]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/log
pkg/ggml/sources/llama.cpp
2 months agoHIP: remove the use of __HIP_PLATFORM_AMD__, explicitly support only AMD targets...
uvos [Tue, 29 Jul 2025 18:23:04 +0000 (20:23 +0200)]
HIP: remove the use of __HIP_PLATFORM_AMD__, explicitly support only AMD targets (#14945)

2 months agoHIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. (#14930)
uvos [Tue, 29 Jul 2025 15:44:30 +0000 (17:44 +0200)]
HIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. (#14930)

This is useful for testing for regressions on GCN with CDNA hardware.

With GGML_HIP_MMQ_MFMA=Off and GGML_CUDA_FORCE_MMQ=On we can conveniently test the GCN code path on CDNA. As CDNA is just GCN renamed with MFMA added and limited use ACC registers, this provides a good alternative for regression testing when GCN hardware is not available.

2 months agoHIP: Ignore unsupported unroll transformation in fattn-vec (#14931)
uvos [Tue, 29 Jul 2025 15:43:43 +0000 (17:43 +0200)]
HIP: Ignore unsupported unroll transformation in fattn-vec (#14931)

llvm with the amdgcn target dose not support unrolling loops with conditional break statements, when those statements can not be resolved at compile time. Similar to other places in GGML lets simply ignore this warning.

2 months agocommon : avoid logging partial messages (which can contain broken UTF-8 sequences...
kallewoof [Tue, 29 Jul 2025 15:05:38 +0000 (00:05 +0900)]
common : avoid logging partial messages (which can contain broken UTF-8 sequences) (#14937)

* bug-fix: don't attempt to log partial parsed messages to avoid crash due to unfinished UTF-8 sequences

2 months agoCANN: Add ggml_set_rows (#14943)
hipudding [Tue, 29 Jul 2025 14:36:43 +0000 (22:36 +0800)]
CANN: Add ggml_set_rows (#14943)

2 months agocuda : add softcap fusion (#14907)
Sigbjørn Skjæret [Tue, 29 Jul 2025 12:22:03 +0000 (14:22 +0200)]
cuda : add softcap fusion (#14907)

2 months agoserver-bench: make seed choice configurable (#14929)
Johannes Gäßler [Tue, 29 Jul 2025 08:40:50 +0000 (10:40 +0200)]
server-bench: make seed choice configurable (#14929)

* server-bench: make seed choice configurable

* Update scripts/server-bench.py

Co-authored-by: Sigbjørn Skjæret <redacted>
* Update scripts/server-bench.py

Co-authored-by: Sigbjørn Skjæret <redacted>
* fix error formatting

* Update scripts/server-bench.py

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Sigbjørn Skjæret <redacted>
2 months agoCUDA: add roll (#14919)
Aman Gupta [Tue, 29 Jul 2025 06:45:18 +0000 (14:45 +0800)]
CUDA: add roll (#14919)

* CUDA: add roll

* Make everything const, use __restrict__

2 months agoopencl : add ops docs (#14910)
lhez [Mon, 28 Jul 2025 16:50:17 +0000 (09:50 -0700)]
opencl : add ops docs (#14910)

2 months agotest-backend-ops : extend test case filtering (#14865)
Leonard Mosescu [Mon, 28 Jul 2025 16:04:27 +0000 (09:04 -0700)]
test-backend-ops : extend test case filtering (#14865)

* Extend test case filtering

1. Allow passing multiple (comma-separated?) ops to test-backend-ops. This can be convenient when working on a set of ops, when you'd want to test them together (but without having to run every single op). For example:

`test-backend-ops.exe test -o "ADD,RMS_NORM,ROPE,SILU,SOFT_MAX"`

2. Support full test-case variation string in addition to basic op names. This would make it easy to select a single variation, either for testing or for benchmarking. It can be particularly useful for profiling a particular variation (ex. a CUDA kernel), for example:

`test-backend-ops.exe perf -b CUDA0 -o "MUL_MAT(type_a=f16,type_b=f32,m=4096,n=512,k=14336,bs=[1,1],nr=[1,1],per=[0,1,2,3],v=2)"`

These two can be combined. As the current `-o`, this change doesn't try to detect/report an error if an filter doesn't name existing ops (ex. misspelled)

* Updating the usage help text

* Update tests/test-backend-ops.cpp

2 months agollama-bench : use local GPUs along with RPC servers (#14917)
Radoslav Gerganov [Mon, 28 Jul 2025 15:59:04 +0000 (18:59 +0300)]
llama-bench : use local GPUs along with RPC servers (#14917)

Currently if RPC servers are specified with '--rpc' and there is a local
GPU available (e.g. CUDA), the benchmark will be performed only on the
RPC device(s) but the backend result column will say "CUDA,RPC" which is
incorrect. This patch is adding all local GPU devices and makes
llama-bench consistent with llama-cli.

2 months agoggml-cpu : deduplicate scalar implementations (#14897)
xctan [Mon, 28 Jul 2025 15:40:24 +0000 (23:40 +0800)]
ggml-cpu : deduplicate scalar implementations (#14897)

* remove redundant code in riscv

* remove redundant code in arm

* remove redundant code in loongarch

* remove redundant code in ppc

* remove redundant code in s390

* remove redundant code in wasm

* remove redundant code in x86

* remove fallback headers

* fix x86 ggml_vec_dot_q8_0_q8_0

2 months agoSYCL: Add set_rows support for quantized types (#14883)
Akarshan Biswas [Mon, 28 Jul 2025 15:02:15 +0000 (20:32 +0530)]
SYCL: Add set_rows support for quantized types  (#14883)

* SYCL: Add set_rows support for quantized types

This commit adds support for GGML_OP_SET_ROWS operation for various
quantized tensor types (Q8_0, Q5_1, Q5_0, Q4_1, Q4_0, IQ4_NL) and BF16
type in the SYCL backend.

