whispercpp 1.3.4 → 1.3.5

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (630) hide show
  1. checksums.yaml +4 -4
  2. data/README.md +60 -43
  3. data/ext/extconf.rb +2 -2
  4. data/ext/ruby_whisper.c +14 -2
  5. data/ext/ruby_whisper.h +39 -0
  6. data/ext/ruby_whisper_context.c +22 -22
  7. data/ext/ruby_whisper_model.c +12 -12
  8. data/ext/ruby_whisper_params.c +47 -23
  9. data/ext/ruby_whisper_segment.c +84 -19
  10. data/ext/ruby_whisper_token.c +351 -0
  11. data/ext/ruby_whisper_transcribe.cpp +1 -1
  12. data/ext/ruby_whisper_vad_context.c +75 -0
  13. data/ext/ruby_whisper_vad_context_detect.cpp +50 -0
  14. data/ext/ruby_whisper_vad_segment.c +139 -0
  15. data/ext/ruby_whisper_vad_segments.c +106 -0
  16. data/ext/sources/CMakeLists.txt +4 -1
  17. data/ext/sources/bindings/javascript/package.json +1 -1
  18. data/ext/sources/cmake/arm64-apple-clang.cmake +16 -0
  19. data/ext/sources/cmake/arm64-windows-llvm.cmake +16 -0
  20. data/ext/sources/cmake/riscv64-spacemit-linux-gnu-gcc.cmake +29 -0
  21. data/ext/sources/cmake/x64-windows-llvm.cmake +5 -0
  22. data/ext/sources/examples/addon.node/vad-example.js +2 -2
  23. data/ext/sources/examples/cli/cli.cpp +121 -112
  24. data/ext/sources/examples/lsp/CMakeLists.txt +2 -1
  25. data/ext/sources/examples/quantize/CMakeLists.txt +2 -1
  26. data/ext/sources/examples/server/server.cpp +10 -11
  27. data/ext/sources/examples/talk-llama/CMakeLists.txt +5 -1
  28. data/ext/sources/examples/talk-llama/llama-adapter.cpp +12 -3
  29. data/ext/sources/examples/talk-llama/llama-adapter.h +7 -1
  30. data/ext/sources/examples/talk-llama/llama-arch.cpp +2046 -1974
  31. data/ext/sources/examples/talk-llama/llama-arch.h +67 -2
  32. data/ext/sources/examples/talk-llama/llama-batch.cpp +75 -33
  33. data/ext/sources/examples/talk-llama/llama-batch.h +17 -4
  34. data/ext/sources/examples/talk-llama/llama-chat.cpp +79 -3
  35. data/ext/sources/examples/talk-llama/llama-chat.h +4 -0
  36. data/ext/sources/examples/talk-llama/llama-context.cpp +775 -78
  37. data/ext/sources/examples/talk-llama/llama-context.h +57 -9
  38. data/ext/sources/examples/talk-llama/llama-cparams.h +1 -0
  39. data/ext/sources/examples/talk-llama/llama-grammar.cpp +288 -53
  40. data/ext/sources/examples/talk-llama/llama-grammar.h +22 -1
  41. data/ext/sources/examples/talk-llama/llama-graph.cpp +381 -64
  42. data/ext/sources/examples/talk-llama/llama-graph.h +103 -13
  43. data/ext/sources/examples/talk-llama/llama-hparams.cpp +26 -2
  44. data/ext/sources/examples/talk-llama/llama-hparams.h +41 -10
  45. data/ext/sources/examples/talk-llama/llama-impl.cpp +7 -3
  46. data/ext/sources/examples/talk-llama/llama-impl.h +1 -1
  47. data/ext/sources/examples/talk-llama/llama-kv-cache-iswa.cpp +5 -3
  48. data/ext/sources/examples/talk-llama/llama-kv-cache.cpp +145 -65
  49. data/ext/sources/examples/talk-llama/llama-kv-cache.h +22 -7
  50. data/ext/sources/examples/talk-llama/llama-kv-cells.h +44 -2
  51. data/ext/sources/examples/talk-llama/llama-memory-hybrid.cpp +12 -10
  52. data/ext/sources/examples/talk-llama/llama-memory-recurrent.cpp +32 -19
  53. data/ext/sources/examples/talk-llama/llama-memory-recurrent.h +2 -2
  54. data/ext/sources/examples/talk-llama/llama-mmap.cpp +172 -37
  55. data/ext/sources/examples/talk-llama/llama-mmap.h +8 -3
  56. data/ext/sources/examples/talk-llama/llama-model-loader.cpp +91 -9
  57. data/ext/sources/examples/talk-llama/llama-model-loader.h +6 -0
  58. data/ext/sources/examples/talk-llama/llama-model-saver.cpp +3 -0
  59. data/ext/sources/examples/talk-llama/llama-model.cpp +1529 -13134
  60. data/ext/sources/examples/talk-llama/llama-model.h +44 -3
  61. data/ext/sources/examples/talk-llama/llama-quant.cpp +8 -23
  62. data/ext/sources/examples/talk-llama/llama-sampling.cpp +1294 -198
  63. data/ext/sources/examples/talk-llama/llama-sampling.h +19 -7
  64. data/ext/sources/examples/talk-llama/llama-vocab.cpp +133 -37
  65. data/ext/sources/examples/talk-llama/llama-vocab.h +45 -40
  66. data/ext/sources/examples/talk-llama/llama.cpp +729 -2
  67. data/ext/sources/examples/talk-llama/llama.h +152 -14
  68. data/ext/sources/examples/talk-llama/models/afmoe.cpp +191 -0
  69. data/ext/sources/examples/talk-llama/models/apertus.cpp +125 -0
  70. data/ext/sources/examples/talk-llama/models/arcee.cpp +135 -0
  71. data/ext/sources/examples/talk-llama/models/arctic.cpp +138 -0
  72. data/ext/sources/examples/talk-llama/models/arwkv7.cpp +86 -0
  73. data/ext/sources/examples/talk-llama/models/baichuan.cpp +122 -0
  74. data/ext/sources/examples/talk-llama/models/bailingmoe.cpp +144 -0
  75. data/ext/sources/examples/talk-llama/models/bailingmoe2.cpp +135 -0
  76. data/ext/sources/examples/talk-llama/models/bert.cpp +178 -0
  77. data/ext/sources/examples/talk-llama/models/bitnet.cpp +160 -0
  78. data/ext/sources/examples/talk-llama/models/bloom.cpp +101 -0
  79. data/ext/sources/examples/talk-llama/models/chameleon.cpp +178 -0
  80. data/ext/sources/examples/talk-llama/models/chatglm.cpp +132 -0
  81. data/ext/sources/examples/talk-llama/models/codeshell.cpp +111 -0
  82. data/ext/sources/examples/talk-llama/models/cogvlm.cpp +102 -0
  83. data/ext/sources/examples/talk-llama/models/cohere2-iswa.cpp +134 -0
  84. data/ext/sources/examples/talk-llama/models/command-r.cpp +122 -0
  85. data/ext/sources/examples/talk-llama/models/dbrx.cpp +123 -0
  86. data/ext/sources/examples/talk-llama/models/deci.cpp +135 -0
  87. data/ext/sources/examples/talk-llama/models/deepseek.cpp +144 -0
  88. data/ext/sources/examples/talk-llama/models/deepseek2.cpp +259 -0
  89. data/ext/sources/examples/talk-llama/models/dots1.cpp +134 -0
  90. data/ext/sources/examples/talk-llama/models/dream.cpp +105 -0
  91. data/ext/sources/examples/talk-llama/models/ernie4-5-moe.cpp +150 -0
  92. data/ext/sources/examples/talk-llama/models/ernie4-5.cpp +110 -0
  93. data/ext/sources/examples/talk-llama/models/exaone.cpp +114 -0
  94. data/ext/sources/examples/talk-llama/models/exaone4.cpp +123 -0
  95. data/ext/sources/examples/talk-llama/models/falcon-h1.cpp +113 -0
  96. data/ext/sources/examples/talk-llama/models/falcon.cpp +120 -0
  97. data/ext/sources/examples/talk-llama/models/gemma-embedding.cpp +116 -0
  98. data/ext/sources/examples/talk-llama/models/gemma.cpp +112 -0
  99. data/ext/sources/examples/talk-llama/models/gemma2-iswa.cpp +128 -0
  100. data/ext/sources/examples/talk-llama/models/gemma3.cpp +155 -0
  101. data/ext/sources/examples/talk-llama/models/gemma3n-iswa.cpp +384 -0
  102. data/ext/sources/examples/talk-llama/models/glm4-moe.cpp +170 -0
  103. data/ext/sources/examples/talk-llama/models/glm4.cpp +150 -0
  104. data/ext/sources/examples/talk-llama/models/gpt2.cpp +105 -0
  105. data/ext/sources/examples/talk-llama/models/gptneox.cpp +144 -0
  106. data/ext/sources/examples/talk-llama/models/granite-hybrid.cpp +196 -0
  107. data/ext/sources/examples/talk-llama/models/granite.cpp +211 -0
  108. data/ext/sources/examples/talk-llama/models/graph-context-mamba.cpp +283 -0
  109. data/ext/sources/examples/talk-llama/models/grok.cpp +159 -0
  110. data/ext/sources/examples/talk-llama/models/grovemoe.cpp +141 -0
  111. data/ext/sources/examples/talk-llama/models/hunyuan-dense.cpp +132 -0
  112. data/ext/sources/examples/talk-llama/models/hunyuan-moe.cpp +154 -0
  113. data/ext/sources/examples/talk-llama/models/internlm2.cpp +120 -0
  114. data/ext/sources/examples/talk-llama/models/jais.cpp +86 -0
  115. data/ext/sources/examples/talk-llama/models/jamba.cpp +106 -0
  116. data/ext/sources/examples/talk-llama/models/lfm2.cpp +175 -0
  117. data/ext/sources/examples/talk-llama/models/llada-moe.cpp +122 -0
  118. data/ext/sources/examples/talk-llama/models/llada.cpp +99 -0
  119. data/ext/sources/examples/talk-llama/models/llama-iswa.cpp +178 -0
  120. data/ext/sources/examples/talk-llama/models/llama.cpp +168 -0
  121. data/ext/sources/examples/talk-llama/models/maincoder.cpp +117 -0
  122. data/ext/sources/examples/talk-llama/models/mamba.cpp +55 -0
  123. data/ext/sources/examples/talk-llama/models/mimo2-iswa.cpp +123 -0
  124. data/ext/sources/examples/talk-llama/models/minicpm3.cpp +199 -0
  125. data/ext/sources/examples/talk-llama/models/minimax-m2.cpp +124 -0
  126. data/ext/sources/examples/talk-llama/models/mistral3.cpp +160 -0
  127. data/ext/sources/examples/talk-llama/models/models.h +569 -0
  128. data/ext/sources/examples/talk-llama/models/modern-bert.cpp +116 -0
  129. data/ext/sources/examples/talk-llama/models/mpt.cpp +126 -0
  130. data/ext/sources/examples/talk-llama/models/nemotron-h.cpp +150 -0
  131. data/ext/sources/examples/talk-llama/models/nemotron.cpp +122 -0
  132. data/ext/sources/examples/talk-llama/models/neo-bert.cpp +104 -0
  133. data/ext/sources/examples/talk-llama/models/olmo.cpp +121 -0
  134. data/ext/sources/examples/talk-llama/models/olmo2.cpp +150 -0
  135. data/ext/sources/examples/talk-llama/models/olmoe.cpp +124 -0
  136. data/ext/sources/examples/talk-llama/models/openai-moe-iswa.cpp +127 -0
  137. data/ext/sources/examples/talk-llama/models/openelm.cpp +124 -0
  138. data/ext/sources/examples/talk-llama/models/orion.cpp +123 -0
  139. data/ext/sources/examples/talk-llama/models/pangu-embedded.cpp +121 -0
  140. data/ext/sources/examples/talk-llama/models/phi2.cpp +121 -0
  141. data/ext/sources/examples/talk-llama/models/phi3.cpp +152 -0
  142. data/ext/sources/examples/talk-llama/models/plamo.cpp +110 -0
  143. data/ext/sources/examples/talk-llama/models/plamo2.cpp +316 -0
  144. data/ext/sources/examples/talk-llama/models/plamo3.cpp +128 -0
  145. data/ext/sources/examples/talk-llama/models/plm.cpp +168 -0
  146. data/ext/sources/examples/talk-llama/models/qwen.cpp +108 -0
  147. data/ext/sources/examples/talk-llama/models/qwen2.cpp +126 -0
  148. data/ext/sources/examples/talk-llama/models/qwen2moe.cpp +151 -0
  149. data/ext/sources/examples/talk-llama/models/qwen2vl.cpp +117 -0
  150. data/ext/sources/examples/talk-llama/models/qwen3.cpp +117 -0
  151. data/ext/sources/examples/talk-llama/models/qwen3moe.cpp +124 -0
  152. data/ext/sources/examples/talk-llama/models/qwen3next.cpp +873 -0
