[TOPI] Fast mode in take op
See original GitHub issueCurrently topi.take
supports two modes, “clip” and “wrap”, following the numpy take convention (https://docs.scipy.org/doc/numpy/reference/generated/numpy.take.html). However, the clipping operation makes the take op 30% slower when the input shape is sufficiently large, for example with input shape [20, 1500] and indices shape [750]. I would like to propose add a “fast” mode, similar to the numpy “raise” mode, with no clipping, wrapping, or raising error when the indices are out of range. The user will be responsible to make sure the indices are in range.
Issue Analytics
- State:
- Created 4 years ago
- Reactions:1
- Comments:6 (6 by maintainers)
Top Results From Across the Web
tvm.topi — tvm 0.11.dev0 documentation - Apache TVM
TOPI is the operator collection library for TVM, to provide sugars for ... Numpy style cumprod op. ... take (a, indices[, axis, batch_dims,...
Read more >[TVM Discuss] [Questions] [ TOPI ] Winograd ... - The Mail Archive
[/quote] I want to ask how to use one template which is implemented by TVM to ... The strategy to select implementations for...
Read more >ARM9E-S Core Technical Reference Manual r2p1
Instruction Cycles Instruction. bus Data. bus
CLZ 1 1S 1I
Data Op 1 1S 1I
Data Op 2 1S+1I 2I
Read more >Hardware acceleration | Android Developers
You can use the Paint parameter to apply color filters, special blending modes, or opacity to a layer. A view can use one...
Read more >Elden Ring - Easy Infinite Rune Location + Map - Gaming Intel
However, they can deal a lot of damage, so be careful! Fast travel to the Farum Greatbridge Site of Grace when you are...
Read more >Top Related Medium Post
No results found
Top Related StackOverflow Question
No results found
Troubleshoot Live Code
Lightrun enables developers to add logs, metrics and snapshots to live code - no restarts or redeploys required.
Start FreeTop Related Reddit Thread
No results found
Top Related Hackernoon Post
No results found
Top Related Tweet
No results found
Top Related Dev.to Post
No results found
Top Related Hashnode Post
No results found
Top GitHub Comments
How about adding a compile time warning as well?
LOG(WARNING)
should do. Please review https://github.com/dmlc/tvm/pull/3325