[BUG] Multi-GPU tensor-split offloads all physical layers to GPU0 on looped architectures
Nadie ha tomado este issue todavía.
Evaluación
- Dificultad
- 3/5
- Tiempo estimado
- Medio día
- Aptitud para principiantes
- 58/100
Línea de trabajo
Start in src/llama-model.cpp at llama_model_base::create_tensor, where tn.bid (the GGUF physical block id) indexes pimpl->dev_layer, which is sized to n_layer_all; read loop_visit and loop_layer in the hparams/model code to confirm the mapping described. Apply the proposed change so the index goes through hparams.loop_visit(tn.bid) and check that non-looped models still get identity mapping. Done means a looped model loaded with --tensor-split places postlude blocks on the second GPU and no CUDA0 OOM occurs, which needs multi-GPU hardware.
Escrito por el modelo de indexación a partir del texto del issue.
Descripción
[BUG] Multi-GPU tensor-split offloads all physical layers to GPU0 on looped architectures
Description
When loading a looped model (such as Inkling-2-Mini, where 24 physical layers expand to 48 graph execution layers) with --tensor-split across multiple GPUs, all physical layers are erroneously assigned to device 0 (CUDA0), starving secondary GPUs (CUDA1) and causing an immediate out-of-memory crash:
ggml_backend_cuda_buffer_type_alloc_buffer: allocating 11383.97 MiB on device 0: cudaMalloc failed: out of memory
ggml_backend_buft_alloc_buffer_n_default: failed to allocate CUDA0 buffer of size 11936954112
Root Cause
-
In GGUF files for looped models, physical layer blocks are numbered
blk.0toblk.23(n_layer_weights = 24), whereas the graph execution sequence containsn_layer_all = 48layers:0..5: prelude (physical0..5)6..17: loop pass 0 (physical6..17)18..29: loop pass 1 (physical6..17)30..41: loop pass 2 (physical6..17)42..47: postlude (physical18..23)
-
In
llama_model_base::create_tensor:if (tn.bid != -1) { buft_list_layer = (size_t) tn.bid < pimpl->dev_layer.size() ? pimpl->dev_layer.at(tn.bid).buft_list : pimpl->dev_output.buft_list; }Here
tn.bidis the physical block index in the GGUF file (0..23).
pimpl->dev_layeris sized ton_layer_all(48), wherepimpl->dev_layer[il] = get_layer_buft_list(loop_layer(il)).When
create_tensoris called for postlude physical blocks18..23:
It indexesdev_layer.at(tn.bid), which evaluates todev_layer.at(18..23).
Because graph slots18..23are pass 1 of the looped core,loop_layer(18..23)maps to6..11(GPU0).
Consequently, physical layers18..23(which were intended for GPU1) receive the device assignment of physical layers6..11(GPU0).
Physical layers0..17are also mapped to0..17(GPU0).
Thus, all 24 physical layers in the file are allocated on GPU0, regardless of--tensor-split.
Proposed Fix
In src/llama-model.cpp:
Map tn.bid to its canonical graph visit slot via hparams.loop_visit(tn.bid):
ggml_tensor * llama_model_base::create_tensor(llama_model_loader & ml, const LLM_TN_IMPL & tn, const std::initializer_list<int64_t> & ne, int flags) {
const buft_list_t * buft_list_layer = nullptr;
if (tn.bid != -1) {
// blocks that are not model layers (e.g. the blocks of a head) go with the output
const size_t iv = (size_t) hparams.loop_visit(tn.bid);
buft_list_layer = iv < pimpl->dev_layer.size() ? pimpl->dev_layer.at(iv).buft_list : pimpl->dev_output.buft_list;
}
return ml.create_tensor(
hparams, &pimpl->cpu_buft_list, pimpl->dev_input.buft_list, pimpl->dev_output.buft_list, buft_list_layer,
tn, ne, flags);
}
For non-looped models, loop_visit(tn.bid) == tn.bid, preserving existing behavior.
For looped models, loop_visit(18..23) yields 42..47, correctly retrieving the device placement assigned to the postlude layers.
- Lenguaje dominante
- C++
- Estrellas
- 1.2k
- Forks
- 80
- Merge medio
- 4 d 20 h
- PR fusionados (30 d)
- 2
Preparar el entorno
Este proyecto no incluye contenedor de desarrollo, Dockerfile ni guía de contribución, así que la configuración corre por tu cuenta: empieza por su README y consulta nuestra guía para la primera contribución para los pasos generales.
Primeros pasos
- Lee el issue completo y luego la guía de contribución del proyecto.
- Comenta en el issue que vas a ocuparte — evita que dos personas hagan lo mismo.
- Haz un fork del repositorio y trabaja en una rama.
- Abre un pull request que haga referencia al número del issue.
Más de Anbeeld/beellama.cpp
-
[BUG] KVarN assertion failure in build_inp_mem_hybrid_iswa() due to uninitialized self_kvarn_rotAbierto
Dificultad 2/5 1-3 horas Aptitud para principiantes 68/100
Anbeeld/beellama.cpp#205 ·
-
Dificultad 3/5 1-2 días Aptitud para principiantes 57/100
Anbeeld/beellama.cpp#204 ·
-
Dificultad 4/5 3-5 días Aptitud para principiantes 45/100
Anbeeld/beellama.cpp#203 · 2 comentarios ·
-
Dificultad 4/5 3-5 días Aptitud para principiantes 45/100
Anbeeld/beellama.cpp#202 · 1 comentario ·
-
Dificultad 4/5 3-5 días Aptitud para principiantes 48/100
Anbeeld/beellama.cpp#200 ·
Todos los issues de Anbeeld/beellama.cpp
Issues similares
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
CachyOS/packageinstaller#144 ·
Los mantenedores suelen responder en 1 día
-
documentation
Dificultad 1/5 Menos de una hora Aptitud para principiantes 82/100
TomerAberbach/profiler-md#951 ·
Los mantenedores suelen responder en 1 día
-
[Bug] Default key routing (JavaStringHash) picks a different partition than the Java client for about half of keysPosiblemente ocupada @0z5a la tomó hoy. Abierto
Dificultad 2/5 1-3 horas Aptitud para principiantes 82/100
apache/pulsar-client-go#1575 ·
Los mantenedores suelen responder en 2 días
-
Dificultad 2/5 1-3 horas Aptitud para principiantes 72/100
pocoproject/poco#5764 ·
Los mantenedores suelen responder en 2 días
-
bug cpp-core-guidelines low-priority
Dificultad 2/5 1-3 horas Aptitud para principiantes 62/100
NatronGitHub/openfx-supportext#164 ·
Los mantenedores suelen responder en 1 día