  153. data/ext/sources/examples/talk-llama/models/qwen3vl-moe.cpp +149 -0
  154. data/ext/sources/examples/talk-llama/models/qwen3vl.cpp +141 -0
  155. data/ext/sources/examples/talk-llama/models/refact.cpp +94 -0
  156. data/ext/sources/examples/talk-llama/models/rnd1.cpp +126 -0
  157. data/ext/sources/examples/talk-llama/models/rwkv6-base.cpp +162 -0
  158. data/ext/sources/examples/talk-llama/models/rwkv6.cpp +94 -0
  159. data/ext/sources/examples/talk-llama/models/rwkv6qwen2.cpp +86 -0
  160. data/ext/sources/examples/talk-llama/models/rwkv7-base.cpp +135 -0
  161. data/ext/sources/examples/talk-llama/models/rwkv7.cpp +90 -0
  162. data/ext/sources/examples/talk-llama/models/seed-oss.cpp +124 -0
  163. data/ext/sources/examples/talk-llama/models/smallthinker.cpp +126 -0
  164. data/ext/sources/examples/talk-llama/models/smollm3.cpp +128 -0
  165. data/ext/sources/examples/talk-llama/models/stablelm.cpp +146 -0
  166. data/ext/sources/examples/talk-llama/models/starcoder.cpp +100 -0
  167. data/ext/sources/examples/talk-llama/models/starcoder2.cpp +121 -0
  168. data/ext/sources/examples/talk-llama/models/t5-dec.cpp +166 -0
  169. data/ext/sources/examples/talk-llama/models/t5-enc.cpp +96 -0
  170. data/ext/sources/examples/talk-llama/models/wavtokenizer-dec.cpp +149 -0
  171. data/ext/sources/examples/talk-llama/models/xverse.cpp +108 -0
  172. data/ext/sources/examples/talk-llama/unicode.cpp +102 -16
  173. data/ext/sources/examples/vad-speech-segments/CMakeLists.txt +1 -1
  174. data/ext/sources/examples/whisper.wasm/index-tmpl.html +1 -1
  175. data/ext/sources/ggml/CMakeLists.txt +82 -54
  176. data/ext/sources/ggml/include/ggml-alloc.h +9 -0
  177. data/ext/sources/ggml/include/ggml-backend.h +4 -1
  178. data/ext/sources/ggml/include/ggml-cpu.h +1 -0
  179. data/ext/sources/ggml/include/ggml-hexagon.h +19 -0
  180. data/ext/sources/ggml/include/ggml-rpc.h +8 -11
  181. data/ext/sources/ggml/include/ggml-zendnn.h +22 -0
  182. data/ext/sources/ggml/include/ggml.h +190 -12
  183. data/ext/sources/ggml/src/CMakeLists.txt +82 -11
  184. data/ext/sources/ggml/src/ggml-alloc.c +124 -41
  185. data/ext/sources/ggml/src/ggml-backend-impl.h +1 -4
  186. data/ext/sources/ggml/src/ggml-backend-reg.cpp +27 -3
  187. data/ext/sources/ggml/src/ggml-backend.cpp +71 -21
  188. data/ext/sources/ggml/src/ggml-blas/CMakeLists.txt +17 -3
  189. data/ext/sources/ggml/src/ggml-blas/ggml-blas.cpp +5 -9
  190. data/ext/sources/ggml/src/ggml-cann/acl_tensor.cpp +57 -45
  191. data/ext/sources/ggml/src/ggml-cann/acl_tensor.h +138 -47
  192. data/ext/sources/ggml/src/ggml-cann/aclnn_ops.cpp +2179 -1696
  193. data/ext/sources/ggml/src/ggml-cann/aclnn_ops.h +238 -317
  194. data/ext/sources/ggml/src/ggml-cann/common.h +283 -208
  195. data/ext/sources/ggml/src/ggml-cann/ggml-cann.cpp +626 -776
  196. data/ext/sources/ggml/src/ggml-cpu/CMakeLists.txt +156 -86
  197. data/ext/sources/ggml/src/ggml-cpu/amx/amx.cpp +1 -0
  198. data/ext/sources/ggml/src/ggml-cpu/arch/arm/cpu-feats.cpp +4 -0
  199. data/ext/sources/ggml/src/ggml-cpu/arch/arm/quants.c +428 -26
  200. data/ext/sources/ggml/src/ggml-cpu/arch/arm/repack.cpp +1004 -0
  201. data/ext/sources/ggml/src/ggml-cpu/arch/loongarch/quants.c +4 -5
  202. data/ext/sources/ggml/src/ggml-cpu/arch/riscv/cpu-feats.cpp +38 -0
  203. data/ext/sources/ggml/src/ggml-cpu/arch/riscv/quants.c +108 -49
  204. data/ext/sources/ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp +50 -0
  205. data/ext/sources/ggml/src/ggml-cpu/arch/x86/repack.cpp +6 -6
  206. data/ext/sources/ggml/src/ggml-cpu/arch-fallback.h +50 -2
  207. data/ext/sources/ggml/src/ggml-cpu/ggml-cpu-impl.h +5 -3
  208. data/ext/sources/ggml/src/ggml-cpu/ggml-cpu.c +195 -71
  209. data/ext/sources/ggml/src/ggml-cpu/ggml-cpu.cpp +4 -0
  210. data/ext/sources/ggml/src/ggml-cpu/kleidiai/kernels.cpp +573 -106
  211. data/ext/sources/ggml/src/ggml-cpu/kleidiai/kernels.h +33 -44
  212. data/ext/sources/ggml/src/ggml-cpu/kleidiai/kleidiai.cpp +298 -112
  213. data/ext/sources/ggml/src/ggml-cpu/llamafile/sgemm-ppc.h +333 -0
  214. data/ext/sources/ggml/src/ggml-cpu/llamafile/sgemm.cpp +819 -125
  215. data/ext/sources/ggml/src/ggml-cpu/llamafile/sgemm.h +6 -0
  216. data/ext/sources/ggml/src/ggml-cpu/ops.cpp +708 -431
  217. data/ext/sources/ggml/src/ggml-cpu/ops.h +5 -4
  218. data/ext/sources/ggml/src/ggml-cpu/repack.cpp +671 -31
  219. data/ext/sources/ggml/src/ggml-cpu/repack.h +14 -0
  220. data/ext/sources/ggml/src/ggml-cpu/simd-mappings.h +41 -43
  221. data/ext/sources/ggml/src/ggml-cpu/spacemit/ime.cpp +3 -2
  222. data/ext/sources/ggml/src/ggml-cpu/unary-ops.cpp +151 -0
  223. data/ext/sources/ggml/src/ggml-cpu/unary-ops.h +7 -0
  224. data/ext/sources/ggml/src/ggml-cpu/vec.cpp +124 -1
  225. data/ext/sources/ggml/src/ggml-cpu/vec.h +261 -146
  226. data/ext/sources/ggml/src/ggml-cuda/CMakeLists.txt +72 -1
  227. data/ext/sources/ggml/src/ggml-cuda/argmax.cu +2 -2
  228. data/ext/sources/ggml/src/ggml-cuda/argsort.cu +123 -6
  229. data/ext/sources/ggml/src/ggml-cuda/argsort.cuh +16 -0
  230. data/ext/sources/ggml/src/ggml-cuda/binbcast.cu +1 -1
  231. data/ext/sources/ggml/src/ggml-cuda/common.cuh +353 -80
  232. data/ext/sources/ggml/src/ggml-cuda/convert.cuh +10 -0
  233. data/ext/sources/ggml/src/ggml-cuda/cpy-utils.cuh +1 -1
  234. data/ext/sources/ggml/src/ggml-cuda/cpy.cu +339 -246
  235. data/ext/sources/ggml/src/ggml-cuda/cpy.cuh +1 -5
  236. data/ext/sources/ggml/src/ggml-cuda/cumsum.cu +307 -0
  237. data/ext/sources/ggml/src/ggml-cuda/cumsum.cuh +5 -0
  238. data/ext/sources/ggml/src/ggml-cuda/diag.cu +77 -0
  239. data/ext/sources/ggml/src/ggml-cuda/diag.cuh +5 -0
  240. data/ext/sources/ggml/src/ggml-cuda/fattn-common.cuh +31 -21
  241. data/ext/sources/ggml/src/ggml-cuda/fattn-mma-f16.cuh +663 -596
  242. data/ext/sources/ggml/src/ggml-cuda/fattn-tile.cu +35 -741
  243. data/ext/sources/ggml/src/ggml-cuda/fattn-tile.cuh +1241 -0
  244. data/ext/sources/ggml/src/ggml-cuda/fattn-vec.cuh +30 -37
  245. data/ext/sources/ggml/src/ggml-cuda/fattn-wmma-f16.cu +14 -13
  246. data/ext/sources/ggml/src/ggml-cuda/fattn-wmma-f16.cuh +48 -0
  247. data/ext/sources/ggml/src/ggml-cuda/fattn.cu +83 -37
  248. data/ext/sources/ggml/src/ggml-cuda/fill.cu +37 -0
  249. data/ext/sources/ggml/src/ggml-cuda/fill.cuh +3 -0
  250. data/ext/sources/ggml/src/ggml-cuda/ggml-cuda.cu +1155 -164
  251. data/ext/sources/ggml/src/ggml-cuda/mean.cu +5 -4
  252. data/ext/sources/ggml/src/ggml-cuda/mma.cuh +741 -48
  253. data/ext/sources/ggml/src/ggml-cuda/mmf.cu +60 -12
  254. data/ext/sources/ggml/src/ggml-cuda/mmf.cuh +381 -42
  255. data/ext/sources/ggml/src/ggml-cuda/mmid.cu +164 -0
  256. data/ext/sources/ggml/src/ggml-cuda/mmid.cuh +5 -0
  257. data/ext/sources/ggml/src/ggml-cuda/mmq.cu +69 -176
  258. data/ext/sources/ggml/src/ggml-cuda/mmq.cuh +498 -171
  259. data/ext/sources/ggml/src/ggml-cuda/mmvf.cu +375 -79
  260. data/ext/sources/ggml/src/ggml-cuda/mmvf.cuh +3 -2
  261. data/ext/sources/ggml/src/ggml-cuda/mmvq.cu +241 -95
  262. data/ext/sources/ggml/src/ggml-cuda/mmvq.cuh +1 -1
  263. data/ext/sources/ggml/src/ggml-cuda/pad.cu +64 -33
  264. data/ext/sources/ggml/src/ggml-cuda/quantize.cu +151 -0
  265. data/ext/sources/ggml/src/ggml-cuda/quantize.cuh +14 -0
  266. data/ext/sources/ggml/src/ggml-cuda/rope.cu +192 -77
  267. data/ext/sources/ggml/src/ggml-cuda/rope.cuh +2 -0
  268. data/ext/sources/ggml/src/ggml-cuda/set-rows.cu +101 -47
  269. data/ext/sources/ggml/src/ggml-cuda/set.cu +39 -0
  270. data/ext/sources/ggml/src/ggml-cuda/set.cuh +7 -0
  271. data/ext/sources/ggml/src/ggml-cuda/softmax.cu +203 -6
  272. data/ext/sources/ggml/src/ggml-cuda/solve_tri.cu +275 -0
  273. data/ext/sources/ggml/src/ggml-cuda/solve_tri.cuh +3 -0
  274. data/ext/sources/ggml/src/ggml-cuda/ssm-conv.cu +14 -20
  275. data/ext/sources/ggml/src/ggml-cuda/ssm-scan.cu +49 -84
  276. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu +5 -0
  277. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu +5 -0
  278. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu +5 -0
  279. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu +5 -0
  280. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu +5 -0
  281. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu +5 -0
  282. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu +5 -0
  283. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu +5 -0
  284. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu +5 -0
  285. data/ext/sources/ggml/src/ggml-cuda/template-instances/generate_cu_files.py +19 -1
  286. data/ext/sources/ggml/src/ggml-cuda/top-k.cu +96 -0
  287. data/ext/sources/ggml/src/ggml-cuda/top-k.cuh +3 -0
  288. data/ext/sources/ggml/src/ggml-cuda/topk-moe.cu +168 -76
  289. data/ext/sources/ggml/src/ggml-cuda/topk-moe.cuh +11 -4
  290. data/ext/sources/ggml/src/ggml-cuda/tri.cu +136 -0
  291. data/ext/sources/ggml/src/ggml-cuda/tri.cuh +5 -0
  292. data/ext/sources/ggml/src/ggml-cuda/unary.cu +105 -11
  293. data/ext/sources/ggml/src/ggml-cuda/unary.cuh +36 -0
  294. data/ext/sources/ggml/src/ggml-cuda/upscale.cu +163 -7
  295. data/ext/sources/ggml/src/ggml-cuda/vendors/cuda.h +4 -0
  296. data/ext/sources/ggml/src/ggml-cuda/vendors/hip.h +12 -1
  297. data/ext/sources/ggml/src/ggml-cuda/vendors/musa.h +6 -0
  298. data/ext/sources/ggml/src/ggml-hexagon/CMakeLists.txt +80 -0
  299. data/ext/sources/ggml/src/ggml-hexagon/ggml-hexagon.cpp +3151 -0
  300. data/ext/sources/ggml/src/ggml-hexagon/htp/CMakeLists.txt +44 -0
  301. data/ext/sources/ggml/src/ggml-hexagon/htp/act-ops.c +682 -0
  302. data/ext/sources/ggml/src/ggml-hexagon/htp/binary-ops.c +360 -0
  303. data/ext/sources/ggml/src/ggml-hexagon/htp/cmake-toolchain.cmake +157 -0
  304. data/ext/sources/ggml/src/ggml-hexagon/htp/flash-attn-ops.c +566 -0
  305. data/ext/sources/ggml/src/ggml-hexagon/htp/get-rows-ops.c +112 -0
  306. data/ext/sources/ggml/src/ggml-hexagon/htp/htp-ctx.h +35 -0
  307. data/ext/sources/ggml/src/ggml-hexagon/htp/htp-dma.c +63 -0
  308. data/ext/sources/ggml/src/ggml-hexagon/htp/htp-dma.h +157 -0
  309. data/ext/sources/ggml/src/ggml-hexagon/htp/htp-msg.h +165 -0
  310. data/ext/sources/ggml/src/ggml-hexagon/htp/htp-ops.h +92 -0