The quantization/dequantization copy kernels were moved from cpy.cpp
to cpy.hpp to make them available for set_rows.cpp.

This addresses part of the TODOs mentioned in the code.

* Use get_global_linear_id() instead

ggml-ci

* Fix formatting

ggml-ci

* Use const for ne11 and size_t variables in set_rows_sycl_q

ggml-ci

* Increase block size for q kernel to 256

ggml-ci

* Cleanup imports

* Add float.h to cpy.hpp

2 months agomtmd : add support for Voxtral (#14862)
Xuan-Son Nguyen [Mon, 28 Jul 2025 13:01:48 +0000 (15:01 +0200)]
mtmd : add support for Voxtral (#14862)

* mtmd : add support for Voxtral

* clean up

* fix python requirements

* add [BEGIN_AUDIO] token

* also support Devstral conversion

* add docs and tests

* fix regression for ultravox

* minor coding style improvement

* correct project activation fn

* Apply suggestions from code review

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Sigbjørn Skjæret <redacted>
2 months agoCUDA: fix pointer incrementation in FA (#14916)
Johannes Gäßler [Mon, 28 Jul 2025 12:30:22 +0000 (14:30 +0200)]
CUDA: fix pointer incrementation in FA (#14916)

2 months agomodel : add support for SmallThinker series (#14898)
Dongliang Wei [Mon, 28 Jul 2025 11:47:00 +0000 (19:47 +0800)]
model : add support for SmallThinker series (#14898)

* support smallthinker

* support 20b softmax, 4b no sliding window

* new build_moe_ffn_from_probs, and can run 4b

* fix 4b rope bug

* fix python type check

* remove is_moe judge

* remove set_dense_start_swa_pattern function and modify set_swa_pattern function

* trim trailing whitespace

* remove get_vocab_base of SmallThinkerModel in convert_hf_to_gguf.py

Co-authored-by: Sigbjørn Skjæret <redacted>
* better whitespace

Apply suggestions from code review

Co-authored-by: Sigbjørn Skjæret <redacted>
* use GGML_ASSERT for expert count validation

Co-authored-by: Sigbjørn Skjæret <redacted>
* Improve null pointer check for probs

Co-authored-by: Sigbjørn Skjæret <redacted>
* use template parameter for SWA attention logic

* better whitespace

Co-authored-by: Georgi Gerganov <redacted>
* move the creation of inp_out_ids before the layer loop

* remove redundant judge for probs

---------

Co-authored-by: Sigbjørn Skjæret <redacted>
Co-authored-by: Georgi Gerganov <redacted>
2 months agosycl: refactor quantization to q8_1 (#14815)
Alberto Cabrera Pérez [Mon, 28 Jul 2025 10:05:53 +0000 (11:05 +0100)]
sycl: refactor quantization to q8_1  (#14815)

* sycl: quantization to q8_1 refactor

* Refactored src1 copy logic in op_mul_mat

2 months agoops : update BLAS (#14914)
Georgi Gerganov [Mon, 28 Jul 2025 08:01:03 +0000 (11:01 +0300)]
ops : update BLAS (#14914)

2 months agoops : update Metal (#14912)
Georgi Gerganov [Mon, 28 Jul 2025 05:22:56 +0000 (08:22 +0300)]
ops : update Metal (#14912)

2 months agosync : ggml
Georgi Gerganov [Mon, 28 Jul 2025 05:14:20 +0000 (08:14 +0300)]
sync : ggml

2 months agocmake : Indent ggml-config.cmake (ggml/1310)
Kai Pastor [Thu, 24 Jul 2025 17:58:02 +0000 (19:58 +0200)]
cmake : Indent ggml-config.cmake (ggml/1310)

2 months agoquantize : update README.md (#14905)
Ed Addario [Sun, 27 Jul 2025 21:31:11 +0000 (22:31 +0100)]
quantize : update README.md (#14905)

* Update README.md

* Fix trailing whitespace

* Update README.md

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Sigbjørn Skjæret <redacted>
2 months agovulkan: add ops docs (#14900)
Ruben Ortlam [Sun, 27 Jul 2025 13:33:08 +0000 (15:33 +0200)]
vulkan: add ops docs (#14900)

2 months agoSYCL: add ops doc (#14901)
Akarshan Biswas [Sun, 27 Jul 2025 12:22:58 +0000 (17:52 +0530)]
SYCL: add ops doc (#14901)

2 months agollama : clarify comment about pp and tg graphs [no ci] (#14895)
Daniel Bevenius [Sun, 27 Jul 2025 10:10:51 +0000 (12:10 +0200)]
llama : clarify comment about pp and tg graphs [no ci] (#14895)

* llama : clarify comment about pp and tg graphs [no ci]

This commit clarifies the comment in `llama-context.cpp` regarding the
prefill prompt (pp), and token generation (tg) graphs.

The motivation for this is that I've struggled to remember these and had
to look them up more than once, so I thought it would be helpful to add
a comment that makes it clear what these stand for.

