This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| echo "compiling the old vmfb" | |
| iree-compile --iree-hal-target-backends=rocm --iree-hip-target=gfx942 --iree-vm-bytecode-module-output-format=flatbuffer-binary --iree-preprocessing-pass-pipeline="builtin.module(util.func(iree-global-opt-raise-special-ops, iree-flow-canonicalize), iree-preprocessing-transpose-convolution-pipeline, iree-preprocessing-pad-to-intrinsics, util.func(iree-preprocessing-generalize-linalg-matmul-experimental))" --iree-hal-dump-executable-files-to=dump/ --iree-dispatch-creation-enable-aggressive-fusion --iree-dispatch-creation-enable-fuse-horizontal-contractions=false --iree-opt-aggressively-propagate-transposes=true --iree-codegen-llvmgpu-use-vector-distribution=true --iree-opt-data-tiling=false --iree-vm-target-truncate-unsupported-floats --iree-opt-outer-dim-concat=true --iree-hal-indirect-command-buffers=true --iree-stream-resource-memory-model=discrete --iree-hal-memoization=true --iree-opt-strip-assertions --iree-global-opt-propagate-transposes=true --iree-opt-const-eval=false --i |
pashu123
/ compare.py
Created
April 23, 2025 23:41
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| import numpy as np | |
| # Load arrays with FP16 data type | |
| old = np.load('old.npy').astype(np.float16) | |
| new = np.load('new.npy').astype(np.float16) | |
| print(old.shape) | |
| print(new.shape) | |
| # Check shape compatibility |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #map = affine_map<(d0, d1, d2) -> (d2, d0)> | |
| #map1 = affine_map<(d0, d1, d2) -> (d1, d2)> | |
| #map2 = affine_map<(d0, d1, d2) -> (d0, d1)> | |
| module { | |
| func.func @faulty(%arg0: tensor<2816x2xf16>, %arg1: tensor<1280x2816xf16>) -> tensor<2x1280xf32> { | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %0 = tensor.empty() : tensor<2x1280xf32> | |
| %1 = linalg.fill ins(%cst : f32) outs(%0 : tensor<2x1280xf32>) -> tensor<2x1280xf32> | |
| %2 = linalg.generic {indexing_maps = [#map, #map1, #map2], iterator_types = ["parallel", "parallel", "reduction"]} ins(%arg0, %arg1 : tensor<2816x2xf16>, tensor<1280x2816xf16>) outs(%1 : tensor<2x1280xf32>) { | |
| ^bb0(%in: f16, %in_0: f16, %out: f32): |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #map = affine_map<(d0, d1, d2) -> (d2, d0)> | |
| #map1 = affine_map<(d0, d1, d2) -> (d1, d2)> | |
| #map2 = affine_map<(d0, d1, d2) -> (d0, d1)> | |
| #map3 = affine_map<(d0, d1) -> (d0, d1)> | |
| #map4 = affine_map<(d0, d1) -> (d1)> | |
| module { | |
| func.func @faulty(%arg0: tensor<2816x2xf16>, %arg1: tensor<1280x2816xf16>) -> tensor<2x1280xf32> { | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %cst_0 = arith.constant 1.000000e+00 : f16 | |
| %0 = tensor.empty() : tensor<2x1280xf16> |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| echo "compiling the old vmfb" | |
| iree-compile --iree-hal-target-backends=rocm --iree-hip-target=gfx942 --iree-vm-bytecode-module-output-format=flatbuffer-binary --iree-preprocessing-pass-pipeline="builtin.module(util.func(iree-global-opt-raise-special-ops, iree-flow-canonicalize), iree-preprocessing-transpose-convolution-pipeline, iree-preprocessing-pad-to-intrinsics, util.func(iree-preprocessing-generalize-linalg-matmul-experimental))" --iree-hal-dump-executable-files-to=dump/ --iree-dispatch-creation-enable-aggressive-fusion --iree-dispatch-creation-enable-fuse-horizontal-contractions=false --iree-opt-aggressively-propagate-transposes=true --iree-codegen-llvmgpu-use-vector-distribution=true --iree-opt-data-tiling=false --iree-vm-target-truncate-unsupported-floats --iree-opt-outer-dim-concat=true --iree-hal-indirect-command-buffers=true --iree-stream-resource-memory-model=discrete --iree-hal-memoization=true --iree-opt-strip-assertions --iree-global-opt-propagate-transposes=true --iree-opt-const-eval=false --i |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| import numpy as np | |
| import os | |
| # Tensor shapes | |
| shapes = { | |
| "arg0": (2816, 2), | |
| "arg1": (1280, 2816), | |
| } | |
| # Generate and save tensors |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| I am inside the populateWarpAndThreadIndices | |
| The rank is 1 | |
| The threadIds size is2 | |
| I am inside the populateWarpAndThreadIndices | |
| The rank is 0 | |
| The threadIds size is0 | |
| I am inside the populateWarpAndThreadIndices | |
| The rank is 1 | |
| The threadIds size is2 | |
| I am inside the populateWarpAndThreadIndices |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| I am inside the populateWarpAndThreadIndices | |
| The rank is 1 | |
| The threadIds size is2 | |
| I am inside the populateWarpAndThreadIndices | |
| The rank is 1 | |
| The threadIds size is2 | |
| I am inside the populateWarpAndThreadIndices | |
| The rank is 0 | |
| The threadIds size is0 | |
| // -----// IR Dump After LLVMGPUVectorDistributePass (iree-llvmgpu-vector-distribute) //----- // |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Running with warp reduction | |
| --------------------------------------------------------------------------------------------------------- | |
| Benchmark Time CPU Iterations UserCounters... | |
| --------------------------------------------------------------------------------------------------------- | |
| BM__4_14336_4096/process_time/real_time 0.510 ms 0.489 ms 1394 items_per_second=1.96019k/s | |
| BM__4_14336_4096/process_time/real_time 0.512 ms 0.518 ms 1394 items_per_second=1.95449k/s | |
| BM__4_14336_4096/process_time/real_time 0.512 ms 0.519 ms 1394 items_per_second=1.9534k/s | |
| BM__4_14336_4096/process_time/real_time 0.513 ms 0.521 ms 1394 items_per_second=1.95046k/s | |
| BM__4_14336_4096/process_time/real_time 0.514 ms 0.521 ms 1394 items_per_second=1.94656k/s | |
| BM__4_14336_4096/process_time/real_time_mean 0.512 ms 0.513 ms 5 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| --------------------------------------------------------------------------------------------------------- | |
| Benchmark Time CPU Iterations UserCounters... | |
| --------------------------------------------------------------------------------------------------------- | |
| BM__4_14336_4096/process_time/real_time 0.510 ms 0.489 ms 1394 items_per_second=1.96019k/s | |
| BM__4_14336_4096/process_time/real_time 0.512 ms 0.518 ms 1394 items_per_second=1.95449k/s | |
| BM__4_14336_4096/process_time/real_time 0.512 ms 0.519 ms 1394 items_per_second=1.9534k/s | |
| BM__4_14336_4096/process_time/real_time 0.513 ms 0.521 ms 1394 items_per_second=1.95046k/s | |
| BM__4_14336_4096/process_time/real_time 0.514 ms 0.521 ms 1394 items_per_second=1.94656k/s | |
| BM__4_14336_4096/process_time/real_time_mean 0.512 ms 0.513 ms 5 items_per_second=1.95302k/s |