  311. data/ext/sources/ggml/src/ggml-hexagon/htp/htp_iface.idl +16 -0
  312. data/ext/sources/ggml/src/ggml-hexagon/htp/hvx-exp.c +94 -0
  313. data/ext/sources/ggml/src/ggml-hexagon/htp/hvx-inverse.c +72 -0
  314. data/ext/sources/ggml/src/ggml-hexagon/htp/hvx-sigmoid.c +49 -0
  315. data/ext/sources/ggml/src/ggml-hexagon/htp/hvx-utils.c +1020 -0
  316. data/ext/sources/ggml/src/ggml-hexagon/htp/hvx-utils.h +1353 -0
  317. data/ext/sources/ggml/src/ggml-hexagon/htp/main.c +1001 -0
  318. data/ext/sources/ggml/src/ggml-hexagon/htp/matmul-ops.c +2503 -0
  319. data/ext/sources/ggml/src/ggml-hexagon/htp/ops-utils.h +149 -0
  320. data/ext/sources/ggml/src/ggml-hexagon/htp/rope-ops.c +487 -0
  321. data/ext/sources/ggml/src/ggml-hexagon/htp/set-rows-ops.c +168 -0
  322. data/ext/sources/ggml/src/ggml-hexagon/htp/softmax-ops.c +402 -0
  323. data/ext/sources/ggml/src/ggml-hexagon/htp/unary-ops.c +287 -0
  324. data/ext/sources/ggml/src/ggml-hexagon/htp/worker-pool.c +297 -0
  325. data/ext/sources/ggml/src/ggml-hexagon/htp/worker-pool.h +57 -0
  326. data/ext/sources/ggml/src/ggml-hexagon/htp-utils.c +454 -0
  327. data/ext/sources/ggml/src/ggml-hexagon/htp-utils.h +221 -0
  328. data/ext/sources/ggml/src/ggml-hexagon/op-desc.h +153 -0
  329. data/ext/sources/ggml/src/ggml-hip/CMakeLists.txt +8 -13
  330. data/ext/sources/ggml/src/ggml-impl.h +67 -6
  331. data/ext/sources/ggml/src/ggml-metal/ggml-metal-common.cpp +2 -2
  332. data/ext/sources/ggml/src/ggml-metal/ggml-metal-context.m +29 -20
  333. data/ext/sources/ggml/src/ggml-metal/ggml-metal-device.cpp +652 -285
  334. data/ext/sources/ggml/src/ggml-metal/ggml-metal-device.h +103 -56
  335. data/ext/sources/ggml/src/ggml-metal/ggml-metal-device.m +496 -118
  336. data/ext/sources/ggml/src/ggml-metal/ggml-metal-impl.h +231 -9
  337. data/ext/sources/ggml/src/ggml-metal/ggml-metal-ops.cpp +1227 -224
  338. data/ext/sources/ggml/src/ggml-metal/ggml-metal-ops.h +12 -0
  339. data/ext/sources/ggml/src/ggml-metal/ggml-metal.cpp +14 -8
  340. data/ext/sources/ggml/src/ggml-metal/ggml-metal.metal +1972 -704
  341. data/ext/sources/ggml/src/ggml-musa/CMakeLists.txt +3 -1
  342. data/ext/sources/ggml/src/ggml-opencl/CMakeLists.txt +11 -0
  343. data/ext/sources/ggml/src/ggml-opencl/ggml-opencl.cpp +1430 -120
  344. data/ext/sources/ggml/src/ggml-opencl/kernels/cvt.cl +63 -0
  345. data/ext/sources/ggml/src/ggml-opencl/kernels/expm1.cl +82 -0
  346. data/ext/sources/ggml/src/ggml-opencl/kernels/fill.cl +17 -0
  347. data/ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32.cl +4 -3
  348. data/ext/sources/ggml/src/ggml-opencl/kernels/gemm_moe_mxfp4_f32.cl +162 -0
  349. data/ext/sources/ggml/src/ggml-opencl/kernels/gemv_moe_mxfp4_f32.cl +156 -0
  350. data/ext/sources/ggml/src/ggml-opencl/kernels/get_rows.cl +36 -12
  351. data/ext/sources/ggml/src/ggml-opencl/kernels/mean.cl +39 -0
  352. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f16_f32_kq_kqv.cl +273 -0
  353. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f16_f32_l4_lm.cl +24 -10
  354. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f32_f32_l4_lm.cl +24 -10
  355. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_q8_0_f32_l4_lm.cl +154 -0
  356. data/ext/sources/ggml/src/ggml-opencl/kernels/pad.cl +29 -20
  357. data/ext/sources/ggml/src/ggml-opencl/kernels/rms_norm.cl +25 -10
  358. data/ext/sources/ggml/src/ggml-opencl/kernels/rope.cl +50 -24
  359. data/ext/sources/ggml/src/ggml-opencl/kernels/set_rows.cl +35 -16
  360. data/ext/sources/ggml/src/ggml-opencl/kernels/softplus.cl +88 -0
  361. data/ext/sources/ggml/src/ggml-opencl/kernels/sqr.cl +53 -0
  362. data/ext/sources/ggml/src/ggml-opencl/kernels/sqrt.cl +53 -0
  363. data/ext/sources/ggml/src/ggml-opencl/kernels/ssm_conv.cl +77 -0
  364. data/ext/sources/ggml/src/ggml-opencl/kernels/transpose.cl +13 -0
  365. data/ext/sources/ggml/src/ggml-rpc/ggml-rpc.cpp +438 -156
  366. data/ext/sources/ggml/src/ggml-sycl/CMakeLists.txt +48 -3
  367. data/ext/sources/ggml/src/ggml-sycl/add-id.cpp +77 -0
  368. data/ext/sources/ggml/src/ggml-sycl/add-id.hpp +8 -0
  369. data/ext/sources/ggml/src/ggml-sycl/backend.hpp +6 -0
  370. data/ext/sources/ggml/src/ggml-sycl/binbcast.cpp +0 -9
  371. data/ext/sources/ggml/src/ggml-sycl/binbcast.hpp +0 -6
  372. data/ext/sources/ggml/src/ggml-sycl/common.hpp +117 -15
  373. data/ext/sources/ggml/src/ggml-sycl/concat.cpp +55 -44
  374. data/ext/sources/ggml/src/ggml-sycl/convert.cpp +34 -0
  375. data/ext/sources/ggml/src/ggml-sycl/count-equal.cpp +79 -0
  376. data/ext/sources/ggml/src/ggml-sycl/count-equal.hpp +9 -0
  377. data/ext/sources/ggml/src/ggml-sycl/cpy.cpp +0 -3
  378. data/ext/sources/ggml/src/ggml-sycl/dequantize.hpp +18 -0
  379. data/ext/sources/ggml/src/ggml-sycl/dpct/helper.hpp +76 -3
  380. data/ext/sources/ggml/src/ggml-sycl/element_wise.cpp +333 -300
  381. data/ext/sources/ggml/src/ggml-sycl/element_wise.hpp +10 -2
  382. data/ext/sources/ggml/src/ggml-sycl/ggml-sycl.cpp +335 -110
  383. data/ext/sources/ggml/src/ggml-sycl/mmvq.cpp +22 -0
  384. data/ext/sources/ggml/src/ggml-sycl/norm.cpp +156 -0
  385. data/ext/sources/ggml/src/ggml-sycl/norm.hpp +2 -0
  386. data/ext/sources/ggml/src/ggml-sycl/pad.cpp +97 -0
  387. data/ext/sources/ggml/src/ggml-sycl/pad.hpp +24 -0
  388. data/ext/sources/ggml/src/ggml-sycl/pad_reflect_1d.cpp +100 -0
  389. data/ext/sources/ggml/src/ggml-sycl/pad_reflect_1d.hpp +10 -0
  390. data/ext/sources/ggml/src/ggml-sycl/presets.hpp +2 -0
  391. data/ext/sources/ggml/src/ggml-sycl/repeat_back.cpp +76 -0
  392. data/ext/sources/ggml/src/ggml-sycl/repeat_back.hpp +8 -0
  393. data/ext/sources/ggml/src/ggml-sycl/roll.cpp +122 -0
  394. data/ext/sources/ggml/src/ggml-sycl/roll.hpp +20 -0
  395. data/ext/sources/ggml/src/ggml-sycl/rope.cpp +30 -17
  396. data/ext/sources/ggml/src/ggml-sycl/set.cpp +73 -0
  397. data/ext/sources/ggml/src/ggml-sycl/set.hpp +5 -0
  398. data/ext/sources/ggml/src/ggml-sycl/softmax.cpp +327 -162
  399. data/ext/sources/ggml/src/ggml-sycl/softmax.hpp +4 -0
  400. data/ext/sources/ggml/src/ggml-sycl/ssm_conv.cpp +127 -0
  401. data/ext/sources/ggml/src/ggml-sycl/ssm_conv.hpp +5 -0
  402. data/ext/sources/ggml/src/ggml-sycl/vecdotq.hpp +58 -0
  403. data/ext/sources/ggml/src/ggml-vulkan/CMakeLists.txt +38 -18
  404. data/ext/sources/ggml/src/ggml-vulkan/ggml-vulkan.cpp +5013 -2859
  405. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/abs.comp +21 -0
  406. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/acc.comp +2 -2
  407. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add.comp +2 -2
  408. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add1.comp +28 -0
  409. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add_id.comp +1 -1
  410. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/arange.comp +20 -0
  411. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argmax.comp +2 -2
  412. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argsort.comp +33 -26
  413. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argsort_large.comp +114 -0
  414. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/ceil.comp +22 -0
  415. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/clamp.comp +2 -2
  416. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/concat.comp +2 -2
  417. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/contig_copy.comp +2 -2
  418. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv2d_dw.comp +1 -1
  419. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv2d_mm.comp +47 -49
  420. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv_transpose_1d.comp +1 -1
  421. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy.comp +2 -2
  422. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_from_quant.comp +3 -3
  423. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_to_quant.comp +4 -4
  424. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_transpose.comp +67 -0
  425. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cos.comp +2 -2
  426. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/count_equal.comp +2 -2
  427. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/count_experts.comp +51 -0
  428. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cumsum.comp +83 -0
  429. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cumsum_multipass1.comp +60 -0
  430. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cumsum_multipass2.comp +66 -0
  431. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_f32.comp +1 -1
  432. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{dequant_funcs.comp → dequant_funcs.glsl} +9 -21
  433. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{dequant_funcs_cm2.comp → dequant_funcs_cm2.glsl} +18 -4
  434. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{dequant_head.comp → dequant_head.glsl} +1 -1
  435. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq1_m.comp +1 -1
  436. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq1_s.comp +1 -1
  437. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq2_s.comp +1 -1
  438. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq2_xs.comp +1 -1
  439. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq2_xxs.comp +1 -1
  440. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq3_s.comp +1 -1
  441. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq3_xxs.comp +1 -1
  442. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq4_nl.comp +1 -1
  443. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq4_xs.comp +1 -1
  444. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_mxfp4.comp +3 -3
  445. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q2_k.comp +3 -3
  446. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q3_k.comp +1 -1
  447. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q4_0.comp +1 -1
  448. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q4_1.comp +1 -1
  449. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q4_k.comp +3 -3
  450. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q5_0.comp +1 -1
  451. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q5_1.comp +1 -1
  452. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q5_k.comp +3 -3
