whisper.cpp/ggml
Jeff Bolz 0484147ab2
vulkan: fix top_k bug when there are ties in the input (llama/17659)
* vulkan: Reduce temporary memory usage for TOP_K

- Compute row size for the temp buffer based on the output of the first pass.
- Update shader addressing math to use the output row size
- Pass the output row size as "ncols_output", what used to be "ncols_output" is now "k"

For the common case of K=40 and src0=(200000,1,1,1), this reduces the temporary buffer
from about 3.2MB to 500KB.

* vulkan: fix top_k bug when there are ties in the input

I noticed by inspection a bug in the vulkan top_k shader where if the least
value in the top_k appears multiple times we could end up writing those extra
copies out rather than some larger values (if the larger values are on higher
numbered threads).

I rewrote the test verification to handle this case, where the final index set
is not necessarily the same.

* Update tests/test-backend-ops.cpp

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2025-12-12 17:53:19 +02:00
..
cmake ggml: Skip backend library linking code when GGML_BACKEND_DL=ON (llama/15094) 2025-08-18 20:30:45 +03:00
include rpc : fix alloc size logic (llama/17116) 2025-12-12 17:53:18 +02:00
src vulkan: fix top_k bug when there are ties in the input (llama/17659) 2025-12-12 17:53:19 +02:00
.gitignore whisper : reorganize source code + improve CMake (#2256) 2024-06-26 19:34:09 +03:00
CMakeLists.txt build : move _WIN32_WINNT definition to headers (llama/17736) 2025-12-12 17:53:16 +02:00