* squash! llama : clarify comment about pp and tg graphs [no ci]

Change "pp" to "prompt processing".

2 months agovulkan : add fp16 support for the conv_2d kernel (#14872)
Erik Scholz [Sun, 27 Jul 2025 10:04:33 +0000 (12:04 +0200)]
vulkan : add fp16 support for the conv_2d kernel (#14872)

* add f16 to conv_2d testing
* weaken conv2d test error threshold

2 months agovulkan: skip empty set_rows to avoid invalid API usage (#14860)
Jeff Bolz [Sun, 27 Jul 2025 09:05:34 +0000 (04:05 -0500)]
vulkan: skip empty set_rows to avoid invalid API usage (#14860)

2 months agomodel : make rope_yarn_log_mul optional for deepseek2 (#14896)
Gabriel Larson [Sun, 27 Jul 2025 08:18:37 +0000 (03:18 -0500)]
model : make rope_yarn_log_mul optional for deepseek2 (#14896)

* make rope_yarn_log_mul optional for deepseek2

* default rope_yarn_log_mul = 0.0f

2 months agollama : fix kq_scale for the attention layers of PLaMo2 (#14892)
Shunta Saito [Sun, 27 Jul 2025 07:38:44 +0000 (16:38 +0900)]
llama : fix kq_scale for the attention layers of PLaMo2 (#14892)

* Fix dimensions for expand

* Change dimensions to copy states to cache

* Fix the default value for plamo2 conversion

* Fix scale given to build_attn

* Update src/llama-model.cpp

Co-authored-by: Sigbjørn Skjæret <redacted>
* Update src/llama-model.cpp

Co-authored-by: Sigbjørn Skjæret <redacted>
* Update src/llama-model.cpp

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Sigbjørn Skjæret <redacted>
2 months agoDocs: add instructions for adding backends (#14889)
Aman Gupta [Sun, 27 Jul 2025 01:36:43 +0000 (09:36 +0800)]
Docs: add instructions for adding backends (#14889)

2 months agoHIP: Enable Matrix cores for MMQ Kernels, Enable stream-K for CDNA 3 (#14624)
deepsek [Sat, 26 Jul 2025 22:28:14 +0000 (18:28 -0400)]
HIP: Enable Matrix cores for MMQ Kernels, Enable stream-K for CDNA 3 (#14624)

This commit adds support for MFMA instructions to MMQ. CDNA1/GFX908 CDNA2/GFX90a and CDNA3/GFX942 are supported by the MFMA-enabled code path added by this commit. The code path and stream-k is only enabled on CDNA3 for now as it fails to outperform blas in all cases on the other devices.
Blas is currently only consistently outperformed on CDNA3 due to issues in the amd-provided blas libraries.
This commit also improves the awareness of MMQ towards different warp sizes and as a side effect improves the performance of all quant formats besides q4_0 and q4_1, which regress slightly, on GCN gpus.

2 months agoCANN: Implement GLU ops (#14884)
hipudding [Sat, 26 Jul 2025 09:56:18 +0000 (17:56 +0800)]
CANN: Implement GLU ops (#14884)

Implement REGLU, GEGLU, SWIGLU ops according to #14158

2 months agomusa: fix build warnings (unused variable) (#14869)
R0CKSTAR [Sat, 26 Jul 2025 02:36:02 +0000 (10:36 +0800)]
musa: fix build warnings (unused variable) (#14869)

Signed-off-by: Xiaodong Ye <redacted>
2 months agoggml-cpu : disable GGML_NNPA by default due to instability (#14880)
Aaron Teo [Fri, 25 Jul 2025 17:09:03 +0000 (01:09 +0800)]
ggml-cpu : disable GGML_NNPA by default due to instability (#14880)

* docs: update s390x document for sentencepiece

Signed-off-by: Aaron Teo <redacted>
(cherry picked from commit e086c5e3a7ab3463d8e0906efcfa39352db0a48d)

* docs: update huggingface links + reword

Signed-off-by: Aaron Teo <redacted>
(cherry picked from commit 8410b085ea8c46e22be38266147a1e94757ef108)

* ggml-cpu: disable ggml-nnpa compile flag by default

fixes #14877

Signed-off-by: Aaron Teo <redacted>
(cherry picked from commit 412f4c7c88894b8f55846b4719c76892a23cfe09)

* docs: update s390x build docs to reflect nnpa disable

Signed-off-by: Aaron Teo <redacted>
(cherry picked from commit c1eeae1d0c2edc74ab9fbeff2707b0d357cf0b4d)

---------

Signed-off-by: Aaron Teo <redacted>
2 months agometal: SSM_SCAN performance (#14743)
Gabe Goodhart [Fri, 25 Jul 2025 16:47:39 +0000 (10:47 -0600)]
metal: SSM_SCAN performance (#14743)

* feat: Add s_off as a parameter in the args struct

This may not be necessary, but it more closely mirrors the CUDA kernel

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* perf: Parallelize mamba2 SSM_SCAN metal kernel over d_state

This is a first attempt at optimizing the metal kernel. The changes here
are:

- Launch the kernel with a thread group of size d_state
- Use simd groups and shared memory to do the summation for the y
  computation

When tested with G4 tiny preview, this shows roughly a 3x speedup on
prefill and 15% speedup on decode.