  453. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q6_k.comp +1 -1
  454. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q8_0.comp +1 -1
  455. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/diag.comp +29 -0
  456. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/diag_mask_inf.comp +1 -1
  457. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/div.comp +2 -2
  458. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/exp.comp +3 -3
  459. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/fill.comp +19 -0
  460. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp +39 -17
  461. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{flash_attn_base.comp → flash_attn_base.glsl} +19 -1
  462. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp +45 -7
  463. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm2.comp +50 -12
  464. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_split_k_reduce.comp +1 -1
  465. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/floor.comp +22 -0
  466. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu.comp +2 -2
  467. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_erf.comp +2 -2
  468. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_quick.comp +2 -2
  469. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu.comp +2 -2
  470. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_erf.comp +2 -2
  471. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_quick.comp +2 -2
  472. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{generic_binary_head.comp → generic_binary_head.glsl} +17 -2
  473. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{generic_head.comp → generic_head.glsl} +2 -0
  474. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{generic_unary_head.comp → generic_unary_head.glsl} +7 -0
  475. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/get_rows.comp +4 -4
  476. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/get_rows_quant.comp +3 -3
  477. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{glu_head.comp → glu_head.glsl} +1 -1
  478. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/group_norm.comp +2 -2
  479. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardsigmoid.comp +2 -2
  480. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardswish.comp +2 -2
  481. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col.comp +19 -7
  482. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col_3d.comp +2 -3
  483. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/l2_norm.comp +2 -2
  484. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/leaky_relu.comp +2 -2
  485. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/log.comp +18 -0
  486. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul.comp +2 -2
  487. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec.comp +2 -2
  488. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{mul_mat_vec_base.comp → mul_mat_vec_base.glsl} +70 -25
  489. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iface.glsl +35 -0
  490. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq1_m.comp +71 -21
  491. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq1_s.comp +41 -25
  492. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq2_s.comp +2 -2
  493. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq2_xs.comp +44 -26
  494. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq2_xxs.comp +2 -2
  495. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq3_s.comp +2 -2
  496. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq3_xxs.comp +2 -2
  497. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_nc.comp +9 -7
  498. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_p021.comp +9 -7
  499. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q2_k.comp +4 -6
  500. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q3_k.comp +2 -2
  501. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q4_k.comp +4 -6
  502. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q5_k.comp +4 -6
  503. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q6_k.comp +2 -2
  504. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq.comp +39 -36
  505. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq_funcs.glsl +494 -0
  506. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp +78 -103
  507. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_cm2.comp +34 -23
  508. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{mul_mm_funcs.comp → mul_mm_funcs.glsl} +69 -59
  509. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_id_funcs.glsl +72 -0
  510. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq.comp +88 -228
  511. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_funcs.glsl +454 -0
  512. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_shmem_types.glsl +78 -0
  513. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/multi_add.comp +97 -13
  514. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/neg.comp +20 -0
  515. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/norm.comp +2 -2
  516. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_adamw.comp +2 -2
  517. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_sgd.comp +1 -1
  518. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/pad.comp +21 -6
  519. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/pool2d.comp +1 -1
  520. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/quantize_q8_1.comp +10 -10
  521. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/reglu.comp +2 -2
  522. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/relu.comp +2 -2
  523. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/repeat.comp +2 -2
  524. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/repeat_back.comp +2 -2
  525. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm.comp +50 -4
  526. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_back.comp +2 -2
  527. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_partials.comp +2 -2
  528. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/roll.comp +2 -2
  529. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_funcs.glsl +234 -0
  530. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_head.glsl +20 -0
  531. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_multi.comp +6 -50
  532. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_neox.comp +6 -33
  533. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_norm.comp +6 -33
  534. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_params.glsl +28 -0
  535. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_vision.comp +6 -39
  536. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/round.comp +29 -0
  537. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/scale.comp +2 -2
  538. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sigmoid.comp +2 -2
  539. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/silu.comp +2 -2
  540. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/silu_back.comp +2 -2
  541. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sin.comp +2 -2
  542. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max.comp +1 -1
  543. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_back.comp +2 -2
  544. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large1.comp +62 -0
  545. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large2.comp +79 -0
  546. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large3.comp +65 -0
  547. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large_common.glsl +53 -0
  548. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/softplus.comp +23 -0
  549. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/solve_tri.comp +81 -0
  550. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sqrt.comp +2 -2
  551. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/square.comp +2 -2
  552. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/ssm_conv.comp +44 -0
  553. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/ssm_scan.comp +124 -0
  554. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/step.comp +22 -0
  555. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sub.comp +2 -2
  556. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sum_rows.comp +2 -25
  557. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sum_rows.glsl +25 -0
  558. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu.comp +2 -2
  559. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu_oai.comp +2 -2
  560. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/tanh.comp +2 -2
  561. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/timestep_embedding.comp +1 -1
  562. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/topk_argsort.comp +118 -0
  563. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/topk_moe.comp +213 -0
  564. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/topk_nary_search.comp +246 -0
  565. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/tri.comp +43 -0
  566. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/trunc.comp +22 -0
  567. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{types.comp → types.glsl} +345 -26
  568. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/upscale.comp +90 -12
  569. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp +335 -151
  570. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/xielu.comp +35 -0
  571. data/ext/sources/ggml/src/ggml-webgpu/CMakeLists.txt +28 -2
  572. data/ext/sources/ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp +169 -0
  573. data/ext/sources/ggml/src/ggml-webgpu/ggml-webgpu.cpp +1964 -435
  574. data/ext/sources/ggml/src/ggml-webgpu/pre_wgsl.hpp +778 -0
  575. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/bin_op.tmpl.wgsl +188 -0
  576. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/cpy.tmpl.wgsl +101 -0
  577. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/embed_wgsl.py +33 -10
  578. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn.wgsl +591 -0
  579. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/get_rows.tmpl.wgsl +1 -1
  580. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/glu.tmpl.wgsl +323 -0
  581. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat.tmpl.wgsl +6 -6
