Code

sync : llama.cpp#1060

Code

Merged

ggerganov merged 22 commits intomasterggml-org/ggml:masterfrom sync-llama.cpp-25-01-03ggml-org/ggml:sync-llama.cpp-25-01-03Copy head branch name to clipboardJan 3, 2025

+1,555-1,313

Commits

Commits on Jan 3, 2025

• tests: add tests for GGUF (llama/10830)

  

  JohannesGaessler

  authored and

  ggerganov

  committed
  089373d

  Copy full SHA for 089373d

• tts : add OuteTTS support (llama/10784)

  Show description for 1c45f7d
  ggerganov

  committed
  1c45f7d

  Copy full SHA for 1c45f7d

• ggml : fix arm build (llama/10890)

  Show description for 9937f58
  

  2 peopleauthored and

  ggerganov

  committed
  9937f58

  Copy full SHA for 9937f58

• ggml: fix arm build with gcc (llama/10895)

  Show description for 66d5a45
  

  angt

  authored and

  ggerganov

  committed
  66d5a45

  Copy full SHA for 66d5a45

• ggml : add test for SVE and disable when it fails (llama/10906)

  

  slaren

  authored and

  ggerganov

  committed
  88ac055

  Copy full SHA for 88ac055

• SYCL: Migrate away from deprecated ggml_tensor->backend (llama/10840)

  Show description for 5a7354f
  

  qnixsynapse

  authored and

  ggerganov

  committed
  5a7354f

  Copy full SHA for 5a7354f

• ggml-cpu: replace NEON asm with intrinsics in ggml_gemv_q4_0_4x8_q8_0() (llama/10874)

  Show description for ee74a78
  

  angt

  authored and

  ggerganov

  committed
  ee74a78

  Copy full SHA for ee74a78

• vulkan: optimize coopmat2 dequant functions (llama/10855)

  Show description for 8b18e52
  

  jeffbolznv

  authored and

  ggerganov

  committed
  8b18e52

  Copy full SHA for 8b18e52

• vulkan: build fixes for 32b (llama/10927)

  Show description for 02db05d
  

  jeffbolznv

  authored and

  ggerganov

  committed
  02db05d

  Copy full SHA for 02db05d

• ggml : fix run-time on FreeBSD in get_executable_path() (llama/10948)

  

  yuri@FreeBSD

  authored and

  ggerganov

  committed
  f77c213

  Copy full SHA for f77c213

• ggml : fix const usage in SSE path (llama/10962)

  

  slaren

  authored and

  ggerganov

  committed
  2f8aea9

  Copy full SHA for 2f8aea9

• ggml : fix arm enabled features check (llama/10961)

  

  slaren

  authored and

  ggerganov

  committed
  250b245

  Copy full SHA for 250b245

• ggml : use wstring for backend search paths (llama/10960)

  Show description for 7f292ed
  

  slaren

  authored and

  ggerganov

  committed
  7f292ed

  Copy full SHA for 7f292ed

• ggml : more perfo with llamafile tinyblas on x86_64 (llama/10714)

  Show description for 636a15a
  

  Djip007

  authored and

  ggerganov

  committed
  636a15a

  Copy full SHA for 636a15a

• examples, ggml : fix GCC compiler warnings (llama/10983)

  Show description for d71439a
  

  peter277

  authored and

  ggerganov

  committed
  d71439a

  Copy full SHA for d71439a

• vulkan: multi-row k quants (llama/10846)

  Show description for e659119
  

  netrunnereve

  authored and

  ggerganov

  committed
  e659119

  Copy full SHA for e659119

• vulkan: Use push constant offset to handle misaligned descriptors (llama/10987)

  

  jeffbolznv

  authored and

  ggerganov

  committed
  92d38ff

  Copy full SHA for 92d38ff

• vulkan: im2col and matmul optimizations for stable diffusion (llama/10942)

  Show description for 3bcc231
  

  jeffbolznv

  authored and

  ggerganov

  committed
  3bcc231

  Copy full SHA for 3bcc231

• vulkan: optimize mul_mat for small values of N (llama/10991)

  Show description for 97d1ca9
  

  jeffbolznv

  authored and

  ggerganov

  committed
  97d1ca9

  Copy full SHA for 97d1ca9

• ggml : fixes for AVXVNNI instruction set with MSVC and Clang (llama/11027)

  Show description for 43c2a5c
  

  2 peopleauthored and

  ggerganov

  committed
  43c2a5c

  Copy full SHA for 43c2a5c

• metal : avoid uint (llama/11019)

  ggerganov

  committed
  0937be4

  Copy full SHA for 0937be4

• sync : llama.cpp

  Show description for add9f12
  ggerganov

  committed
  add9f12

  Copy full SHA for add9f12