Signed-off-by: Gabe Goodhart <redacted>
* fix: Update logic to correctly do the multi-layer parallel sum

Signed-off-by: Gabe Goodhart <redacted>
* fix: Correctly size the shared memory bufer and assert expected size relationships

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* refactor: Compute block offsets once rather than once per token

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* feat: Use local variable for state recursion

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* feat: Use a secondary simd_sum instead of a for loop

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* feat: Add assertion and comment about relationship between simd size and num simd groups

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* feat: Parallelize of d_state for mamba-1

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* feat: Parallel sum in SSM_CONV

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
* Revert "feat: Parallel sum in SSM_CONV"

After discussion with @compilade, the size of the parallelism here is
not worth the cost in complexity or overhead of the parallel for.

https://github.com/ggml-org/llama.cpp/pull/14743#discussion_r2223395357

This reverts commit 16bc059660c1c59e566628201c0ca2c20c9f4bc3.

Signed-off-by: Gabe Goodhart <redacted>
* refactor: Simplify shared memory sizing

Branch: GraniteFourPerf

Signed-off-by: Gabe Goodhart <redacted>
Co-Authored-By: Georgi Gerganov <redacted>
---------

Signed-off-by: Gabe Goodhart <redacted>
Co-authored-by: Georgi Gerganov <redacted>
2 months agoopencl: add fused `rms_norm_mul` (#14841)
lhez [Fri, 25 Jul 2025 15:12:13 +0000 (08:12 -0700)]
opencl: add fused `rms_norm_mul` (#14841)

* opencl: add fused `rms_norm` + `mul`

* opencl: improve workgroup size for `rms_norm_mul`

2 months agodocs : update HOWTO‑add‑model.md for ModelBase and new model classes (#14874)
wooksong [Fri, 25 Jul 2025 14:25:05 +0000 (23:25 +0900)]
docs : update HOWTO‑add‑model.md for ModelBase and new model classes (#14874)

This patch updates the example in docs/development/HOWTO-add-model.md to
reflect recent changes after `TextModel` and `MmprojModel` were introduced.

It replaces the outdated `Model` base class with `TextModel` or `MmprojModel`
and updates the registration example accordingly.

Signed-off-by: Wook Song <redacted>
2 months agoggml : remove invalid portPos specifiers from dot files (#14838)
Oliver Simons [Fri, 25 Jul 2025 11:29:57 +0000 (13:29 +0200)]
ggml : remove invalid portPos specifiers from dot files (#14838)

Neither "g" nor "x" are valid portPos specifiers per the official
[graphviz documents](https://graphviz.org/docs/attr-types/portPos/):

> If a compass point is used, it must have the form "n","ne","e","se","s","sw","w","nw","c","_".

I tested locally for it to fall back to default portPos specifier if an
invalid portPos is specified. As a consequence, we can remove associated
code.

2 months agocontext : restore preemptive sched reset when LLAMA_SET_ROWS=0 (#14870)
Georgi Gerganov [Fri, 25 Jul 2025 11:28:06 +0000 (14:28 +0300)]
context : restore preemptive sched reset when LLAMA_SET_ROWS=0 (#14870)

ggml-ci

2 months agomtmd : fix 32-bit narrowing issue in export-lora and mtmd clip (#14503)
kiwi [Fri, 25 Jul 2025 11:08:04 +0000 (19:08 +0800)]
mtmd : fix 32-bit narrowing issue in export-lora and mtmd clip (#14503)

* [fix] Fix 32-bit narrowing issue in export-lora and mtmd clip

* Update export-lora.cpp

* Update clip.cpp

* Update export-lora.cpp

* format: use space to replace tab

2 months agorpc : check for null buffers in get/set/copy tensor endpoints (#14868)
Chris Rohlf [Fri, 25 Jul 2025 10:17:02 +0000 (06:17 -0400)]
rpc : check for null buffers in get/set/copy tensor endpoints (#14868)

2 months agosched : fix multiple evaluations of the same graph with pipeline parallelism (#14855)
Diego Devesa [Fri, 25 Jul 2025 08:07:26 +0000 (01:07 -0700)]
sched : fix multiple evaluations of the same graph with pipeline parallelism (#14855)

ggml-ci

2 months agomusa: upgrade musa sdk to rc4.2.0 (#14498)
R0CKSTAR [Thu, 24 Jul 2025 19:05:37 +0000 (03:05 +0800)]
musa: upgrade musa sdk to rc4.2.0 (#14498)

* musa: apply mublas API changes

Signed-off-by: Xiaodong Ye <redacted>
* musa: update musa version to 4.2.0

Signed-off-by: Xiaodong Ye <redacted>
* musa: restore MUSA graph settings in CMakeLists.txt

Signed-off-by: Xiaodong Ye <redacted>
* musa: disable mudnnMemcpyAsync by default

Signed-off-by: Xiaodong Ye <redacted>
* musa: switch back to non-mudnn images

Signed-off-by: Xiaodong Ye <redacted>
* minor changes

Signed-off-by: Xiaodong Ye <redacted>
* musa: restore rc in docker image tag

Signed-off-by: Xiaodong Ye <redacted>
---------

Signed-off-by: Xiaodong Ye <redacted>
2 months agosync : ggml
Georgi Gerganov [Thu, 24 Jul 2025 15:30:33 +0000 (18:30 +0300)]
sync : ggml

ggml-ci

2 months agocmake : fix usage issues (ggml/1257)
Kai Pastor [Tue, 22 Jul 2025 18:13:21 +0000 (20:13 +0200)]
cmake : fix usage issues (ggml/1257)

* CMake config: Create target only once

Fix error on repeated find_package(ggml).
For simplicity, check only for the top-level ggml::ggml.

