* ggml : FA with different K, V head sizes (CPU)
ggml-ci
* metal : add FA with HS=192
* metal : extend FA to support different K and V head sizes
ggml-ci
* metal : add FA vector kernels for heads K 192 and V 128
ggml-ci
* ggml : restrict op on other backends to equal head sizes
ggml-ci
* metal : optimize FA-vec kernel
ggml-ci
* metal : FA remove mq registers
* metal : improve MoE mul_mat_id condition
ggml-ci
* metal : fix comments + remove unnecessary addition
ggml-ci
* metal : avoid too much shared memory usage with mul_mat_id
ggml-ci
* vulkan: fix coopmat shader generation when cross-compiling
Previously the status of coopmat{,2} support isn't passed to the
vulkan-shaders-gen project building on the host, which leads to build
failure because of the cross-compiling code expecting coopmat{,2}
shaders that didn't get generated.
Fix this by passing the coopmat{,2} support status to vulkan-shaders
subproject.
Signed-off-by: Icenowy Zheng <uwu@icenowy.me>
* Only call coop-mat shaders once
* Fix whitespace
---------
Signed-off-by: Icenowy Zheng <uwu@icenowy.me>
Co-authored-by: bandoti <141645996+bandoti@users.noreply.github.com>
* Include speculative decoding stats when timings_per_token is true
New fields added to the `timings` object:
- draft_n : number of draft tokens generated
- draft_accepted_n : number of draft tokens accepted
- draft_accept_ratio: ratio of accepted/generated
* Remove redundant draft_accept_ratio var
* add draft acceptance rate to server console output
This patch enables usage of MMA when one of the
dimensions of the matrix(ie either M or N) is 1. This
is useful in case of token generation where N < 2.
The concept of 'GEMV Forwarding' is used where when one
of the matrix has a single row/column, the elements are
broadcasted, instead of using packing routine to prepack
the matrix elements.
This change results in 5% - 15% improvement in total
speed(ie all tokens/total time), across various batch
sizes. This is in comparision with the corresponding
dot product implementation.
The patch is tested with FP32 models of Meta-Lllama-3-8B,
Mistral-7B, Llama-2-7B-chat-hf on a IBM POWER10 machine.
Signed-off-by: Amrita H S <amritahs@linux.vnet.ibm.com>
* rpc : send hash when tensor data is above some fixed threshold
ref #10095
* rpc : put cache under $HOME/.cache/llama.cpp
* try to fix win32 build
* another try to fix win32 build
* remove llama as dependency
* server : Bump cpp-httplib to include AF_UNIX windows support
Signed-off-by: Piotr Stankiewicz <piotr.stankiewicz@docker.com>
* server : Allow running the server example on a unix socket
Signed-off-by: Piotr Stankiewicz <piotr.stankiewicz@docker.com>
---------
Signed-off-by: Piotr Stankiewicz <piotr.stankiewicz@docker.com>
* add edgellm model arch[conversation feature doesn't work]
* remove output.weight layer for edgellm arch
* [Model] update the name of the model
* update the name of model arch in convert gguf
* [Model] Refarctor the model arch into llama-model
* [Bug] Fix the bug in create attn kv
* [Code] Fix editorconfig erros
* [Code] Remove Trailing whitespace
* [Code] Remove Trailing whitespace
* [Code] Change the order of model arch in list
* [Code] Fix flake8 Lint errors
* Remove trailing white space
* [Code] Remove call in model arch
This change upstreams llamafile's cpu matrix
multiplication kernels for ppc64le ISA using MMA
builtins. This patch handles matrix multiplication
between quantised datatypes, block_q4_0 and
block_q8_0.
This change results in 5% - 50% improvement
in total speed(ie all tokens/total time), across
various batch sizes.
The patch is tested with Meta-Lllama-3-8B,
Mistral-7B, Llama-2-7B-chat-hf models on a
IBM POWER10 machine.
Signed-off-by: Amrita H S <amritahs@linux.vnet.ibm.com>
* [Fix] Compiling clip-quantize-cli and running it in a CUDA environment will cause ggml_fp16_to_fp32 to report an error when trying to access video memory. You need to switch to the CPU backend to run quantize.
After the fix, it will automatically run in the CPU backend and will no longer be bound to CUDA.
* [Fix]Roll back the signature and implementation of clip_model_load, and change the call in clip_model_quantize to clip_init.
* convert : fix squeeze for ssm_conv tensors
* convert : match ssm_conv tensors by type
---------
Co-authored-by: Francis Couture-Harpin <git@compilade.net>
The OOB calculation could be wrong if the last iteration was during one of
the unrolled loops. Adjust the unrolling counts to avoid this. Add a couple
new backend tests that hit this failure on NVIDIA GPUs.
Add verbose output to server_task_result_cmpl_final::to_json_oaicompat_chat_stream, making it conform with server_task_result_cmpl_final::to_json_oaicompat_chat, as well as the other to_json methods.
