Type conversion rule for single element vector and materialization
function to support the conversion has a mismatch.
Update materialization function to match the type conversion rule.
During the XeGPU-to-XeVM type conversion, a memref is lowered to its
base address. This PR extends the conversion to correctly handle memrefs
that include an offset, such as those generated by memref.subview.
Some usage case or shapes for 2D block op with sub byte types can be
emulated with 2D block operations for non-sub byte types. Add sub byte
type i4 as a valid XeGPU type. And add lowering of certain 2D
block operations by emulating with larger element types.
Current lowering pattern for create_nd_tdesc restricts source memref to
static shape.
In case of a dynamic ranked memref, create_nd_tdesc already provides
shape as an argument.
Lowering can use those values instead of returning a mismatch error.
This PR adds lowering of xegpu.load_matrix/store_matrix to
xevm.blockload/blockstore or and llvm.load/store, depending on wi level
attributes.
It includes a few components:
1. adds wi-level attributes: subgroup_block_io.
2. expand load_matrix/store_matrix op definition to support scalar data
(besides vector data).
2. adds a member function to mem_desc to compute the linearized address
for a nd offsets.
3. add lowering depending on wi-level attributes:
a) if subgroup_block_io attribute presents, lower to
xevm.blockload/blockstore
c) else lower to llvm.load/store. If result is a vector, lower to
llvm.load/store with vector operand.
Use wrappers around `std::accumulate` to make the code more concise and
less bug-prone: https://github.com/llvm/llvm-project/pull/162129.
With `std::accumulate`, it's the initial value that determines the
accumulator type. `llvm::sum_of` and `llvm::product_of` pick the right
accumulator type based on the range element type.
Found some funny bugs like a local accumulate helper that calculated a
sum with initial value of 1 -- we didn't hit the bug because the code
was actually dead...
Fixes two issue with XeGPU to XeVM pass
1. xegpu.update_nd_offset op lower generated incorrect code sequence
2. xegpu.store_nd did not lower single element vector