* CMake config: Add CUDA link libs

* CMake config: Add OpenCL link libs

* CMake config: Use canonical find_dependency

Use set and append to control link lib variables.
Apply more $<LINK_ONLY...>.

* CMake config: Wire OpenMP dependency

2 months agoggml-cpu : remove stdlib include from repack.cpp (ggml/1276)
Daniel Bevenius [Mon, 21 Jul 2025 13:53:12 +0000 (15:53 +0200)]
ggml-cpu : remove stdlib include from repack.cpp (ggml/1276)

This commit removes the inclusion of `<cstdlib>`.

The motivation for this change is that this source file does not seem to
use any functions from this header and the comment about `qsort` is a
little misleading/confusing.

2 months agocontext : perform output reorder lazily upon access after sync (#14853)
Georgi Gerganov [Thu, 24 Jul 2025 13:31:48 +0000 (16:31 +0300)]
context : perform output reorder lazily upon access after sync (#14853)

* context : perform output reorder after lazily upon access after sync

ggml-ci

* cont : add TODO

2 months agochat : fix kimi-k2 chat template (#14852)
Xuan-Son Nguyen [Thu, 24 Jul 2025 11:59:56 +0000 (13:59 +0200)]
chat : fix kimi-k2 chat template (#14852)

2 months agosycl: fixed semantics of block offset calculation (#14814)
Alberto Cabrera Pérez [Thu, 24 Jul 2025 10:09:57 +0000 (11:09 +0100)]
sycl: fixed semantics of block offset calculation (#14814)

2 months agollama : fix MiniCPM inference after Granite Four changes (#14850)
yummy [Thu, 24 Jul 2025 09:50:51 +0000 (17:50 +0800)]
llama : fix MiniCPM inference after Granite Four changes (#14850)

MiniCPM models use the llm_build_granite constructor which was changed
in the Granite Four PR to use hparams.rope_finetuned instead of a
use_rope parameter. MiniCPM models need rope enabled by default.

Fixes inference from gibberish to correct responses.

2 months agodocs: add libcurl-dev install hint for Linux distros (#14801)
Pouya [Thu, 24 Jul 2025 09:26:44 +0000 (12:26 +0300)]
docs: add libcurl-dev install hint for Linux distros (#14801)

* docs: add libcurl-dev install hint for Linux distros

Signed-off-by: PouyaGhahramanian <redacted>
* Update docs/build.md

---------

Signed-off-by: PouyaGhahramanian <redacted>
Co-authored-by: Xuan-Son Nguyen <redacted>
2 months agometal : fix fusion across different encoders (#14849)
Georgi Gerganov [Thu, 24 Jul 2025 07:24:05 +0000 (10:24 +0300)]
metal : fix fusion across different encoders (#14849)

* metal : fix fusion across different encoders

ggml-ci

* cont : add assertion

ggml-ci

2 months agosycl: fix undefined variable in work group size check (#14843)
Donghyeon Jeong [Thu, 24 Jul 2025 04:50:41 +0000 (13:50 +0900)]
sycl: fix undefined variable in work group size check (#14843)

2 months agoconvert : text-only support for GLM-4.1V-9B-Thinking (#14823)
jacekpoplawski [Wed, 23 Jul 2025 21:23:57 +0000 (23:23 +0200)]
convert : text-only support for GLM-4.1V-9B-Thinking (#14823)

* use language_model part only, ignore visual layers

* fix rope_dim calculation

2 months agoCUDA: fix overflow in FA, tune performance (#14840)
Johannes Gäßler [Wed, 23 Jul 2025 19:43:25 +0000 (21:43 +0200)]
CUDA: fix overflow in FA, tune performance (#14840)

2 months agoCUDA: fix compilation with GGML_CUDA_F16 (#14837)
Johannes Gäßler [Wed, 23 Jul 2025 16:22:30 +0000 (18:22 +0200)]
CUDA: fix compilation with GGML_CUDA_F16 (#14837)

2 months agoci : correct label refactor->refactoring (#14832)
Sigbjørn Skjæret [Wed, 23 Jul 2025 12:27:54 +0000 (14:27 +0200)]
ci : correct label refactor->refactoring (#14832)

2 months agoCUDA: fix quantized KV cache + multiple sequences (#14822)
Johannes Gäßler [Wed, 23 Jul 2025 10:35:53 +0000 (12:35 +0200)]
CUDA: fix quantized KV cache + multiple sequences (#14822)

* CUDA: fix quantized KV cache + multiple sequences

* Update ggml/src/ggml-cuda/fattn-common.cuh

Co-authored-by: Georgi Gerganov <redacted>
---------

Co-authored-by: Georgi Gerganov <redacted>
2 months agotests : add non-cont K,V FA tests
Georgi Gerganov [Fri, 18 Jul 2025 10:36:27 +0000 (13:36 +0300)]
tests : add non-cont K,V FA tests

ggml-ci

2 months agomemory : handle saving/loading null layers in recurrent memory (#14675)
l3utterfly [Wed, 23 Jul 2025 08:16:41 +0000 (16:16 +0800)]
memory : handle saving/loading null layers in recurrent memory (#14675)

