[Mlir-commits] [mlir] [mlir][tosa] Support more float types in resource transpose folding (PR #213226)

[Luke Hutton]

================
@@ -231,6 +254,19 @@ DenseElementsAttr transpose(ElementsAttr attr, ShapedType inputType,
     if (auto data = tryGetDenseResourceValues<float>(attr);
         data && elementTy.isF32())
       return transposeType(*data, inputType, outputType, permValues);
+
+    if (auto data = tryGetDenseResourceValues<uint8_t>(attr);
+        data && isa<Float4E2M1FNType, Float8E4M3FNType, Float8E5M2Type,
+                    Float8E8M0FNUType>(elementTy))
+      return transposeRawType(*data, inputType, outputType, permValues);
----------------
lhutton1 wrote:

This appears to assume supplied data is unpacked for sub-byte types, I'm not sure if that's always the case for dense resource attrs?

Slightly tangental: perhaps it's safer to go through APFloat, similar to https://github.com/llvm/llvm-project/pull/213226/changes#diff-e3e29fb7d252978e2d0330ba35e34893bce21bf009f4ab1227700152db7f8b88R227?

https://github.com/llvm/llvm-project/pull/213226