Benoit Jacob d8a656ffaf [MLIR] AMDGPUToROCDL: Use a bitcast op to reintepret a vector of i8 as single integer. (#111400) ...

Found by inspecting AMDGPU assembly - so the arithmetic ops created
there were definitely making their way into the target ISA. A
`LLVM::BitcastOp` seems equivalent, and evaporates as expected in the
target asm.

Along the way, I thought that this helper function `mfmaConcatIfNeeded`
could be renamed to `convertMFMAVectorOperand` to better convey its
contract; so I don't need to think about whether a bitcast is a
legitimate "concat" :-)

---------

Signed-off-by: Benoit Jacob <jacob.benoit.1@gmail.com>

2024-10-07 14:14:18 -04:00

..

AMDGPUToROCDL.cpp

[MLIR] AMDGPUToROCDL: Use a bitcast op to reintepret a vector of i8 as single integer. (#111400)

2024-10-07 14:14:18 -04:00

CMakeLists.txt

…