| commit | e026545ad5d91928b3cb9afde16e74fad612bfdb | [log] [tgz] |
|---|---|---|
| author | Ben Vanik <ben.vanik@gmail.com> | Tue Jan 31 13:21:06 2023 -0800 |
| committer | GitHub <noreply@github.com> | Tue Jan 31 13:21:06 2023 -0800 |
| tree | 7fe66e338e118f9dfc2cc8684b22895c2c982f24 | |
| parent | 71aa720a990bed848628e0c3972574ea4ae075b6 [diff] | |
| parent | 7ab025c8a18d3c3c7a93e914f4cabb9f055dd0cf [diff] |
Extending `iree_hal_allocator_t` with memory heaps and compat queries. (#12002) The new `iree_hal_allocator_query_memory_heaps` API allows allocators to expose the available memory heaps to external layers that need to make decisions based on the characteristics of the underlying allocator. The update to `iree_hal_allocator_query_buffer_compatibility` provides pooling layers the ability to ask implementations to adjust the parameters and allocation size of a request without performing the allocation. A pooling layer that was trying to find an appropriate bucket or virtual addressing layer that needed to query for reservation ranges can take the user params, have the implementation adjust them, and assume that any allocations made would have those exact parameters. To aid flag/configuration utilities some bitfield parsing has been added for HAL memory types that can be extended to all enums in the future.
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.