* Update llama-memory-recurrent.cpp

handle saving/loading null layers in recurrent memory

* fixed styling issues and updated comments

* fix styling issue

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Sigbjørn Skjæret <redacted>
2 months agoggml: fix loongarch quantize_row_q8_1 error (#14827)
lixing-star [Wed, 23 Jul 2025 06:39:51 +0000 (14:39 +0800)]
ggml: fix loongarch quantize_row_q8_1 error (#14827)

2 months agoCANN: weight format to NZ for Ascend310P3 (#14407)
chen fan [Wed, 23 Jul 2025 03:58:00 +0000 (11:58 +0800)]
CANN: weight format to NZ for Ascend310P3 (#14407)

* weight format to nz for 310p

* remove quant weight format to nz

* clean code

* fix

* make the conditions for converting weights to NZ format consistent

* clean code

2 months agoCUDA: add fused rms norm (#14800)
Aman Gupta [Wed, 23 Jul 2025 01:25:42 +0000 (09:25 +0800)]
CUDA: add fused rms norm (#14800)

2 months agoggml : model card yaml tab->2xspace (#14819)
Csaba Kecskemeti [Tue, 22 Jul 2025 16:29:43 +0000 (09:29 -0700)]
ggml : model card yaml tab->2xspace (#14819)

2 months agovulkan: fix rms_norm_mul to handle broadcasting dim0 (#14817)
Jeff Bolz [Tue, 22 Jul 2025 15:35:21 +0000 (10:35 -0500)]
vulkan: fix rms_norm_mul to handle broadcasting dim0 (#14817)

2 months agollama : add model type detection for rwkv7 7B&14B (#14816)
Molly Sophia [Tue, 22 Jul 2025 15:01:29 +0000 (23:01 +0800)]
llama : add model type detection for rwkv7 7B&14B (#14816)

Signed-off-by: Molly Sophia <redacted>
2 months agoimatrix: add option to display importance score statistics for a given imatrix file...
Ed Addario [Tue, 22 Jul 2025 12:33:37 +0000 (13:33 +0100)]
imatrix: add option to display importance score statistics for a given imatrix file (#12718)

* Add --show-statistics option

* Add --show-statistics logic

* Add tensor name parsing

* Tidy output format

* Fix typo in title

* Improve tensor influence ranking

* Add better statistics

* Change statistics' sort order

* Add Cosine Similarity

* Add header search path

* Change header search path to private

* Add weighted statistics per layer

* Update report title

* Refactor compute_statistics out of main

* Refactor compute_cossim out of load_imatrix

* Refactor compute_statistics out of load_imatrix

* Move imatrix statistics calculation into its own functions

* Add checks and validations

* Remove unnecessary include directory

* Rename labels

* Add m_stats getter and refactor compute_statistics out of load_imatrix

* Refactor variable names

* Minor cosmetic change

* Retrigger checks (empty commit)

* Rerun checks (empty commit)

* Fix unnecessary type promotion

Co-authored-by: compilade <redacted>
* Reverting change to improve code readability

* Rerun checks (empty commit)

* Rerun checks (empty commit)

* Rerun checks - third time's the Charm 🤞 (empty commit)

* Minor cosmetic change

* Update README

* Fix typo

* Update README

* Rerun checks (empty commit)

* Re-implement changes on top of #9400

* Update README.md

* Update README

* Update README.md

Co-authored-by: compilade <redacted>
* Update README.md

Co-authored-by: compilade <redacted>
* Update README.md

* Remove duplicate option in print_usage()

* Update README.md

* Update README.md

Co-authored-by: compilade <redacted>
* Update README.md

Co-authored-by: compilade <redacted>
* Remove input check

* Remove commented out code

---------

Co-authored-by: compilade <redacted>
2 months agoMtmd: add a way to select device for vision encoder (#14236)
stduhpf [Tue, 22 Jul 2025 10:51:03 +0000 (12:51 +0200)]
Mtmd: add a way to select device for vision encoder (#14236)

* Mtmd: add a way to select device for vision encoder

* simplify

* format

* Warn user if manual device selection failed

* initialize backend to nullptr

2 months agocuda : implement bf16 cpy ops and enable bf16 cont (#14763)
Sigbjørn Skjæret [Tue, 22 Jul 2025 10:33:10 +0000 (12:33 +0200)]
cuda : implement bf16 cpy ops and enable bf16 cont (#14763)

* implement bf16 cpy ops and enable bf16 cont

* deduplicate copy functions

* deduplicate checks

2 months agoopencl: remove unreachable `return` (#14806)
lhez [Tue, 22 Jul 2025 06:53:30 +0000 (23:53 -0700)]
opencl: remove unreachable `return` (#14806)

2 months agoserver : allow setting `--reverse-prompt` arg (#14799)
Molly Sophia [Tue, 22 Jul 2025 01:24:22 +0000 (09:24 +0800)]
server : allow setting `--reverse-prompt` arg (#14799)

Signed-off-by: Molly Sophia <redacted>
2 months agocuda: remove linking to cublasLt (#14790)
R0CKSTAR [Mon, 21 Jul 2025 23:45:26 +0000 (07:45 +0800)]
cuda: remove linking to cublasLt (#14790)

Signed-off-by: Xiaodong Ye <redacted>
2 months agoopencl: fix `im2col` when `KW!=KH` (#14803)
Sigbjørn Skjæret [Mon, 21 Jul 2025 20:55:10 +0000 (22:55 +0200)]
opencl: fix `im2col` when `KW!=KH` (#14803)

