feat(backend): add turboquant llama.cpp-fork backend#9355
feat(backend): add turboquant llama.cpp-fork backend#9355
Conversation
turboquant is a llama.cpp fork (TheTom/llama-cpp-turboquant, branch feature/turboquant-kv-cache) that adds a TurboQuant KV-cache scheme. It ships as a first-class backend reusing backend/cpp/llama-cpp sources via a thin wrapper Makefile: each variant target copies ../llama-cpp into a sibling build dir and invokes llama-cpp's build-llama-cpp-grpc-server with LLAMA_REPO/LLAMA_VERSION overridden to point at the fork. No duplication of grpc-server.cpp — upstream fixes flow through automatically. Wires up the full matrix (CPU, CUDA 12/13, L4T, L4T-CUDA13, ROCm, SYCL f32/f16, Vulkan) in backend.yml and the gallery entries in index.yaml, adds a tests-turboquant-grpc e2e job driven by BACKEND_TEST_CACHE_TYPE_K/V=q8_0 to exercise the KV-cache config path (backend_test.go gains dedicated env vars wired into ModelOptions.CacheTypeKey/Value — a generic improvement usable by any llama.cpp-family backend), and registers a nightly auto-bump PR in bump_deps.yaml tracking feature/turboquant-kv-cache. scripts/changed-backends.js gets a special-case so edits to backend/cpp/llama-cpp/ also retrigger the turboquant CI pipeline, since the wrapper reuses those sources.
turboquant branched from llama.cpp before upstream commit 66060008
("server: respect the ignore eos flag", #21203) which added the
`logit_bias_eog` field to `server_context_meta` and a matching
parameter to `server_task::params_from_json_cmpl`. The shared
backend/cpp/llama-cpp/grpc-server.cpp depends on that field, so
building it against the fork unmodified fails.
Cherry-pick that commit as a patch file under
backend/cpp/turboquant/patches/ and apply it to the cloned fork
sources via a new apply-patches.sh hook called from the wrapper
Makefile. Simplifies the build flow too: instead of hopping through
llama-cpp's build-llama-cpp-grpc-server indirection, the wrapper now
drives the copied Makefile directly (clone -> patch -> build).
Drop the corresponding patch whenever the fork catches up with
upstream — the build fails fast if a patch stops applying, which
is the signal to retire it.
Document the new turboquant (llama.cpp fork with TurboQuant KV-cache) backend alongside the existing llama-cpp / ik-llama-cpp sections in features/text-generation.md: when to pick it, how to install it from the gallery, and a YAML example showing backend: turboquant together with cache_type_k / cache_type_v. Also expand the cache_type_k / cache_type_v table rows in advanced/model-configuration.md to spell out the accepted llama.cpp quantization values and note that these fields apply to all llama.cpp-family backends, not just vLLM.
The fork adds new GGML ops bringing GGML_OP_COUNT to 97, but ggml/include/ggml-rpc.h static-asserts it equals 96, breaking the GGML_RPC=ON build paths (turboquant-grpc / turboquant-rpc-server). Carry a one-line patch that updates the expected count so the assertion holds. Drop this patch whenever the fork fixes it upstream.
The shared backend/cpp/llama-cpp/grpc-server.cpp carries its own allow-list of accepted KV-cache types (kv_cache_types[]) and rejects anything outside it before the value reaches llama.cpp's parser. That list only contains the standard llama.cpp types — turbo2/turbo3/turbo4 would throw "Unsupported cache type" at LoadModel time, meaning nothing the LocalAI gRPC layer accepted was actually fork-specific. Add a build-time augmentation step (patch-grpc-server.sh, called from the turboquant wrapper Makefile) that inserts GGML_TYPE_TURBO2_0/3_0/4_0 into the allow-list of the *copied* grpc-server.cpp under turboquant-<flavor>-build/. The original file under backend/cpp/llama-cpp/ is never touched, so the stock llama-cpp build keeps compiling against vanilla upstream which has no notion of those enum values. Switch test-extra-backend-turboquant to set BACKEND_TEST_CACHE_TYPE_K=turbo3 / _V=turbo3 so the e2e gRPC suite actually runs the fork's TurboQuant KV-cache code paths (turbo3 also auto-enables flash_attention in the fork). Picking q8_0 here would only re-test the standard llama.cpp path that the upstream llama-cpp backend already covers. Refresh the docs (text-generation.md + model-configuration.md) to list turbo2/turbo3/turbo4 explicitly and call out that you only get the TurboQuant code path with this backend + a turbo* cache type.
The builder image (ubuntu:24.04 stage-2 in Dockerfile.turboquant) does not install python3, so the python-based augmentation step errored with `python3: command not found` at make time. Switch to awk, which ships in coreutils and is already available everywhere the rest of the wrapper Makefile runs.
|
@TheTom this is the good one :) will merge it soon and cut a LocalAI release with it, will keep you posted! |
this is great, thank you for putting in the work on the full backend integration. the CI wiring and nightly auto-bump tracking the branch is exactly right. let me know if you hit any issues with the fork API surface and i'll make sure nothing breaks on my end. one suggestion: the e2e test currently uses q8_0 for both K and V. the recommended production config from our testing is asymmetric: q8_0 K + turbo3 V. symmetric q8_0/q8_0 works but doesn't exercise the actual TurboQuant compression path where the savings come from (V compression is essentially free on quality, K is where precision matters). switching the test to BACKEND_TEST_CACHE_TYPE_K=q8_0 BACKEND_TEST_CACHE_TYPE_V=turbo3 would validate the path users will actually run. |
Signed-off-by: Ettore Di Giacinto <mudler@users.noreply.github.com>
Good catch, thanks! actually the comment was misleading as e2e tests were actually running turbo3/turbo3 - now I've correctly wired as q8_0/turbo3, thanks for the tip! |
2 participants
turboquant is a llama.cpp fork (TheTom/llama-cpp-turboquant, branch feature/turboquant-kv-cache) that adds a TurboQuant KV-cache scheme. It ships as a first-class backend reusing backend/cpp/llama-cpp sources via a thin wrapper Makefile: each variant target copies ../llama-cpp into a sibling build dir and invokes llama-cpp's build-llama-cpp-grpc-server with LLAMA_REPO/LLAMA_VERSION overridden to point at the fork. No duplication of grpc-server.cpp: upstream fixes flow through automatically.
Wires up the full matrix (CPU, CUDA 12/13, L4T, L4T-CUDA13, ROCm, SYCL f32/f16, Vulkan) in backend.yml and the gallery entries in index.yaml, adds a tests-turboquant-grpc e2e job driven by BACKEND_TEST_CACHE_TYPE_K/V=q8_0 to exercise the KV-cache config path (backend_test.go gains dedicated env vars wired into ModelOptions.CacheTypeKey/Value — a generic improvement usable by any llama.cpp-family backend), and registers a nightly auto-bump PR in bump_deps.yaml tracking feature/turboquant-kv-cache.
scripts/changed-backends.js gets a special-case so edits to backend/cpp/llama-cpp/ also retrigger the turboquant CI pipeline, since the wrapper reuses those sources.