  582. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_decls.tmpl +97 -0
  583. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_reg_tile.tmpl.wgsl +247 -0
  584. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_subgroup_matrix.tmpl.wgsl +302 -0
  585. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_vec.tmpl.wgsl +267 -0
  586. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rms_norm.wgsl +83 -17
  587. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rope.tmpl.wgsl +295 -0
  588. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/scale.tmpl.wgsl +90 -0
  589. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/set_rows.tmpl.wgsl +112 -0
  590. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/soft_max.tmpl.wgsl +345 -0
  591. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/unary_op.wgsl +483 -0
  592. data/ext/sources/ggml/src/ggml-zendnn/CMakeLists.txt +92 -0
  593. data/ext/sources/ggml/src/ggml-zendnn/ggml-zendnn.cpp +466 -0
  594. data/ext/sources/ggml/src/ggml.c +425 -33
  595. data/ext/sources/include/whisper.h +1 -0
  596. data/ext/sources/src/CMakeLists.txt +3 -1
  597. data/ext/sources/src/whisper.cpp +101 -35
  598. data/ext/sources/tests/CMakeLists.txt +2 -2
  599. data/ext/sources/tests/test-vad-full.cpp +4 -2
  600. data/ext/sources/tests/test-vad.cpp +1 -1
  601. data/extsources.rb +1 -0
  602. data/lib/whisper/model/uri.rb +17 -18
  603. data/sig/whisper.rbs +119 -2
  604. data/test/test_params.rb +16 -8
  605. data/test/test_segment.rb +0 -1
  606. data/test/test_token.rb +70 -0
  607. data/test/test_vad.rb +1 -1
  608. data/test/test_vad_context.rb +50 -0
  609. data/test/test_vad_segment.rb +19 -0
  610. data/test/test_vad_segments.rb +16 -0
  611. data/test/test_whisper.rb +7 -0
  612. data/whispercpp.gemspec +1 -1
  613. metadata +287 -34
  614. data/ext/sources/build-xcframework.sh +0 -571
  615. data/ext/sources/ggml/src/ggml-cann/Doxyfile +0 -2579
  616. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_funcs.comp +0 -105
  617. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_head.comp +0 -55
  618. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/add.tmpl.wgsl +0 -44
  619. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/add_in_place.tmpl.wgsl +0 -41
  620. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/cpy.wgsl +0 -60
  621. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul.tmpl.wgsl +0 -44
  622. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_in_place.tmpl.wgsl +0 -41
  623. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rms_norm_in_place.wgsl +0 -48
  624. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{test_bfloat16_support.comp → feature-tests/bfloat16.comp} +0 -0
  625. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{test_coopmat_support.comp → feature-tests/coopmat.comp} +0 -0
  626. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{test_coopmat2_support.comp → feature-tests/coopmat2.comp} +0 -0
  627. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{test_integer_dot_support.comp → feature-tests/integer_dot.comp} +0 -0
  628. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{glu_main.comp → glu_main.glsl} +0 -0
  629. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{rte.comp → rte.glsl} +0 -0
  630. /data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/{utils.comp → utils.glsl} +0 -0
metadata CHANGED
@@ -1,15 +1,14 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: whispercpp
3
3
  version: !ruby/object:Gem::Version
4
- version: 1.3.4
4
+ version: 1.3.5
5
5
  platform: ruby
6
6
  authors:
7
7
  - Georgi Gerganov
8
8
  - Todd A. Fisher
9
- autorequire:
10
9
  bindir: bin
11
10
  cert_chain: []
12
- date: 2025-10-08 00:00:00.000000000 Z
11
+ date: 1980-01-02 00:00:00.000000000 Z
13
12
  dependencies: []
14
13
  description: High-performance inference of OpenAI's Whisper automatic speech recognition
15
14
  (ASR) model via Ruby
@@ -36,8 +35,13 @@ files:
36
35
  - ext/ruby_whisper_model.c
37
36
  - ext/ruby_whisper_params.c
38
37
  - ext/ruby_whisper_segment.c
38
+ - ext/ruby_whisper_token.c
39
39
  - ext/ruby_whisper_transcribe.cpp
40
+ - ext/ruby_whisper_vad_context.c
41
+ - ext/ruby_whisper_vad_context_detect.cpp
40
42
  - ext/ruby_whisper_vad_params.c
43
+ - ext/ruby_whisper_vad_segment.c
44
+ - ext/ruby_whisper_vad_segments.c
41
45
  - ext/sources/CMakeGraphVizOptions.cmake
42
46
  - ext/sources/CMakeLists.txt
43
47
  - ext/sources/bindings/javascript/CMakeLists.txt
@@ -46,13 +50,16 @@ files:
46
50
  - ext/sources/bindings/javascript/package-tmpl.json
47
51
  - ext/sources/bindings/javascript/package.json
48
52
  - ext/sources/bindings/javascript/whisper.js
49
- - ext/sources/build-xcframework.sh
50
53
  - ext/sources/cmake/DefaultTargetOptions.cmake
51
54
  - ext/sources/cmake/FindFFmpeg.cmake
55
+ - ext/sources/cmake/arm64-apple-clang.cmake
56
+ - ext/sources/cmake/arm64-windows-llvm.cmake
52
57
  - ext/sources/cmake/build-info.cmake
53
58
  - ext/sources/cmake/git-vars.cmake
59
+ - ext/sources/cmake/riscv64-spacemit-linux-gnu-gcc.cmake
54
60
  - ext/sources/cmake/whisper-config.cmake.in
55
61
  - ext/sources/cmake/whisper.pc.in
62
+ - ext/sources/cmake/x64-windows-llvm.cmake
56
63
  - ext/sources/examples/CMakeLists.txt
57
64
  - ext/sources/examples/addon.node/CMakeLists.txt
58
65
  - ext/sources/examples/addon.node/__test__/whisper.spec.js
@@ -165,6 +172,110 @@ files:
165
172
  - ext/sources/examples/talk-llama/llama-vocab.h
166
173
  - ext/sources/examples/talk-llama/llama.cpp
167
174
  - ext/sources/examples/talk-llama/llama.h
175
+ - ext/sources/examples/talk-llama/models/afmoe.cpp
176
+ - ext/sources/examples/talk-llama/models/apertus.cpp
177
+ - ext/sources/examples/talk-llama/models/arcee.cpp
178
+ - ext/sources/examples/talk-llama/models/arctic.cpp
179
+ - ext/sources/examples/talk-llama/models/arwkv7.cpp
180
+ - ext/sources/examples/talk-llama/models/baichuan.cpp
181
+ - ext/sources/examples/talk-llama/models/bailingmoe.cpp
182
+ - ext/sources/examples/talk-llama/models/bailingmoe2.cpp
183
+ - ext/sources/examples/talk-llama/models/bert.cpp
184
+ - ext/sources/examples/talk-llama/models/bitnet.cpp
185
+ - ext/sources/examples/talk-llama/models/bloom.cpp
186
+ - ext/sources/examples/talk-llama/models/chameleon.cpp
187
+ - ext/sources/examples/talk-llama/models/chatglm.cpp
188
+ - ext/sources/examples/talk-llama/models/codeshell.cpp
189
+ - ext/sources/examples/talk-llama/models/cogvlm.cpp
190
+ - ext/sources/examples/talk-llama/models/cohere2-iswa.cpp
191
+ - ext/sources/examples/talk-llama/models/command-r.cpp
192
+ - ext/sources/examples/talk-llama/models/dbrx.cpp
193
+ - ext/sources/examples/talk-llama/models/deci.cpp
194
+ - ext/sources/examples/talk-llama/models/deepseek.cpp
195
+ - ext/sources/examples/talk-llama/models/deepseek2.cpp
196
+ - ext/sources/examples/talk-llama/models/dots1.cpp
197
+ - ext/sources/examples/talk-llama/models/dream.cpp
198
+ - ext/sources/examples/talk-llama/models/ernie4-5-moe.cpp
199
+ - ext/sources/examples/talk-llama/models/ernie4-5.cpp
200
+ - ext/sources/examples/talk-llama/models/exaone.cpp
201
+ - ext/sources/examples/talk-llama/models/exaone4.cpp
202
+ - ext/sources/examples/talk-llama/models/falcon-h1.cpp
203
+ - ext/sources/examples/talk-llama/models/falcon.cpp
204
+ - ext/sources/examples/talk-llama/models/gemma-embedding.cpp
205
+ - ext/sources/examples/talk-llama/models/gemma.cpp
206
+ - ext/sources/examples/talk-llama/models/gemma2-iswa.cpp
207
+ - ext/sources/examples/talk-llama/models/gemma3.cpp
208
+ - ext/sources/examples/talk-llama/models/gemma3n-iswa.cpp
209
+ - ext/sources/examples/talk-llama/models/glm4-moe.cpp
210
+ - ext/sources/examples/talk-llama/models/glm4.cpp
211
+ - ext/sources/examples/talk-llama/models/gpt2.cpp
212
+ - ext/sources/examples/talk-llama/models/gptneox.cpp
213
+ - ext/sources/examples/talk-llama/models/granite-hybrid.cpp
214
+ - ext/sources/examples/talk-llama/models/granite.cpp
215
+ - ext/sources/examples/talk-llama/models/graph-context-mamba.cpp
216
+ - ext/sources/examples/talk-llama/models/grok.cpp
217
+ - ext/sources/examples/talk-llama/models/grovemoe.cpp
218
+ - ext/sources/examples/talk-llama/models/hunyuan-dense.cpp
219
+ - ext/sources/examples/talk-llama/models/hunyuan-moe.cpp
220
+ - ext/sources/examples/talk-llama/models/internlm2.cpp
221
+ - ext/sources/examples/talk-llama/models/jais.cpp
222
+ - ext/sources/examples/talk-llama/models/jamba.cpp
223
+ - ext/sources/examples/talk-llama/models/lfm2.cpp
224
+ - ext/sources/examples/talk-llama/models/llada-moe.cpp
225
+ - ext/sources/examples/talk-llama/models/llada.cpp
226
+ - ext/sources/examples/talk-llama/models/llama-iswa.cpp
227
+ - ext/sources/examples/talk-llama/models/llama.cpp
228
+ - ext/sources/examples/talk-llama/models/maincoder.cpp
229
+ - ext/sources/examples/talk-llama/models/mamba.cpp
230
+ - ext/sources/examples/talk-llama/models/mimo2-iswa.cpp
231
+ - ext/sources/examples/talk-llama/models/minicpm3.cpp
232
+ - ext/sources/examples/talk-llama/models/minimax-m2.cpp
233
+ - ext/sources/examples/talk-llama/models/mistral3.cpp
234
+ - ext/sources/examples/talk-llama/models/models.h
235
+ - ext/sources/examples/talk-llama/models/modern-bert.cpp
236
+ - ext/sources/examples/talk-llama/models/mpt.cpp
237
+ - ext/sources/examples/talk-llama/models/nemotron-h.cpp
238
+ - ext/sources/examples/talk-llama/models/nemotron.cpp
239
+ - ext/sources/examples/talk-llama/models/neo-bert.cpp
240
+ - ext/sources/examples/talk-llama/models/olmo.cpp
241
+ - ext/sources/examples/talk-llama/models/olmo2.cpp
242
+ - ext/sources/examples/talk-llama/models/olmoe.cpp
243
+ - ext/sources/examples/talk-llama/models/openai-moe-iswa.cpp
244
+ - ext/sources/examples/talk-llama/models/openelm.cpp
245
+ - ext/sources/examples/talk-llama/models/orion.cpp
246
+ - ext/sources/examples/talk-llama/models/pangu-embedded.cpp
247
+ - ext/sources/examples/talk-llama/models/phi2.cpp
248
+ - ext/sources/examples/talk-llama/models/phi3.cpp
249
+ - ext/sources/examples/talk-llama/models/plamo.cpp
250
+ - ext/sources/examples/talk-llama/models/plamo2.cpp
251
+ - ext/sources/examples/talk-llama/models/plamo3.cpp
252