2 months agoopencl: add conv2d kernel (#14403)
rmatif [Mon, 21 Jul 2025 17:03:19 +0000 (19:03 +0200)]
opencl: add conv2d kernel (#14403)

* add conv2d kernel

* fix trailing whitespace

* whitespace fixe

* handle f16 input and f16 kernel, more opt

* resolve conflicts

* use enqueue_ndrange_kernel

2 months agosycl: Fix im2col (#14797)
Romain Biessy [Mon, 21 Jul 2025 16:39:29 +0000 (18:39 +0200)]
sycl: Fix im2col (#14797)

2 months agokleidiai: add support for get_rows (#14676)
Charles Xu [Mon, 21 Jul 2025 13:49:52 +0000 (15:49 +0200)]
kleidiai: add support for get_rows (#14676)

* kleidiai: add support for get_rows

* apply fixes based on code review

* apply more fixes based on code review

2 months agodocs : fix backends table in README.md (#14796)
Radoslav Gerganov [Mon, 21 Jul 2025 12:03:49 +0000 (15:03 +0300)]
docs : fix backends table in README.md (#14796)

2 months agovulkan/cuda: Fix im2col when KW!=KH (#14789)
Jeff Bolz [Mon, 21 Jul 2025 11:35:40 +0000 (06:35 -0500)]
vulkan/cuda: Fix im2col when KW!=KH (#14789)

The tid is decomposed into "ow + ky*OW + kx*OW*KH". Change "ksize" to match.

2 months agollama : fix `--reverse-prompt` crashing issue (#14794)
Molly Sophia [Mon, 21 Jul 2025 09:38:36 +0000 (17:38 +0800)]
llama : fix `--reverse-prompt` crashing issue (#14794)

Signed-off-by: Molly Sophia <redacted>
2 months agoserver : add parse_special option to /tokenize endpoint (#14783)
IsaacDynamo [Mon, 21 Jul 2025 07:24:51 +0000 (09:24 +0200)]
server : add parse_special option to /tokenize endpoint (#14783)

2 months agodocs : fix link for tools/perplexity in README.md (#14780)
Aman Gupta [Sun, 20 Jul 2025 18:13:47 +0000 (02:13 +0800)]
docs : fix link for tools/perplexity in README.md (#14780)

2 months agoDocumentation: Further revisions to the Vulkan section in build.md (#14785)
rspOverflow [Sun, 20 Jul 2025 16:55:32 +0000 (23:55 +0700)]
Documentation: Further revisions to the Vulkan section in build.md (#14785)

* Documentation: Revised and further improved the Vulkan instructions for Linux users in build.md.

* Minor: Revise step 2 of the Vulkan instructions for Linux users in build.md

2 months agoClang-format: local files first + fix BinPacking (#14779)
Aman Gupta [Sun, 20 Jul 2025 11:42:34 +0000 (19:42 +0800)]
Clang-format: local files first + fix BinPacking (#14779)

2 months agoContrib: add 0cc4m as codeowner for Vulkan backend (#14775)
0cc4m [Sat, 19 Jul 2025 20:47:21 +0000 (22:47 +0200)]
Contrib: add 0cc4m as codeowner for Vulkan backend (#14775)

2 months agoggml: adds CONV_2D op and direct GEMM Vulkan implementation (#14316)
Ervin Áron Tasnádi [Sat, 19 Jul 2025 19:59:08 +0000 (21:59 +0200)]
ggml: adds CONV_2D op and direct GEMM Vulkan implementation (#14316)

* ggml/ggml-vulkan/test-backend-ops: adds CONV_2D for Vulkan

* ggml-vulkan: adds f32 scalar shader to compute 2D convolution directly
with gemm (no need for im2col),

* test-backend-ops: adds test_case_ref to check the validity/performance of ops
against reference implementations having different graphs, adds tests

* * Performance fixes: minimized branch divergence, uses collectives to
  eliminate redundant calculation, macros removed.

* Kernel shared memory size check

* Updates test-backend-ops to support graphs for performance
  measurement.

* * Apple/Win32 compile errors fixed

* Subgroup size used to determine tile size -> fixes llvmpipe errors.

* Collectives disabled by default.

* Intel support is disabled as the performance is poor.

* Conv2d enabled for Intel with disabled collectives, disabled for Apple

* test-backend-ops modifications are reverted

* Trailing spaces and missing override fixed.

* Triggering pipeline relaunch.

* Code formatted with .clang-format.

2 months agoimatrix : use GGUF to store importance matrices (#9400)
compilade [Sat, 19 Jul 2025 16:51:22 +0000 (12:51 -0400)]
imatrix : use GGUF to store importance matrices (#9400)

* imatrix : allow processing multiple chunks per batch

* perplexity : simplify filling the batch

* imatrix : fix segfault when using a single chunk per batch

* imatrix : use GGUF to store imatrix data

* imatrix : fix conversion problems

* imatrix : use FMA and sort tensor names

* py : add requirements for legacy imatrix convert script

* perplexity : revert changes

* py : include imatrix converter requirements in toplevel requirements

* imatrix : avoid using designated initializers in C++

* imatrix : remove unused n_entries

* imatrix : allow loading mis-ordered tensors

Sums and counts tensors no longer need to be consecutive.

