TFLM: Rename FinishTensorAllocation to FinishModelAllocation in the comments.
The name of this function has been changed. PiperOrigin-RevId: 316790532 Change-Id: I31a77d3c2e7fe5af3138c99317ccd075ecbdf4ff
This commit is contained in:
parent
cdb6e80b21
commit
81cff8fc90
@ -40,7 +40,7 @@ TfLiteStatus InitializeTfLiteTensorFromFlatbuffer(
|
||||
|
||||
// A handle tracking scratch buffer allocation. This handle is created by
|
||||
// `RequestScratchBufferInArena`. `data` field is populated in
|
||||
// `FinishTensorAllocation` after static memory planning.
|
||||
// `FinishModelAllocation` after static memory planning.
|
||||
// TODO(b/150257460) As a future optimization, this struct could be replaced by
|
||||
// a union, since once `data` is populated, `bytes` and `node_idx` is not
|
||||
// needed.
|
||||
@ -126,7 +126,7 @@ class MicroAllocator {
|
||||
|
||||
// Register a scratch buffer of size `bytes` for Node with `node_id`.
|
||||
// This method only allocates a BufferHandle holding information for memory
|
||||
// planning. The buffer ptr is ready after `FinishTensorAllocation` and can
|
||||
// planning. The buffer ptr is ready after `FinishModelAllocation` and can
|
||||
// be retrieved by `GetScratchBuffer` method using the returned buffer_idx.
|
||||
// Note that there should be no tail allocation between two consecutive
|
||||
// `RequestScratchBufferInArena` calls.
|
||||
@ -136,7 +136,7 @@ class MicroAllocator {
|
||||
void* GetScratchBuffer(int buffer_idx) const;
|
||||
|
||||
// Returns the arena usage in bytes, only available after
|
||||
// `FinishTensorAllocation`. Otherwise, it will return 0.
|
||||
// `FinishModelAllocation`. Otherwise, it will return 0.
|
||||
size_t used_bytes() const;
|
||||
|
||||
protected:
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user