Skip to content

Fix inflate operator: Reset max output volume. Use size in bytes, not elements. #6283

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
mzient merged 1 commit into from
Apr 9, 2026
Merged

Fix inflate operator: Reset max output volume. Use size in bytes, not elements. #6283

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
4 changes: 3 additions & 1 deletion dali/operators/decoder/inflate/inflate.h
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
// Copyright (c) 2022-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
// Copyright (c) 2022-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
Expand Down Expand Up @@ -38,6 +38,7 @@ class InflateOpImplBase : public OpImplBase<Backend> {
IsFloatingPoint(dtype_) || IsIntegral(dtype_),
make_string("The inflate output type must have floating point or integral type, got `",
dtype_, "` instead."));
element_type_ = TypeTable::GetTypeInfo(dtype_);
}

bool SetupImpl(std::vector<OutputDesc> &output_desc, const Workspace &ws) override {
Expand All @@ -62,6 +63,7 @@ class InflateOpImplBase : public OpImplBase<Backend> {

protected:
DALIDataType dtype_;
TypeInfo element_type_;
inflate::ShapeParams<Backend> params_;
};

Expand Down
11 changes: 7 additions & 4 deletions dali/operators/decoder/inflate/inflate_gpu.cc
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
// Copyright (c) 2022-2025, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
// Copyright (c) 2022-2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
Expand Down Expand Up @@ -58,9 +58,12 @@ class InflateOpGpuLZ4Impl : public InflateOpImplBase<GPUBackend> {
stream, params_.GetInChunkSizes(), input_ptrs_, inflated_sizes_, inflated_ptrs_);

size_t tempSize;
CUDA_CALL(nvcompBatchedLZ4DecompressGetTempSizeAsync(total_chunks_num,
params_.GetMaxOutChunkVol(),
{}, &tempSize, params_.GetMaxOutVol()));
CUDA_CALL(nvcompBatchedLZ4DecompressGetTempSizeAsync(
total_chunks_num,
params_.GetMaxOutChunkVol() * element_type_.size(),
{},
&tempSize,
params_.GetMaxOutVol() * element_type_.size()));

void *temp = scratchpad.AllocateGPU<uint8_t>(tempSize);
nvcompStatus_t *device_statuses = scratchpad.AllocateGPU<nvcompStatus_t>(total_chunks_num);
Expand Down
3 changes: 2 additions & 1 deletion dali/operators/decoder/inflate/inflate_params.h
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,4 +1,4 @@
// Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
// Copyright (c) 2022, 2026, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
Expand Down Expand Up @@ -354,6 +354,7 @@ class ShapeParams {
int sample_dim = provided_shape[0].num_elements();
TensorListShape<> shape(num_samples, sample_dim);
max_output_vol_ = 0;
max_output_sample_vol_ = 0;
for (int sample_idx = 0; sample_idx < provided_shape.num_samples(); sample_idx++) {
const int *data = provided_shape.tensor_data(sample_idx);
for (int d = 0; d < sample_dim; d++) {
Expand Down
Loading