+ - ext/sources/examples/talk-llama/models/plm.cpp
253
+ - ext/sources/examples/talk-llama/models/qwen.cpp
254
+ - ext/sources/examples/talk-llama/models/qwen2.cpp
255
+ - ext/sources/examples/talk-llama/models/qwen2moe.cpp
256
+ - ext/sources/examples/talk-llama/models/qwen2vl.cpp
257
+ - ext/sources/examples/talk-llama/models/qwen3.cpp
258
+ - ext/sources/examples/talk-llama/models/qwen3moe.cpp
259
+ - ext/sources/examples/talk-llama/models/qwen3next.cpp
260
+ - ext/sources/examples/talk-llama/models/qwen3vl-moe.cpp
261
+ - ext/sources/examples/talk-llama/models/qwen3vl.cpp
262
+ - ext/sources/examples/talk-llama/models/refact.cpp
263
+ - ext/sources/examples/talk-llama/models/rnd1.cpp
264
+ - ext/sources/examples/talk-llama/models/rwkv6-base.cpp
265
+ - ext/sources/examples/talk-llama/models/rwkv6.cpp
266
+ - ext/sources/examples/talk-llama/models/rwkv6qwen2.cpp
267
+ - ext/sources/examples/talk-llama/models/rwkv7-base.cpp
268
+ - ext/sources/examples/talk-llama/models/rwkv7.cpp
269
+ - ext/sources/examples/talk-llama/models/seed-oss.cpp
270
+ - ext/sources/examples/talk-llama/models/smallthinker.cpp
271
+ - ext/sources/examples/talk-llama/models/smollm3.cpp
272
+ - ext/sources/examples/talk-llama/models/stablelm.cpp
273
+ - ext/sources/examples/talk-llama/models/starcoder.cpp
274
+ - ext/sources/examples/talk-llama/models/starcoder2.cpp
275
+ - ext/sources/examples/talk-llama/models/t5-dec.cpp
276
+ - ext/sources/examples/talk-llama/models/t5-enc.cpp
277
+ - ext/sources/examples/talk-llama/models/wavtokenizer-dec.cpp
278
+ - ext/sources/examples/talk-llama/models/xverse.cpp
168
279
  - ext/sources/examples/talk-llama/prompts/talk-alpaca.txt
169
280
  - ext/sources/examples/talk-llama/speak
170
281
  - ext/sources/examples/talk-llama/speak.bat
@@ -200,6 +311,7 @@ files:
200
311
  - ext/sources/ggml/include/ggml-cpp.h
201
312
  - ext/sources/ggml/include/ggml-cpu.h
202
313
  - ext/sources/ggml/include/ggml-cuda.h
314
+ - ext/sources/ggml/include/ggml-hexagon.h
203
315
  - ext/sources/ggml/include/ggml-metal.h
204
316
  - ext/sources/ggml/include/ggml-opencl.h
205
317
  - ext/sources/ggml/include/ggml-opt.h
@@ -208,6 +320,7 @@ files:
208
320
  - ext/sources/ggml/include/ggml-vulkan.h
209
321
  - ext/sources/ggml/include/ggml-webgpu.h
210
322
  - ext/sources/ggml/include/ggml-zdnn.h
323
+ - ext/sources/ggml/include/ggml-zendnn.h
211
324
  - ext/sources/ggml/include/ggml.h
212
325
  - ext/sources/ggml/include/gguf.h
213
326
  - ext/sources/ggml/src/CMakeLists.txt
@@ -218,7 +331,6 @@ files:
218
331
  - ext/sources/ggml/src/ggml-blas/CMakeLists.txt
219
332
  - ext/sources/ggml/src/ggml-blas/ggml-blas.cpp
220
333
  - ext/sources/ggml/src/ggml-cann/CMakeLists.txt
221
- - ext/sources/ggml/src/ggml-cann/Doxyfile
222
334
  - ext/sources/ggml/src/ggml-cann/acl_tensor.cpp
223
335
  - ext/sources/ggml/src/ggml-cann/acl_tensor.h
224
336
  - ext/sources/ggml/src/ggml-cann/aclnn_ops.cpp
@@ -239,8 +351,10 @@ files:
239
351
  - ext/sources/ggml/src/ggml-cpu/arch/loongarch/quants.c
240
352
  - ext/sources/ggml/src/ggml-cpu/arch/powerpc/cpu-feats.cpp
241
353
  - ext/sources/ggml/src/ggml-cpu/arch/powerpc/quants.c
354
+ - ext/sources/ggml/src/ggml-cpu/arch/riscv/cpu-feats.cpp
242
355
  - ext/sources/ggml/src/ggml-cpu/arch/riscv/quants.c
243
356
  - ext/sources/ggml/src/ggml-cpu/arch/riscv/repack.cpp
357
+ - ext/sources/ggml/src/ggml-cpu/arch/s390/cpu-feats.cpp
244
358
  - ext/sources/ggml/src/ggml-cpu/arch/s390/quants.c
245
359
  - ext/sources/ggml/src/ggml-cpu/arch/wasm/quants.c
246
360
  - ext/sources/ggml/src/ggml-cpu/arch/x86/cpu-feats.cpp
@@ -259,6 +373,7 @@ files:
259
373
  - ext/sources/ggml/src/ggml-cpu/kleidiai/kernels.h
260
374
  - ext/sources/ggml/src/ggml-cpu/kleidiai/kleidiai.cpp
261
375
  - ext/sources/ggml/src/ggml-cpu/kleidiai/kleidiai.h
376
+ - ext/sources/ggml/src/ggml-cpu/llamafile/sgemm-ppc.h
262
377
  - ext/sources/ggml/src/ggml-cpu/llamafile/sgemm.cpp
263
378
  - ext/sources/ggml/src/ggml-cpu/llamafile/sgemm.h
264
379
  - ext/sources/ggml/src/ggml-cpu/ops.cpp
@@ -314,7 +429,11 @@ files:
314
429
  - ext/sources/ggml/src/ggml-cuda/cpy.cuh
315
430
  - ext/sources/ggml/src/ggml-cuda/cross-entropy-loss.cu
316
431
  - ext/sources/ggml/src/ggml-cuda/cross-entropy-loss.cuh
432
+ - ext/sources/ggml/src/ggml-cuda/cumsum.cu
433
+ - ext/sources/ggml/src/ggml-cuda/cumsum.cuh
317
434
  - ext/sources/ggml/src/ggml-cuda/dequantize.cuh
435
+ - ext/sources/ggml/src/ggml-cuda/diag.cu
436
+ - ext/sources/ggml/src/ggml-cuda/diag.cuh
318
437
  - ext/sources/ggml/src/ggml-cuda/diagmask.cu
319
438
  - ext/sources/ggml/src/ggml-cuda/diagmask.cuh
320
439
  - ext/sources/ggml/src/ggml-cuda/fattn-common.cuh
@@ -326,6 +445,8 @@ files:
326
445
  - ext/sources/ggml/src/ggml-cuda/fattn-wmma-f16.cuh
327
446
  - ext/sources/ggml/src/ggml-cuda/fattn.cu
328
447
  - ext/sources/ggml/src/ggml-cuda/fattn.cuh
448
+ - ext/sources/ggml/src/ggml-cuda/fill.cu
449
+ - ext/sources/ggml/src/ggml-cuda/fill.cuh
329
450
  - ext/sources/ggml/src/ggml-cuda/getrows.cu
330
451
  - ext/sources/ggml/src/ggml-cuda/getrows.cuh
331
452
  - ext/sources/ggml/src/ggml-cuda/ggml-cuda.cu
@@ -338,6 +459,8 @@ files:
338
459
  - ext/sources/ggml/src/ggml-cuda/mma.cuh
339
460
  - ext/sources/ggml/src/ggml-cuda/mmf.cu
340
461
  - ext/sources/ggml/src/ggml-cuda/mmf.cuh
462
+ - ext/sources/ggml/src/ggml-cuda/mmid.cu
463
+ - ext/sources/ggml/src/ggml-cuda/mmid.cuh
341
464
  - ext/sources/ggml/src/ggml-cuda/mmq.cu
342
465
  - ext/sources/ggml/src/ggml-cuda/mmq.cuh
343
466
  - ext/sources/ggml/src/ggml-cuda/mmvf.cu
@@ -369,10 +492,14 @@ files:
369
492
  - ext/sources/ggml/src/ggml-cuda/scale.cuh
370
493
  - ext/sources/ggml/src/ggml-cuda/set-rows.cu
371
494
  - ext/sources/ggml/src/ggml-cuda/set-rows.cuh
495
+ - ext/sources/ggml/src/ggml-cuda/set.cu
496
+ - ext/sources/ggml/src/ggml-cuda/set.cuh
372
497
  - ext/sources/ggml/src/ggml-cuda/softcap.cu
373
498
  - ext/sources/ggml/src/ggml-cuda/softcap.cuh
374
499
  - ext/sources/ggml/src/ggml-cuda/softmax.cu
375
500
  - ext/sources/ggml/src/ggml-cuda/softmax.cuh
501
+ - ext/sources/ggml/src/ggml-cuda/solve_tri.cu
502
+ - ext/sources/ggml/src/ggml-cuda/solve_tri.cuh
376
503
  - ext/sources/ggml/src/ggml-cuda/ssm-conv.cu
377
504
  - ext/sources/ggml/src/ggml-cuda/ssm-conv.cuh
378
505
  - ext/sources/ggml/src/ggml-cuda/ssm-scan.cu
@@ -400,6 +527,15 @@ files:
400
527
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu
401
528
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu
402
529
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu
530
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq112-dv112.cu
531
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq128-dv128.cu
532
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq256-dv256.cu
533
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq40-dv40.cu
534
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq576-dv512.cu
535
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq64-dv64.cu
536
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq72-dv72.cu
537
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq80-dv80.cu
538
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-tile-instance-dkq96-dv96.cu
403
539
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu
404
540
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_0.cu
405
541
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_1.cu
@@ -472,8 +608,12 @@ files:
472
608
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu
473
609
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu
474
610
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu
611
+ - ext/sources/ggml/src/ggml-cuda/top-k.cu
612
+ - ext/sources/ggml/src/ggml-cuda/top-k.cuh
475
613
  - ext/sources/ggml/src/ggml-cuda/topk-moe.cu
476
614
  - ext/sources/ggml/src/ggml-cuda/topk-moe.cuh
615
+ - ext/sources/ggml/src/ggml-cuda/tri.cu
616
+ - ext/sources/ggml/src/ggml-cuda/tri.cuh
477
617
  - ext/sources/ggml/src/ggml-cuda/tsembd.cu
478
618
  - ext/sources/ggml/src/ggml-cuda/tsembd.cuh
479
619
  - ext/sources/ggml/src/ggml-cuda/unary.cu
@@ -486,6 +626,37 @@ files:
486
626
  - ext/sources/ggml/src/ggml-cuda/vendors/musa.h
487
627
  - ext/sources/ggml/src/ggml-cuda/wkv.cu
488
628
  - ext/sources/ggml/src/ggml-cuda/wkv.cuh
629
+ - ext/sources/ggml/src/ggml-hexagon/CMakeLists.txt
630
+ - ext/sources/ggml/src/ggml-hexagon/ggml-hexagon.cpp
631
+ - ext/sources/ggml/src/ggml-hexagon/htp-utils.c
632
+ - ext/sources/ggml/src/ggml-hexagon/htp-utils.h
633
+ - ext/sources/ggml/src/ggml-hexagon/htp/CMakeLists.txt
634
+ - ext/sources/ggml/src/ggml-hexagon/htp/act-ops.c
635
+ - ext/sources/ggml/src/ggml-hexagon/htp/binary-ops.c
636
+ - ext/sources/ggml/src/ggml-hexagon/htp/cmake-toolchain.cmake
637
+ - ext/sources/ggml/src/ggml-hexagon/htp/flash-attn-ops.c
638
+ - ext/sources/ggml/src/ggml-hexagon/htp/get-rows-ops.c
639
+ - ext/sources/ggml/src/ggml-hexagon/htp/htp-ctx.h
640
+ - ext/sources/ggml/src/ggml-hexagon/htp/htp-dma.c
641
+ - ext/sources/ggml/src/ggml-hexagon/htp/htp-dma.h
642
+ - ext/sources/ggml/src/ggml-hexagon/htp/htp-msg.h
643
+ - ext/sources/ggml/src/ggml-hexagon/htp/htp-ops.h
644
+ - ext/sources/ggml/src/ggml-hexagon/htp/htp_iface.idl
645
+ - ext/sources/ggml/src/ggml-hexagon/htp/hvx-exp.c
646
+ - ext/sources/ggml/src/ggml-hexagon/htp/hvx-inverse.c
647
+ - ext/sources/ggml/src/ggml-hexagon/htp/hvx-sigmoid.c
648
+ - ext/sources/ggml/src/ggml-hexagon/htp/hvx-utils.c
649
+ - ext/sources/ggml/src/ggml-hexagon/htp/hvx-utils.h
650
+ - ext/sources/ggml/src/ggml-hexagon/htp/main.c
651
+ - ext/sources/ggml/src/ggml-hexagon/htp/matmul-ops.c
652
+ - ext/sources/ggml/src/ggml-hexagon/htp/ops-utils.h
653
+ - ext/sources/ggml/src/ggml-hexagon/htp/rope-ops.c
654
+ - ext/sources/ggml/src/ggml-hexagon/htp/set-rows-ops.c
655
+ - ext/sources/ggml/src/ggml-hexagon/htp/softmax-ops.c
656
+ - ext/sources/ggml/src/ggml-hexagon/htp/unary-ops.c
657
+ - ext/sources/ggml/src/ggml-hexagon/htp/worker-pool.c
658
+ - ext/sources/ggml/src/ggml-hexagon/htp/worker-pool.h
659
+ - ext/sources/ggml/src/ggml-hexagon/op-desc.h
489
660
  - ext/sources/ggml/src/ggml-hip/CMakeLists.txt
490
661
  - ext/sources/ggml/src/ggml-impl.h
491
662
  - ext/sources/ggml/src/ggml-metal/CMakeLists.txt
@@ -518,10 +689,14 @@ files:
