them with explicit
llvm casts.
Depends on D142852
Reviewed By: rriddle
Differential Revision: https://reviews.llvm.org/D142855
* `getAliasingOpOperand` => `getAliasingOpOperands`
* `getAliasingOpResult` => `getAliasingOpResults`
Also a few minor code cleanups and better documentation.
Differential Revision: https://reviews.llvm.org/D142979
This is to prepare for implementing a hybrid quick sort, which switches to heap
sort when the recursive depth exceeds certain limits.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D142731
The name of the method was confusing. It is bufferizesToMemoryWrite, but from the perspective of OpResults.
`bufferizesToMemoryWrite(OpResult)` now supports ops with regions that do not have aliasing OpOperands (such as `scf.if`). These ops no longer need to implement `isMemoryWrite`.
Differential Revision: https://reviews.llvm.org/D141684
Currently, all the non-stable sorting algorithms are implemented via the
straightforward quick sort. This will be fixed in the following PR.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D142678
Previously, we choose the value at (lo + hi)/2 as a pivot for partitioning the
data in [lo, hi). We now choose the median for the three values at lo, (lo +
hi)/2, and (hi-1) as a pivot to match the std::qsort implementation.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D142679
This patch moves some utils into CodegenEnv class, it should make the code easier to follow and it eliminates several indirect value assignment that use `ptr**`.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D142040
The handling of unknown ops will be tightened in a subsequent change. All sparse_tensor ops should implement BufferizableOpInterface, otherwise, they are treated as "unknown" and additional buffer allocs/copies may be inserted around them.
Differential Revision: https://reviews.llvm.org/D142005
The bulk of D142074 seems to have gotten overwritten due to some sort of merge conflict (afaict there's no record of it having been reverted intentionally). So this commit redoes those changes. In addition to the original changes, this commit also:
* moves the definition of `getRankedTensorType` (from `Transforms/CodegenUtils.h` to `IR/SparseTensor.h`), so that it can be used by `IR/SparseTensorDialect.cpp`.
* adds `getMemRefType` as another abbreviation.
Reviewed By: aartbik
Differential Revision: https://reviews.llvm.org/D142503
Use SparseTensorDescriptor whenever not calling setters, to avoid needing to create a temporal buffer for simple query purposes.
Reviewed By: bixia, wrengr
Differential Revision: https://reviews.llvm.org/D141953
Previously, we rely on InsertOp to add values to the result, in the same way we
add values to a sparse tensor with compressed dimensions. We now direct store
values to the values buffer.
Reviewed By: Peiming
Differential Revision: https://reviews.llvm.org/D141517
