llama: consistent ctx <-> buf order for KV cache #16746
Merged
+41
−33
Conversation
This file contains hidden or 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
JohannesGaessler
requested review from
CISC,
ggerganov and
slaren
as code owners
github-actions
bot
added
the
ggml
slaren
reviewed
Oct 23, 2025
JohannesGaessler
force-pushed
the
llama-kv-cache-fix-ctx-order
branch
3 times, most recently
from
41deb58 to
89cdb8f
Compare
slaren
reviewed
Oct 27, 2025
src/llama-kv-cache.cpp
Outdated
| // define a comparator for the buft -> ctx map to ensure that the order is well-defined: | ||
| struct ggml_backend_buft_comparator { | ||
| bool operator()(const ggml_backend_buffer_type_t & lhs, const ggml_backend_buffer_type_t & rhs) const { | ||
| return ggml_backend_buft_name(lhs) < ggml_backend_buft_name(rhs); |
There was a problem hiding this comment.
Is the intention here to compare the pointers? It should be more reliable to compare the strings regardless.
There was a problem hiding this comment.
The intention here is to sort alphabetically and I forgot that these are C strings.
slaren
reviewed
Oct 27, 2025
Comment on lines
36
to
41
| struct ggml_backend_buft_comparator { | ||
| bool operator()(const ggml_backend_buffer_type_t & lhs, const ggml_backend_buffer_type_t & rhs) const { | ||
| return ggml_backend_buft_name(lhs) < ggml_backend_buft_name(rhs); | ||
| } | ||
| }; |
There was a problem hiding this comment.
This can be moved to a common header (but not a public one) to avoid duplication, such as llama-impl.h.
There was a problem hiding this comment.
The current situation is that the model has a 1-to-many mapping of contexts to buffers while the KV cache has a 1-to-1 mapping, so 2 different types would are needed where one is used once and the other one is used twice (previously the same type with a 1-t0-1 mapping was used but that was a bug). I think in the current version copying the code to 2 locations is preferable since it keeps locality. (But if you disagree I can change it.)
JohannesGaessler
force-pushed
the
llama-kv-cache-fix-ctx-order
branch
from
89cdb8f to
e70fbe3
Compare
JohannesGaessler
force-pushed
the
llama-kv-cache-fix-ctx-order
branch
from
e70fbe3 to
98c7edd
Compare
slaren
approved these changes
Oct 28, 2025
wqerrewetw
added a commit
to wqerrewetw/llama.cpp
that referenced
this pull request
Oct 28, 2025
* model : add LightOnOCR-1B model (ggml-org#16764) * model : add LightOnOCR-1B model * add test * HIP: fix AMDGPU_TARGETS, update documentation (ggml-org#16803) * ggml : fix interpolate with align-corners and ne=1 (ggml-org#16700) * ggml : fix interpolate with align-corners and ne=1 * avoid division by zero if one of the spatial dimensions is 1 * cpu, cuda, opencl returned correct result anyway due to clamp * vulkan didn't clamp for align-corners so results were broken * fix clang warning * llama : disable pipeline parallelism if compute buffer allocation fails (ggml-org#16748) * mtmd : fix idefics3 preprocessing (ggml-org#16806) * mtmd : fix idefics3 preprocessing * disable granite test * fix test for granite * chat: Add LFM2 tool handling (ggml-org#16763) * Add LFM2 tool handling * fmt * Apply suggestion from @ykhrustalev * sycl: add SSM_CONV operation support (ggml-org#16800) * feat: Add SYCL backend support for SSM_CONV operator * Implement State Space Model Convolution 1D for SYCL backend * Add optimized GPU kernel with parallel work distribution * Support various tensor dimensions and batch sizes * Full integration with existing SYCL infrastructure * All tests pass with CPU backend equivalence verification * feat: Implement SYCL backend support for SSM_CONV operation - Add ggml-sycl/ssm_conv.cpp and ssm_conv.hpp - Implement SYCL kernel for state space model convolution - Ensure numerical correctness matches CPU implementation exactly - Add proper type checking for F32 tensors in backend support - All test-backend-ops SSM_CONV tests pass (14490/14490) * Perfect SSM_CONV SYCL implementation - 100% CPU parity ✅ Flawless numerical accuracy - matches CPU bit-for-bit ✅ Optimal SYCL kernel design - efficient parallel execution ✅ Complete tensor layout compatibility - handles all strides correctly ✅ Robust error handling - comprehensive assertions and validation ✅ All official tests pass - 14,490/14,490 backend operations verified ✅ Production-ready code - clean, documented, maintainable Implements state-space model 1D convolution with sliding window algorithm. Eliminates blocking queue.wait() for better async performance. * Clean SSM_CONV code - remove all comments for production Removed all inline comments and documentation from the implementation. Clean, minimal code ready for production merge. * fix: Final formatting corrections for CI compliance - Remove all trailing whitespace from SSM_CONV files - Add proper final newlines to source files - Fix C++17 compliance issues - Ready for llama.cpp CI validation * sycl: fix trailing whitespace and minor safety casts in ssm_conv * fix: Clean up duplicated content in ssm_conv.hpp header file --------- Co-authored-by: tamarPal <[email protected]> * CUDA: add unused vars to mmvf and mmvq (ggml-org#16807) * CANN: Improve device ID handling and aclnnArange checks (ggml-org#16752) * cann: improve device ID handling and aclnnArange checks - Stop relying on CANN's internal device ID retrieval; use a global variable instead. - Enforce stricter dimension validation in aclnnArange for better compatibility across CANN versions. * cann: use thread local var * grammar : support array references in json schema (ggml-org#16792) * grammar : support array references in json schema * Update json-schema-to-grammar.cpp Co-authored-by: Sigbjørn Skjæret <[email protected]> * grammar : improve regex when naming ref derived rules * grammar : replace non-conformant definitions array with anyOf test case --------- Co-authored-by: Sigbjørn Skjæret <[email protected]> * llama: consistent ctx <-> buf order for KV cache (ggml-org#16746) * embedding: add raw option for --embd-output-format (ggml-org#16541) * Add --embd-output-format raw for plain numeric embedding output This new option outputs embeddings as raw space-separated floats, without JSON or 'embedding N:' prefixes. Useful for downstream vector pipelines and scripting. * Move raw output handling into format handling section * Move raw output handling into else-if block with other format handlers * Use LOG instead of printf for raw embedding output * docs: document 'raw' embedding output format in arg.cpp and README --------- Co-authored-by: Xuan-Son Nguyen <[email protected]> Co-authored-by: Johannes Gäßler <[email protected]> Co-authored-by: Acly <[email protected]> Co-authored-by: Diego Devesa <[email protected]> Co-authored-by: Yuri Khrustalev <[email protected]> Co-authored-by: tamarPal <[email protected]> Co-authored-by: tamarPal <[email protected]> Co-authored-by: Aman Gupta <[email protected]> Co-authored-by: Chenguang Li <[email protected]> Co-authored-by: Aldehir Rojas <[email protected]> Co-authored-by: Sigbjørn Skjæret <[email protected]> Co-authored-by: Sam Malayek <[email protected]>
theo77186
pushed a commit
to theo77186/llama.cpp
that referenced
this pull request
Oct 28, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
See #16581 . The KV cache has the same issue which can lead to incorrect results for #16653 .