Lower cache_read and cache_write to Hexagon DMA via tensorize #10365
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
adstraw
requested review from
vinx13,
tqchen,
liangfu,
areusch,
tmoreau89,
masahi,
kazum,
ZihengJiang,
junrushao,
kparzysz-quic,
Hzfengsy,
comaniac,
jroesch,
yzhliu,
merrymercy and
icemelon
as code owners
|
HUGE thanks to @masahi who also contributed to this PR. |
masahi
reviewed
Feb 24, 2022
masahi
reviewed
Feb 24, 2022
| * Same semantics as memcpy(destination, source, size) | ||
| * Allows for device specific implementations e.g. direct memory access (DMA) | ||
| */ | ||
| TVM_DLL const Op& mem_copy(); |
There was a problem hiding this comment.
cc @tqchen @junrushao1994 if they want a better name here.
There was a problem hiding this comment.
Is it possible to make it more specific? We have a few examples above like texture2d_load/store. What about dma_mem_copy() or any hexagon-specific name?
There was a problem hiding this comment.
I was thinking about something with "DMA" in the name but I realized that DMA is more of an implementation detail. I ended up referencing DMA in the comments instead. Happy to take suggestions including dma_mem_copy.
There was a problem hiding this comment.
I thought about:
mem_copy_local(to emphasize the local nature of the intended usage, unlikeTVMArrayCopyFromTo@junrushao1994 mentioned which might involve RPC)mem_copy_1d(if we intend to add 2d rectangular memcpy later)
But both of them don't particularly sound better.
masahi
reviewed
Feb 24, 2022
masahi
reviewed
Feb 24, 2022
masahi
approved these changes
Feb 24, 2022
pfk-beta
pushed a commit
to pfk-beta/tvm
that referenced
this pull request
Apr 11, 2022
…#10365) * Lower cache_read and cache_write to Hexagon DMA via tensorize * rework test to be compatible with launcher * remove cpu device api mem_copy implementation and test
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.
Adds support to lower
cache_readandcache_writeschedule primitives to Hexagon User DMA usingtensorizeand newmem_copyintrinsic.