diff options
Diffstat (limited to 'mlir/g3doc/Rationale.md')
| -rw-r--r-- | mlir/g3doc/Rationale.md | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/mlir/g3doc/Rationale.md b/mlir/g3doc/Rationale.md index bafd029ce0f..8b22e93598c 100644 --- a/mlir/g3doc/Rationale.md +++ b/mlir/g3doc/Rationale.md @@ -568,9 +568,9 @@ func @search_body(%A: memref<?x?xi32>, %S: memref<?xi32>, %key: i32) { As per the [MLIR spec](LangRef.md), the restrictions on dimensions and symbol identifiers to be used with the affine.apply instruction only apply to accesses -inside `affine.for` and `if` instructions. However, an analysis of accesses -inside the called function (`@search_body`) is necessary to determine if the -`%i` loop could be parallelized: such function access analysis is calling +inside `affine.for` and `affine.if` instructions. However, an analysis of +accesses inside the called function (`@search_body`) is necessary to determine +if the `%i` loop could be parallelized: such function access analysis is calling context sensitive. ### Non-affine loop bounds {#non-affine-loop-bounds} @@ -665,7 +665,7 @@ func @conv2d(memref<16x1024x1024x3xf32, #lm0, vmem> %input, %h_pad_low, %w_pad_low] // Check if access is not in padding. - if #domain(%1_0, %1_1) + affine.if #domain(%1_0, %1_1) [%h_base_dilation, %w_kernel_dilation, %h_bound, %w_bound] { %2_0 = affine.apply #map2 (%1_0, %1_1) %2_1 = affine.apply #map2 (%1_0, %1_1) @@ -899,10 +899,10 @@ func @dma_hbm_to_vmem(memref<1024 x f32, #layout_map0, hbm> %a, representation. 2(b) requires no change, but impacts how cost models look at index and layout maps. -### `if` and `affine.for` Extensions for "Escaping Scalars" {#extensions-for-"escaping-scalars"} +### `affine.if` and `affine.for` Extensions for "Escaping Scalars" {#extensions-for-"escaping-scalars"} We considered providing a representation for SSA values that are live out of -`if/else` conditional bodies and loop carried in `affine.for` loops. We +`affine.if/else` conditional bodies and loop carried in `affine.for` loops. We ultimately abandoned this approach due to its complexity. In the current design of MLIR, scalar variables cannot escape for loops or if instructions. In situations, where escaping is necessary, we use zero-dimensional tensors and @@ -948,7 +948,7 @@ func int32 @sum(%A : memref<?xi32>, %N : i32) -> (i32) { Syntax: ``` {.ebnf} -<out-var-list> = if (<cond-list>) {...} [else {...}] +<out-var-list> = affine.if (<cond-list>) {...} [else {...}] ``` Out-var-list is a list of SSA values defined by the if-instruction. The values @@ -965,7 +965,7 @@ Example: func int32 @sum_half(%A, %N) { %s0 = 0 %s1 = affine.for %i = 1 ... N step 1 with %s2 (%s0) { - %s3 = if (%i >= %N / 2) { + %s3 = affine.if (%i >= %N / 2) { %v0 = load %A[%i] %s4 = %s2 + %v0 yield %s4 |
