Distributed IO #533
Merged
Distributed IO #533
Conversation
This file contains 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
…mentation and testing
nirandaperera
requested changes
Dec 8, 2021
cpp/src/cylon/table_api.cpp
Outdated
Comment on lines
364
to
369
| buffers.resize(ctx->GetWorldSize()); | ||
| int32_t buf_start = 0; | ||
| for (int i = 0; i < ctx->GetWorldSize(); ++i) { | ||
| buffers[i] = std::make_shared<arrow::Buffer>(received_buf.data() + buf_start, all_buffer_sizes[i]); | ||
| buf_start += all_buffer_sizes[i]; | ||
| } |
There was a problem hiding this comment.
Suggested change
| buffers.resize(ctx->GetWorldSize()); | |
| int32_t buf_start = 0; | |
| for (int i = 0; i < ctx->GetWorldSize(); ++i) { | |
| buffers[i] = std::make_shared<arrow::Buffer>(received_buf.data() + buf_start, all_buffer_sizes[i]); | |
| buf_start += all_buffer_sizes[i]; | |
| } | |
| buffers.reserve(ctx->GetWorldSize()); | |
| int32_t buf_start = 0; | |
| for (int i = 0; i < ctx->GetWorldSize(); ++i) { | |
| buffers.push_back(std::make_shared<arrow::Buffer>(received_buf.data() + buf_start, all_buffer_sizes[i])); | |
| buf_start += all_buffer_sizes[i]; | |
| } |
nirandaperera
requested changes
Dec 13, 2021
There was a problem hiding this comment.
added some minor comments. I am happy with the approach though 🙂
Good job @ahmet-uyar
cpp/src/cylon/net/ops/gather.cpp
Outdated
| return cylon::Status(cylon::Code::ExecutionError, "MPI_Gather failed when receiving buffer sizes!"); | ||
| } | ||
|
|
||
| std::shared_ptr<arrow::Buffer> all_buf = std::move(arrow::AllocateBuffer(0).MoveValueUnsafe()); |
There was a problem hiding this comment.
Suggested change
| std::shared_ptr<arrow::Buffer> all_buf = std::move(arrow::AllocateBuffer(0).MoveValueUnsafe()); | |
| std::shared_ptr<arrow::Buffer> all_buf; |
we can leave the shared_ptr as it is (ie. nullptr)
|
@ahmet-uyar what about that |
|
@nirandaperera I have added arrow::buffer broadcast function. One can easily serialize a schema and broadcast. I think, we do not need a separate function for that. |
nirandaperera
approved these changes
Dec 13, 2021
nirandaperera
pushed a commit
that referenced
this pull request
Dec 16, 2021
* initial implementation of read_csv * added write_csv * added json read and write methods * implemented allgather for arrow.Buffer * added schema checking after reading and initializing empty dataframes * added test cases for read/write csv/json * added json data files for read tests * added read_parquet function * added write_parquet and test files for parquet io * added parquet files for io testing * added allgathering metadata, improved row_group distribution among workers * added MPI gathering arrow buffer function * implemented _metadata writing for parquet files * improved write function signatures * added reading parquet directories with _metadata files, improved documentation and testing * moved io util functions to pycylon/util directory * avoided buffer copying when gathering and allgathering arrow Buffers * added arrow buffer broadcasting * error checking on arrow::Buffer allocation and using std::partial_sum * improved looping in gather
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Implemented distributed read/write for csv, json and parquet files.
Added utility codes to Cylon code base, so that Cylon IO can be easily implemented.
Details can be found in the following design document:
https://docs.google.com/document/d/1y7TAU_15qNkQ9DBkw43El1CimhkAMqemW5dcGAwbZmk/edit?usp=sharing