[Mlir-commits] [mlir] [mlir][sparse] Expand LevelType to 64 bits and implement parsing n out of m (PR #79935)

[Aart Bik]

================
@@ -145,7 +145,7 @@ def SparseTensorEncodingAttr : SparseTensor_Attr<"SparseTensorEncoding",
     - **compressed** : only nonzeros along this level are stored
     - **loose_compressed** : as compressed, but allows for free space between regions
     - **singleton** : a variant of the compressed format, where coordinates have no siblings
-    - **block2_4** : the compression uses a 2:4 encoding per 1x4 block
+    - **block[n, m]** : the compression uses a n:m encoding per 1xm block
----------------
aartbik wrote:

the 1xm block actually implies a 2-dim envelop at least
perhaps we should just say

uses a n:m encoding (viz. n out of m consecutive elements are nonzero)

https://github.com/llvm/llvm-project/pull/79935