* imatrix : more sanity checks when loading multiple imatrix files

* imatrix : use ggml_format_name instead of std::string concatenation

Co-authored-by: Xuan Son Nguyen <redacted>
* quantize : use unused imatrix chunk_size with LLAMA_TRACE

* common : use GGUF for imatrix output by default

* imatrix : two-way conversion between old format and GGUF

* convert : remove imatrix to gguf python script

* imatrix : use the function name in more error messages

* imatrix : don't use FMA explicitly

This should make comparisons between the formats easier
because this matches the behavior of the previous version.

* imatrix : avoid returning from void function save_imatrix

* imatrix : support 3d tensors with MUL_MAT

* quantize : fix dataset name loading from gguf imatrix

* common : move string_remove_suffix from quantize and imatrix

Co-authored-by: Sigbjørn Skjæret <redacted>
* imatrix : add warning when legacy format is written

* imatrix : warn when writing partial data, to help guess dataset coverage

Also make the legacy format store partial data
by using neutral values for missing data.
This matches what is done at read-time for the new format,
and so should get the same quality in case the old format is still used.

* imatrix : avoid loading model to convert or combine imatrix

* imatrix : avoid using imatrix.dat in README

---------

Co-authored-by: Xuan Son Nguyen <redacted>
Co-authored-by: Sigbjørn Skjæret <redacted>
2 months agovulkan: Add logging for bf16 features to ggml_vk_print_gpu_info (#13274) (#14707)
Peter0x44 [Sat, 19 Jul 2025 15:58:03 +0000 (16:58 +0100)]
vulkan: Add logging for bf16 features to ggml_vk_print_gpu_info (#13274) (#14707)

2 months agoVulkan: Fix fprintf format-security warning (#14770)
0cc4m [Sat, 19 Jul 2025 15:47:53 +0000 (17:47 +0200)]
Vulkan: Fix fprintf format-security warning (#14770)

2 months agoDocumentation: Update build.md's Vulkan section (#14736)
rspOverflow [Sat, 19 Jul 2025 10:18:36 +0000 (17:18 +0700)]
Documentation: Update build.md's Vulkan section (#14736)

* Documentation: Rewrote and updated the "Without docker" portion of the Vulkan backend build documentation.

* Documentation: Reorganize build.md's Vulkan section.

2 months agosync : ggml
Georgi Gerganov [Sat, 19 Jul 2025 08:46:12 +0000 (11:46 +0300)]
sync : ggml

2 months agometal : fuse add, mul + add tests (#14596)
Georgi Gerganov [Fri, 18 Jul 2025 17:37:26 +0000 (20:37 +0300)]
metal : fuse add, mul + add tests (#14596)

ggml-ci

2 months agograph : fix graph reuse reset of params (#14760)
Georgi Gerganov [Fri, 18 Jul 2025 17:08:33 +0000 (20:08 +0300)]
graph : fix graph reuse reset of params (#14760)

ggml-ci

2 months agoparallel : add option for different RNG seeds (#14757)
Georgi Gerganov [Fri, 18 Jul 2025 14:33:41 +0000 (17:33 +0300)]
parallel : add option for different RNG seeds (#14757)

ggml-ci

2 months agocuda : Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs (#14741)
Oliver Simons [Fri, 18 Jul 2025 11:35:32 +0000 (13:35 +0200)]
cuda : Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs (#14741)

* Fix Gemma3n not executed as CUDA_GRAPH on NVGPUs

Gemma3n uses Matrix-Matrix addition as part of their input processing,
wrongly triggering CUDA_GRAPH disablement on NVGPUs even when batch-size
of 1 is used.

* Exclude `project_per_layer_input` by matching node names

This ensures that all other graphs which don't exhibit this pattern do
not have their behavior changed.

* Revert unnecessary formatting changes

2 months agograph : avoid huge warm-up graphs for MoE models (#14753)
Georgi Gerganov [Fri, 18 Jul 2025 11:31:15 +0000 (14:31 +0300)]
graph : avoid huge warm-up graphs for MoE models (#14753)

* graph : avoid huge warm-up graphs for MoE models

ggml-ci

* cont : bump max nodes to 8x model tensors

2 months agomodel : fix build after merge conflict (#14754)
Georgi Gerganov [Fri, 18 Jul 2025 08:53:55 +0000 (11:53 +0300)]
model : fix build after merge conflict (#14754)

2 months agomodel : add EXAONE 4.0 support (#14630)
lgai-exaone [Fri, 18 Jul 2025 08:45:49 +0000 (17:45 +0900)]
model : add EXAONE 4.0 support (#14630)

2 months agoCUDA: set_rows + cpy.cu refactor (#14712)
Aman Gupta [Fri, 18 Jul 2025 06:54:18 +0000 (14:54 +0800)]
CUDA: set_rows + cpy.cu refactor (#14712)

2 months agograph : refactor context to not pass gf explicitly (#14629)
Georgi Gerganov [Fri, 18 Jul 2025 05:29:28 +0000 (08:29 +0300)]
graph : refactor context to not pass gf explicitly (#14629)

ggml-ci

2 months agograph : Pass the graph placeholder message in debug mode (#14748)
Nexes the Elder [Fri, 18 Jul 2025 04:25:54 +0000 (06:25 +0200)]
graph : Pass the graph placeholder message in debug mode (#14748)

Without that condition, this debug log clutters the screen every batch treated in the prompt processing, or every token generated in Kobold.cpp.