This is an archive of the discontinued LLVM Phabricator instance.

Differential D112609

[mlir] Reduce the number of iterations in async microbenchmarks
ClosedPublic

Authored by ezhulenev on Oct 27 2021, 3:18 AM.

Details

Reviewers
nicolasvasilache
Commits
rG0d9b47893281: [mlir] Reduce the number of iterations in async microbenchmarks

Diff Detail

Event Timeline

ezhulenev created this revision.Oct 27 2021, 3:18 AM
Herald added subscribers: wenzhicui, wrengr, Chia-hungDuan and 19 others. · View Herald TranscriptOct 27 2021, 3:18 AM
ezhulenev requested review of this revision.Oct 27 2021, 3:18 AM
Herald added a project: Restricted Project. · View Herald TranscriptOct 27 2021, 3:18 AM
Herald added subscribers: stephenneuendorffer, nicolasvasilache. · View Herald Transcript
ezhulenev added a reviewer: nicolasvasilache.Oct 27 2021, 3:19 AM
This revision was not accepted when it landed; it landed in state Needs Review.Oct 27 2021, 3:20 AM
This revision was landed with ongoing or failed builds.
Closed by commit rG0d9b47893281: [mlir] Reduce the number of iterations in async microbenchmarks (authored by ezhulenev). · Explain Why
This revision was automatically updated to reflect the committed changes.
ezhulenev added a commit: rG0d9b47893281: [mlir] Reduce the number of iterations in async microbenchmarks.
Harbormaster completed remote builds in B130903: Diff 382586.Oct 27 2021, 3:45 AM
nicolasvasilache added a comment.Oct 27 2021, 6:02 AM

Thanks!

Revision Contents

PathSize
mlir/
test/
Integration/
Dialect/
Async/
CPU/
4 lines
4 lines

Diff 382587

mlir/test/Integration/Dialect/Async/CPU/microbench-linalg-async-parallel-for.mlir

Show First 20 Lines • Show All 51 Lines • ▼ Show 20 Linesfunc @linalg_generic(%lhs: memref<?x?xf32>,
return return
} }
func @entry() { func @entry() {
%f1 = arith.constant 1.0 : f32 %f1 = arith.constant 1.0 : f32
%f4 = arith.constant 4.0 : f32 %f4 = arith.constant 4.0 : f32
%c0 = arith.constant 0 : index %c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index %c1 = arith.constant 1 : index
%cM = arith.constant 1000 : index %cN = arith.constant 50 : index
// //
// Sanity check for the function under test. // Sanity check for the function under test.
// //
%LHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32> %LHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32>
%RHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32> %RHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32>
%DST10 = memref.alloc() {alignment = 64} : memref<1x10xf32> %DST10 = memref.alloc() {alignment = 64} : memref<1x10xf32>
Show All 35 Linesfunc @entry() {
call @linalg_generic(%LHS0, %RHS0, %DST0) call @linalg_generic(%LHS0, %RHS0, %DST0)
: (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> () : (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> ()
// //
// Measure execution time. // Measure execution time.
// //
%t0 = call @rtclock() : () -> f64 %t0 = call @rtclock() : () -> f64
scf.for %i = %c0 to %cM step %c1 { scf.for %i = %c0 to %cN step %c1 {
call @linalg_generic(%LHS0, %RHS0, %DST0) call @linalg_generic(%LHS0, %RHS0, %DST0)
: (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> () : (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> ()
} }
%t1 = call @rtclock() : () -> f64 %t1 = call @rtclock() : () -> f64
%t1024 = arith.subf %t1, %t0 : f64 %t1024 = arith.subf %t1, %t0 : f64
// Print timings. // Print timings.
vector.print %t1024 : f64 vector.print %t1024 : f64
Show All 13 Lines

mlir/test/Integration/Dialect/Async/CPU/microbench-scf-async-parallel-for.mlir

Show First 20 Lines • Show All 72 Lines • ▼ Show 20 Linesfunc @scf_parallel(%lhs: memref<?x?xf32>,
return return
} }
func @entry() { func @entry() {
%f1 = arith.constant 1.0 : f32 %f1 = arith.constant 1.0 : f32
%f4 = arith.constant 4.0 : f32 %f4 = arith.constant 4.0 : f32
%c0 = arith.constant 0 : index %c0 = arith.constant 0 : index
%c1 = arith.constant 1 : index %c1 = arith.constant 1 : index
%cM = arith.constant 1000 : index %cN = arith.constant 50 : index
// //
// Sanity check for the function under test. // Sanity check for the function under test.
// //
%LHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32> %LHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32>
%RHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32> %RHS10 = memref.alloc() {alignment = 64} : memref<1x10xf32>
%DST10 = memref.alloc() {alignment = 64} : memref<1x10xf32> %DST10 = memref.alloc() {alignment = 64} : memref<1x10xf32>
Show All 35 Linesfunc @entry() {
call @scf_parallel(%LHS0, %RHS0, %DST0) call @scf_parallel(%LHS0, %RHS0, %DST0)
: (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> () : (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> ()
// //
// Measure execution time. // Measure execution time.
// //
%t0 = call @rtclock() : () -> f64 %t0 = call @rtclock() : () -> f64
scf.for %i = %c0 to %cM step %c1 { scf.for %i = %c0 to %cN step %c1 {
call @scf_parallel(%LHS0, %RHS0, %DST0) call @scf_parallel(%LHS0, %RHS0, %DST0)
: (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> () : (memref<?x?xf32>, memref<?x?xf32>, memref<?x?xf32>) -> ()
} }
%t1 = call @rtclock() : () -> f64 %t1 = call @rtclock() : () -> f64
%t1024 = arith.subf %t1, %t0 : f64 %t1024 = arith.subf %t1, %t0 : f64
// Print timings. // Print timings.
vector.print %t1024 : f64 vector.print %t1024 : f64
Show All 13 Lines