Skip to content

AutoBatching: batch through a nested loop whose trip count the data gives #3437

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
wsmoses merged 1 commit into from
Oct 8, 2026
Merged

AutoBatching: batch through a nested loop whose trip count the data gives #3437

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
11 changes: 6 additions & 5 deletions src/enzyme_ad/jax/Passes/AutoBatching.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3043,9 +3043,10 @@ namespace {
// Batches the body of an enzymexla.parallel while over its iterations: a value
// that varies with the iteration gains a leading dimension of the trip count,
// a write into a carried buffer becomes one scatter of every iteration's
// write, and a constant-trip loop nested in the body keeps running, over the
// batched values (the iterations of the parallel loop are independent, so it
// and the parallel loop interchange).
// write, and a loop nested in the body whose trip count is the same for
// every iteration keeps running, over the batched values (the iterations of
// the parallel loop are independent, so it and the parallel loop
// interchange).
struct ParallelWhileBatcher {
PatternRewriter &rewriter;
Location loc;
Expand Down Expand Up @@ -3165,8 +3166,8 @@ struct ParallelWhileBatcher {
// iteration of the parallel loop, over values that may vary with it.
LogicalResult analyzeWhile(stablehlo::WhileOp w) {
enzyme::WhileLoopInfo wi(w);
if (failed(wi.computeInfo()) || !wi.isValid() || !wi.isConstant() ||
!isMemoryEffectFree(w))
if (failed(wi.computeInfo()) || !wi.isValid() || !wi.isConstantStart() ||
!wi.isConstantStep() || !isMemoryEffectFree(w))
return failure();
Value wiv = wi.getInductionVariable();
if (!wiv)
Expand Down
105 changes: 105 additions & 0 deletions test/lit_tests/parallel_while_invariant_trip.mlir
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,105 @@
// RUN: enzymexlamlir-opt %s --enzyme-hlo-generate-td="patterns=parallel_while_to_batched_scatter" --transform-interpreter --enzyme-hlo-remove-transform | FileCheck %s