518
689
  - ext/sources/ggml/src/ggml-opencl/kernels/diag_mask_inf.cl
519
690
  - ext/sources/ggml/src/ggml-opencl/kernels/div.cl
520
691
  - ext/sources/ggml/src/ggml-opencl/kernels/embed_kernel.py
692
+ - ext/sources/ggml/src/ggml-opencl/kernels/expm1.cl
693
+ - ext/sources/ggml/src/ggml-opencl/kernels/fill.cl
521
694
  - ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f16.cl
522
695
  - ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32.cl
523
696
  - ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32_f16.cl
524
697
  - ext/sources/ggml/src/ggml-opencl/kernels/gelu.cl
698
+ - ext/sources/ggml/src/ggml-opencl/kernels/gemm_moe_mxfp4_f32.cl
699
+ - ext/sources/ggml/src/ggml-opencl/kernels/gemv_moe_mxfp4_f32.cl
525
700
  - ext/sources/ggml/src/ggml-opencl/kernels/gemv_noshuffle.cl
526
701
  - ext/sources/ggml/src/ggml-opencl/kernels/gemv_noshuffle_general.cl
527
702
  - ext/sources/ggml/src/ggml-opencl/kernels/get_rows.cl
@@ -529,11 +704,14 @@ files:
529
704
  - ext/sources/ggml/src/ggml-opencl/kernels/group_norm.cl
530
705
  - ext/sources/ggml/src/ggml-opencl/kernels/im2col_f16.cl
531
706
  - ext/sources/ggml/src/ggml-opencl/kernels/im2col_f32.cl
707
+ - ext/sources/ggml/src/ggml-opencl/kernels/mean.cl
532
708
  - ext/sources/ggml/src/ggml-opencl/kernels/mul.cl
533
709
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mat_Ab_Bi_8x4.cl
534
710
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mat_f16_f32.cl
711
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f16_f32_kq_kqv.cl
535
712
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f16_f32_l4_lm.cl
536
713
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f32_f32_l4_lm.cl
714
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_q8_0_f32_l4_lm.cl
537
715
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f16.cl
538
716
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f32.cl
539
717
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f32_1row.cl
@@ -568,6 +746,10 @@ files:
568
746
  - ext/sources/ggml/src/ggml-opencl/kernels/softmax_4_f32.cl
569
747
  - ext/sources/ggml/src/ggml-opencl/kernels/softmax_f16.cl
570
748
  - ext/sources/ggml/src/ggml-opencl/kernels/softmax_f32.cl
749
+ - ext/sources/ggml/src/ggml-opencl/kernels/softplus.cl
750
+ - ext/sources/ggml/src/ggml-opencl/kernels/sqr.cl
751
+ - ext/sources/ggml/src/ggml-opencl/kernels/sqrt.cl
752
+ - ext/sources/ggml/src/ggml-opencl/kernels/ssm_conv.cl
571
753
  - ext/sources/ggml/src/ggml-opencl/kernels/sub.cl
572
754
  - ext/sources/ggml/src/ggml-opencl/kernels/sum_rows.cl
573
755
  - ext/sources/ggml/src/ggml-opencl/kernels/tanh.cl
@@ -580,6 +762,8 @@ files:
580
762
  - ext/sources/ggml/src/ggml-rpc/CMakeLists.txt
581
763
  - ext/sources/ggml/src/ggml-rpc/ggml-rpc.cpp
582
764
  - ext/sources/ggml/src/ggml-sycl/CMakeLists.txt
765
+ - ext/sources/ggml/src/ggml-sycl/add-id.cpp
766
+ - ext/sources/ggml/src/ggml-sycl/add-id.hpp
583
767
  - ext/sources/ggml/src/ggml-sycl/backend.hpp
584
768
  - ext/sources/ggml/src/ggml-sycl/binbcast.cpp
585
769
  - ext/sources/ggml/src/ggml-sycl/binbcast.hpp
@@ -591,6 +775,8 @@ files:
591
775
  - ext/sources/ggml/src/ggml-sycl/conv.hpp
592
776
  - ext/sources/ggml/src/ggml-sycl/convert.cpp
593
777
  - ext/sources/ggml/src/ggml-sycl/convert.hpp
778
+ - ext/sources/ggml/src/ggml-sycl/count-equal.cpp
779
+ - ext/sources/ggml/src/ggml-sycl/count-equal.hpp
594
780
  - ext/sources/ggml/src/ggml-sycl/cpy.cpp
595
781
  - ext/sources/ggml/src/ggml-sycl/cpy.hpp
596
782
  - ext/sources/ggml/src/ggml-sycl/dequantize.hpp
@@ -615,15 +801,27 @@ files:
615
801
  - ext/sources/ggml/src/ggml-sycl/norm.hpp
616
802
  - ext/sources/ggml/src/ggml-sycl/outprod.cpp
617
803
  - ext/sources/ggml/src/ggml-sycl/outprod.hpp
804
+ - ext/sources/ggml/src/ggml-sycl/pad.cpp
805
+ - ext/sources/ggml/src/ggml-sycl/pad.hpp
806
+ - ext/sources/ggml/src/ggml-sycl/pad_reflect_1d.cpp
807
+ - ext/sources/ggml/src/ggml-sycl/pad_reflect_1d.hpp
618
808
  - ext/sources/ggml/src/ggml-sycl/presets.hpp
619
809
  - ext/sources/ggml/src/ggml-sycl/quantize.hpp
620
810
  - ext/sources/ggml/src/ggml-sycl/quants.hpp
811
+ - ext/sources/ggml/src/ggml-sycl/repeat_back.cpp
812
+ - ext/sources/ggml/src/ggml-sycl/repeat_back.hpp
813
+ - ext/sources/ggml/src/ggml-sycl/roll.cpp
814
+ - ext/sources/ggml/src/ggml-sycl/roll.hpp
621
815
  - ext/sources/ggml/src/ggml-sycl/rope.cpp
622
816
  - ext/sources/ggml/src/ggml-sycl/rope.hpp
817
+ - ext/sources/ggml/src/ggml-sycl/set.cpp
818
+ - ext/sources/ggml/src/ggml-sycl/set.hpp
623
819
  - ext/sources/ggml/src/ggml-sycl/set_rows.cpp
624
820
  - ext/sources/ggml/src/ggml-sycl/set_rows.hpp
625
821
  - ext/sources/ggml/src/ggml-sycl/softmax.cpp
626
822
  - ext/sources/ggml/src/ggml-sycl/softmax.hpp
823
+ - ext/sources/ggml/src/ggml-sycl/ssm_conv.cpp
824
+ - ext/sources/ggml/src/ggml-sycl/ssm_conv.hpp
627
825
  - ext/sources/ggml/src/ggml-sycl/sycl_hw.cpp
628
826
  - ext/sources/ggml/src/ggml-sycl/sycl_hw.hpp
629
827
  - ext/sources/ggml/src/ggml-sycl/tsembd.cpp
@@ -637,11 +835,16 @@ files:
637
835
  - ext/sources/ggml/src/ggml-vulkan/cmake/host-toolchain.cmake.in
638
836
  - ext/sources/ggml/src/ggml-vulkan/ggml-vulkan.cpp
639
837
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/CMakeLists.txt
838
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/abs.comp
640
839
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/acc.comp
641
840
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add.comp
841
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add1.comp
642
842
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add_id.comp
843
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/arange.comp
643
844
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argmax.comp
644
845
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argsort.comp
846
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argsort_large.comp
847
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/ceil.comp
645
848
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/clamp.comp
646
849
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/concat.comp
647
850
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/contig_copy.comp
@@ -651,12 +854,17 @@ files:
651
854
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy.comp
652
855
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_from_quant.comp
653
856
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_to_quant.comp
857
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_transpose.comp
654
858
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cos.comp
655
859
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/count_equal.comp
860
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/count_experts.comp
861
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cumsum.comp
862
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cumsum_multipass1.comp
863
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/cumsum_multipass2.comp
656
864
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_f32.comp
657
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs.comp
658
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs_cm2.comp
659
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_head.comp
865
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs.glsl
866
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs_cm2.glsl
867
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_head.glsl
660
868
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq1_m.comp
661
869
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq1_s.comp
662
870
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq2_s.comp
@@ -677,27 +885,34 @@ files:
677
885
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q5_k.comp
678
886
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q6_k.comp
679
887
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q8_0.comp
888
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/diag.comp
680
889
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/diag_mask_inf.comp
681
890
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/div.comp
682
891
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/exp.comp
892
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/feature-tests/bfloat16.comp
893
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/feature-tests/coopmat.comp
894
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/feature-tests/coopmat2.comp
895
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/feature-tests/integer_dot.comp
896
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/fill.comp
683
897
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp
684
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.comp
898
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.glsl
685
899
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp
686
900
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm2.comp
687
901
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_split_k_reduce.comp
902
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/floor.comp
688
903
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu.comp
689
904
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_erf.comp
690
905
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_quick.comp
691
906
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu.comp
692
907
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_erf.comp
693
908
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_quick.comp
694
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_binary_head.comp
695
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_head.comp
696
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_unary_head.comp
909
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_binary_head.glsl
910
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_head.glsl
911
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_unary_head.glsl
697
912
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/get_rows.comp
698
913
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/get_rows_quant.comp
699
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_head.comp
700
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_main.comp
