shylie/llvm-project
1
0
Fork 0
Code Issues Pull Requests Actions 6 Packages Projects Releases Wiki Activity
495,344 Commits 1,713 Branches 298 Tags
Commit Graph

10 Commits

Author SHA1 Message Date
Yinying Li
c5a67e16b6
[mlir][sparse] Use variable instead of inlining sparse encoding (#72561) 
Example:

#CSR = #sparse_tensor.encoding<{
  map = (d0, d1) -> (d0 : dense, d1 : compressed),
}>

// CHECK: #[[$CSR.*]] = #sparse_tensor.encoding<{ map = (d0, d1) -> (d0
: dense, d1 : compressed) }>
// CHECK-LABEL: func private @sparse_csr(
// CHECK-SAME: tensor<?x?xf32, **#[[$CSR]]**>)
func.func private @sparse_csr(tensor<?x?xf32, #CSR>)
 2023-11-16 19:30:21 -05:00
Aart Bik
5f32bcfbae
[mlir][sparse][gpu] re-enable all GPU libgen tests (#72185) 
Previous change no longer properly used the GPU libgen pass (even though
most tests still passed falling back to CPU). This revision puts the
proper pass order into place. Also bit of a cleanup of CPU codegen vs.
libgen setup.
 2023-11-14 09:06:15 -08:00
Yinying Li
3dc621124f
[mlir][sparse] Migrate tests to use new syntax (#66543) 
**COO**
`lvlTypes = [ "compressed_nu", "singleton" ]` to `map = (d0, d1) -> (d0
: compressed(nonunique), d1 : singleton)`
`lvlTypes = [ "compressed_nu_no", "singleton_no" ]` to `map = (d0, d1)
-> (d0 : compressed(nonunique, nonordered), d1 : singleton(nonordered))`

**SortedCOO**
`lvlTypes = [ "compressed_nu", "singleton" ]` to `map = (d0, d1) -> (d0
: compressed(nonunique), d1 : singleton)`

**BCOO**
`lvlTypes = [ "dense", "compressed_hi_nu", "singleton" ]` to `map = (d0,
d1, d2) -> (d0 : dense, d1 : compressed(nonunique, high), d2 :
singleton)`

**BCSR**
`lvlTypes = [ "compressed", "compressed", "dense", "dense" ], dimToLvl =
affine_map<(d0, d1) -> (d0 floordiv 2, d1 floordiv 3, d0 mod 2, d1 mod
3)>` to
`map = ( i, j ) ->
      ( i floordiv 2 : compressed,
        j floordiv 3 : compressed,
        i mod 2 : dense,
        j mod 3 : dense
      )`

**Tensor and other supported formats(e.g. CCC, CDC, CCCC)**

Currently, ELL and slice are not supported yet in the new syntax and the
CHECK tests will be updated once printing is set to output the new
syntax.

Previous PRs: #66146, #66309, #66443
 2023-09-15 16:12:20 -04:00
Yinying Li
51ebecf309 [mlir][sparse] Changed sparsity properties to use _ instead of - 
Example: compressed-no -> compressed_no

Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D158567
 2023-08-23 17:00:27 +00:00
Kun Wu
be2dd22b8f [mlir][sparse][gpu] reuse CUDA environment handle throughout instance lifetime 
Differential Revision: https://reviews.llvm.org/D153173
 2023-06-30 21:52:34 +00:00
Kun Wu
97f4c22b3a [mlir][sparse][gpu] unify dnmat and dnvec handle and ops 
Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D152465
 2023-06-09 17:16:48 +00:00
Kun Wu
8ed59c53de [mlir][sparse][gpu] add sm8.0+ tensor core 2:4 sparsity support 
Reviewed By: aartbik

Differential Revision: https://reviews.llvm.org/D151775
 2023-06-06 23:13:21 +00:00
Aart Bik
bcb698bfdc [mlir][sparse][gpu] various cuSparse refinements 
(1) keep all cuSparse ops on single stream without wait() in right order
(2) use more type precise memref types for COO
(3) use ToTensor on resulting memref (even though it folds away again)

Reviewed By: K-Wu

Differential Revision: https://reviews.llvm.org/D151404
 2023-05-24 22:32:52 -07:00
wren romano
7f5fb90bbb [mlir][sparse] Fixing GPU tests (followup to D150330) 
The GPU tests weren't updated when rebasing D150330, so this patch fixes that.

Reviewed By: anlunx

Differential Revision: https://reviews.llvm.org/D150822
 2023-05-17 15:29:54 -07:00
Aart Bik
ee42e23614 [mlir][sparse][gpu] first implementation of the GPU libgen approach 
The sparse compiler now has two prototype strategies for GPU acceleration:

* CUDA codegen: this converts sparsified code to CUDA threads
* CUDA libgen: this converts pre-sparsified code to cuSPARSE library calls

This revision introduces the first steps required for the second approach.

Reviewed By: ThomasRaoux

Differential Revision: https://reviews.llvm.org/D150170
 2023-05-15 08:49:38 -07:00