// A CSR transpose padded to its longest row: the nested loop runs to a
// trip count read from the data, the same for every row, so it keeps
// running over the batched rows.
// M = max_i off[i+1] - off[i]; for i: acc = 0; for k < M: j = off[i] + k;
// if j < off[i+1]: acc += x[idx[j]]; y[i] = acc
func.func @csr_padded(%off: tensor<9xi32>, %idx: tensor<40xi32>, %x: tensor<40xf64>, %y: tensor<8xf64>) -> tensor<8xf64> {
%c0 = stablehlo.constant dense<0> : tensor<i64>
%c1 = stablehlo.constant dense<1> : tensor<i64>
%c8 = stablehlo.constant dense<8> : tensor<i64>
%min = stablehlo.constant dense<-2147483648> : tensor<i32>
%zero = stablehlo.constant dense<0.0> : tensor<f64>
%lo = stablehlo.slice %off [0:8] : (tensor<9xi32>) -> tensor<8xi32>
%hi = stablehlo.slice %off [1:9] : (tensor<9xi32>) -> tensor<8xi32>
%len = stablehlo.subtract %hi, %lo : tensor<8xi32>
%m = stablehlo.reduce(%len init: %min) applies stablehlo.maximum across dimensions = [0] : (tensor<8xi32>, tensor<i32>) -> tensor<i32>
%m64 = stablehlo.convert %m : (tensor<i32>) -> tensor<i64>
%0:2 = stablehlo.while(%i = %c0, %buf = %y) : tensor<i64>, tensor<8xf64> attributes {enzymexla.parallel}
cond {
%c = stablehlo.compare LT, %i, %c8 : (tensor<i64>, tensor<i64>) -> tensor<i1>
stablehlo.return %c : tensor<i1>
} do {
%l = stablehlo.dynamic_slice %off, %i, sizes = [1] : (tensor<9xi32>, tensor<i64>) -> tensor<1xi32>
%ls = stablehlo.reshape %l : (tensor<1xi32>) -> tensor<i32>
%i1 = stablehlo.add %i, %c1 : tensor<i64>
%h = stablehlo.dynamic_slice %off, %i1, sizes = [1] : (tensor<9xi32>, tensor<i64>) -> tensor<1xi32>
%hs = stablehlo.reshape %h : (tensor<1xi32>) -> tensor<i32>
%1:2 = stablehlo.while(%k = %c0, %acc = %zero) : tensor<i64>, tensor<f64>
cond {
%c = stablehlo.compare LT, %k, %m64 : (tensor<i64>, tensor<i64>) -> tensor<i1>
stablehlo.return %c : tensor<i1>
} do {
%k32 = stablehlo.convert %k : (tensor<i64>) -> tensor<i32>
%j = stablehlo.add %ls, %k32 : tensor<i32>
%in = stablehlo.compare LT, %j, %hs, SIGNED : (tensor<i32>, tensor<i32>) -> tensor<i1>
%j64 = stablehlo.convert %j : (tensor<i32>) -> tensor<i64>
%jr = stablehlo.reshape %j64 : (tensor<i64>) -> tensor<1xi64>
%col = "stablehlo.gather"(%idx, %jr) <{dimension_numbers = #stablehlo.gather<collapsed_slice_dims = [0], start_index_map = [0]>, indices_are_sorted = false, slice_sizes = array<i64: 1>}> : (tensor<40xi32>, tensor<1xi64>) -> tensor<i32>
%col64 = stablehlo.convert %col : (tensor<i32>) -> tensor<i64>
%cr = stablehlo.reshape %col64 : (tensor<i64>) -> tensor<1xi64>
%v = "stablehlo.gather"(%x, %cr) <{dimension_numbers = #stablehlo.gather<collapsed_slice_dims = [0], start_index_map = [0]>, indices_are_sorted = false, slice_sizes = array<i64: 1>}> : (tensor<40xf64>, tensor<1xi64>) -> tensor<f64>
%vm = stablehlo.select %in, %v, %zero : tensor<i1>, tensor<f64>
%acc2 = stablehlo.add %acc, %vm : tensor<f64>
%kn = stablehlo.add %k, %c1 : tensor<i64>
stablehlo.return %kn, %acc2 : tensor<i64>, tensor<f64>
}
%ir = stablehlo.reshape %i : (tensor<i64>) -> tensor<1xi64>
%b2 = "stablehlo.scatter"(%buf, %ir, %1#1) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<inserted_window_dims = [0], scatter_dims_to_operand_dims = [0]>, unique_indices = true}> ({
^bb0(%p: tensor<f64>, %q: tensor<f64>):
stablehlo.return %q : tensor<f64>
}) : (tensor<8xf64>, tensor<1xi64>, tensor<f64>) -> tensor<8xf64>
%in2 = stablehlo.add %i, %c1 : tensor<i64>
stablehlo.return %in2, %b2 : tensor<i64>, tensor<8xf64>
}
return %0#1 : tensor<8xf64>
}

// CHECK: func.func @csr_padded(%arg0: tensor<9xi32>, %arg1: tensor<40xi32>, %arg2: tensor<40xf64>, %arg3: tensor<8xf64>) -> tensor<8xf64> {
// CHECK-NEXT: %c = stablehlo.constant dense<0> : tensor<i64>
// CHECK-NEXT: %c_0 = stablehlo.constant dense<1> : tensor<i64>
// CHECK-NEXT: %c_1 = stablehlo.constant dense<-2147483648> : tensor<i32>
// CHECK-NEXT: %cst = stablehlo.constant dense<0.000000e+00> : tensor<f64>
// CHECK-NEXT: %0 = stablehlo.slice %arg0 [0:8] : (tensor<9xi32>) -> tensor<8xi32>
// CHECK-NEXT: %1 = stablehlo.slice %arg0 [1:9] : (tensor<9xi32>) -> tensor<8xi32>
// CHECK-NEXT: %2 = stablehlo.subtract %1, %0 : tensor<8xi32>
// CHECK-NEXT: %3 = stablehlo.reduce(%2 init: %c_1) applies stablehlo.maximum across dimensions = [0] : (tensor<8xi32>, tensor<i32>) -> tensor<i32>
// CHECK-NEXT: %4 = stablehlo.convert %3 : (tensor<i32>) -> tensor<i64>
// CHECK-NEXT: %5 = stablehlo.iota dim = 0 : tensor<8xi64>
// CHECK-NEXT: %6 = stablehlo.slice %arg0 [0:8] : (tensor<9xi32>) -> tensor<8xi32>
// CHECK-NEXT: %7 = stablehlo.reshape %6 : (tensor<8xi32>) -> tensor<8x1xi32>
// CHECK-NEXT: %8 = stablehlo.reshape %7 : (tensor<8x1xi32>) -> tensor<8xi32>
// CHECK-NEXT: %9 = stablehlo.slice %arg0 [1:9] : (tensor<9xi32>) -> tensor<8xi32>
// CHECK-NEXT: %10 = stablehlo.reshape %9 : (tensor<8xi32>) -> tensor<8x1xi32>
// CHECK-NEXT: %11 = stablehlo.reshape %10 : (tensor<8x1xi32>) -> tensor<8xi32>
// CHECK-NEXT: %12 = stablehlo.broadcast_in_dim %cst, dims = [] : (tensor<f64>) -> tensor<8xf64>
// CHECK-NEXT: %13:2 = stablehlo.while(%iterArg = %c, %iterArg_2 = %12) : tensor<i64>, tensor<8xf64>
// CHECK-NEXT: cond {
// CHECK-NEXT: %16 = stablehlo.compare LT, %iterArg, %4 : (tensor<i64>, tensor<i64>) -> tensor<i1>
// CHECK-NEXT: stablehlo.return %16 : tensor<i1>
// CHECK-NEXT: } do {
// CHECK-NEXT: %16 = stablehlo.convert %iterArg : (tensor<i64>) -> tensor<i32>
// CHECK-NEXT: %17 = stablehlo.broadcast_in_dim %16, dims = [] : (tensor<i32>) -> tensor<8xi32>
// CHECK-NEXT: %18 = stablehlo.add %8, %17 : tensor<8xi32>
// CHECK-NEXT: %19 = stablehlo.compare LT, %18, %11, SIGNED : (tensor<8xi32>, tensor<8xi32>) -> tensor<8xi1>
// CHECK-NEXT: %20 = stablehlo.convert %18 : (tensor<8xi32>) -> tensor<8xi64>
// CHECK-NEXT: %21 = stablehlo.reshape %20 : (tensor<8xi64>) -> tensor<8x1xi64>
// CHECK-NEXT: %22 = "stablehlo.gather"(%arg1, %21) <{dimension_numbers = #stablehlo.gather<collapsed_slice_dims = [0], start_index_map = [0], index_vector_dim = 1>, indices_are_sorted = false, slice_sizes = array<i64: 1>}> : (tensor<40xi32>, tensor<8x1xi64>) -> tensor<8xi32>
// CHECK-NEXT: %23 = stablehlo.convert %22 : (tensor<8xi32>) -> tensor<8xi64>
// CHECK-NEXT: %24 = stablehlo.reshape %23 : (tensor<8xi64>) -> tensor<8x1xi64>
// CHECK-NEXT: %25 = "stablehlo.gather"(%arg2, %24) <{dimension_numbers = #stablehlo.gather<collapsed_slice_dims = [0], start_index_map = [0], index_vector_dim = 1>, indices_are_sorted = false, slice_sizes = array<i64: 1>}> : (tensor<40xf64>, tensor<8x1xi64>) -> tensor<8xf64>
// CHECK-NEXT: %26 = stablehlo.broadcast_in_dim %cst, dims = [] : (tensor<f64>) -> tensor<8xf64>
// CHECK-NEXT: %27 = stablehlo.broadcast_in_dim %19, dims = [0] : (tensor<8xi1>) -> tensor<8xi1>
// CHECK-NEXT: %28 = stablehlo.select %27, %25, %26 : tensor<8xi1>, tensor<8xf64>
// CHECK-NEXT: %29 = stablehlo.add %iterArg_2, %28 : tensor<8xf64>
// CHECK-NEXT: %30 = stablehlo.add %iterArg, %c_0 : tensor<i64>
// CHECK-NEXT: stablehlo.return %30, %29 : tensor<i64>, tensor<8xf64>
// CHECK-NEXT: }
// CHECK-NEXT: %14 = stablehlo.reshape %5 : (tensor<8xi64>) -> tensor<8x1xi64>
// CHECK-NEXT: %15 = "stablehlo.scatter"(%arg3, %14, %13#1) <{indices_are_sorted = false, scatter_dimension_numbers = #stablehlo.scatter<inserted_window_dims = [0], scatter_dims_to_operand_dims = [0], index_vector_dim = 1>, unique_indices = false}> ({
// CHECK-NEXT: ^bb0(%arg4: tensor<f64>, %arg5: tensor<f64>):
// CHECK-NEXT: stablehlo.return %arg5 : tensor<f64>
// CHECK-NEXT: }) : (tensor<8xf64>, tensor<8x1xi64>, tensor<8xf64>) -> tensor<8xf64>
// CHECK-NEXT: return %15 : tensor<8xf64>
// CHECK-NEXT: }
Loading