914
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_head.glsl
915
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_main.glsl
701
916
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/group_norm.comp
702
917
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardsigmoid.comp
703
918
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardswish.comp
@@ -705,10 +920,12 @@ files:
705
920
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col_3d.comp
706
921
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/l2_norm.comp
707
922
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/leaky_relu.comp
923
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/log.comp
708
924
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul.comp
709
925
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_split_k_reduce.comp
710
926
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec.comp
711
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_base.comp
927
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_base.glsl
928
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iface.glsl
712
929
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq1_m.comp
713
930
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq1_s.comp
714
931
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_iq2_s.comp
@@ -724,12 +941,16 @@ files:
724
941
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q5_k.comp
725
942
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q6_k.comp
726
943
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq.comp
944
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq_funcs.glsl
727
945
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp
728
946
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_cm2.comp
729
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_funcs.comp
947
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_funcs.glsl
948
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_id_funcs.glsl
730
949
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq.comp
731
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_funcs.comp
950
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_funcs.glsl
951
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_shmem_types.glsl
732
952
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/multi_add.comp
953
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/neg.comp
733
954
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/norm.comp
734
955
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_adamw.comp
735
956
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_sgd.comp
@@ -744,12 +965,15 @@ files:
744
965
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_back.comp
745
966
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_partials.comp
746
967
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/roll.comp
747
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_head.comp
968
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_funcs.glsl
969
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_head.glsl
748
970
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_multi.comp
749
971
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_neox.comp
750
972
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_norm.comp
973
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_params.glsl
751
974
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_vision.comp
752
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rte.comp
975
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/round.comp
976
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rte.glsl
753
977
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/scale.comp
754
978
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sigmoid.comp
755
979
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/silu.comp
@@ -757,40 +981,61 @@ files:
757
981
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sin.comp
758
982
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max.comp
759
983
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_back.comp
984
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large1.comp
985
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large2.comp
986
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large3.comp
987
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_large_common.glsl
988
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/softplus.comp
989
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/solve_tri.comp
760
990
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sqrt.comp
761
991
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/square.comp
992
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/ssm_conv.comp
993
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/ssm_scan.comp
994
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/step.comp
762
995
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sub.comp
763
996
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sum_rows.comp
997
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sum_rows.glsl
764
998
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu.comp
765
999
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu_oai.comp
766
1000
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/tanh.comp
767
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/test_bfloat16_support.comp
768
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/test_coopmat2_support.comp
769
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/test_coopmat_support.comp
770
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/test_integer_dot_support.comp
771
1001
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/timestep_embedding.comp
772
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/types.comp
1002
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/topk_argsort.comp
1003
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/topk_moe.comp
1004
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/topk_nary_search.comp
1005
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/tri.comp
1006
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/trunc.comp
1007
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/types.glsl
773
1008
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/upscale.comp
774
- - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/utils.comp
1009
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/utils.glsl
775
1010
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp
776
1011
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/wkv6.comp
777
1012
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/wkv7.comp
1013
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/xielu.comp
778
1014
  - ext/sources/ggml/src/ggml-webgpu/CMakeLists.txt
1015
+ - ext/sources/ggml/src/ggml-webgpu/ggml-webgpu-shader-lib.hpp
779
1016
  - ext/sources/ggml/src/ggml-webgpu/ggml-webgpu.cpp
780
- - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/add.tmpl.wgsl
781
- - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/add_in_place.tmpl.wgsl
1017
+ - ext/sources/ggml/src/ggml-webgpu/pre_wgsl.hpp
1018
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/bin_op.tmpl.wgsl
782
1019
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/binary_head.tmpl
783
1020
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl
784
- - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/cpy.wgsl
1021
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/cpy.tmpl.wgsl
785
1022
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/embed_wgsl.py
1023
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/flash_attn.wgsl
786
1024
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/get_rows.tmpl.wgsl
1025
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/glu.tmpl.wgsl
787
1026
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/memset.wgsl
788
- - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul.tmpl.wgsl
789
- - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_in_place.tmpl.wgsl
790
1027
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat.tmpl.wgsl
1028
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_decls.tmpl
1029
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_reg_tile.tmpl.wgsl
1030
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_subgroup_matrix.tmpl.wgsl
1031
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat_vec.tmpl.wgsl
791
1032
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rms_norm.wgsl
792
- - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rms_norm_in_place.wgsl
1033
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rope.tmpl.wgsl
1034
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/scale.tmpl.wgsl
1035
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/set_rows.tmpl.wgsl
793
1036
  - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/set_rows.wgsl
1037
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/soft_max.tmpl.wgsl
1038
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/unary_op.wgsl
794
1039
  - ext/sources/ggml/src/ggml-zdnn/CMakeLists.txt
795
1040
  - ext/sources/ggml/src/ggml-zdnn/common.hpp
796
1041
  - ext/sources/ggml/src/ggml-zdnn/ggml-zdnn.cpp
@@ -798,6 +1043,8 @@ files:
798
1043
  - ext/sources/ggml/src/ggml-zdnn/mmf.hpp
799
1044
  - ext/sources/ggml/src/ggml-zdnn/utils.cpp
800
1045
  - ext/sources/ggml/src/ggml-zdnn/utils.hpp
1046
+ - ext/sources/ggml/src/ggml-zendnn/CMakeLists.txt
1047
+ - ext/sources/ggml/src/ggml-zendnn/ggml-zendnn.cpp
801
1048
  - ext/sources/ggml/src/ggml.c
802
1049
  - ext/sources/ggml/src/ggml.cpp
803
1050
  - ext/sources/ggml/src/gguf.cpp
@@ -854,15 +1101,18 @@ files:
854
1101
  - test/test_package.rb
855
1102
  - test/test_params.rb
856
1103
  - test/test_segment.rb
1104
+ - test/test_token.rb
857
1105
  - test/test_vad.rb
1106
+ - test/test_vad_context.rb
858
1107
  - test/test_vad_params.rb
1108
+ - test/test_vad_segment.rb
1109
+ - test/test_vad_segments.rb
859
1110
  - test/test_whisper.rb
860
1111
  - whispercpp.gemspec
861
1112
  homepage: https://github.com/ggml-org/whisper.cpp
862
1113
  licenses:
863
1114
  - MIT
864
1115
  metadata: {}
865
- post_install_message:
866
1116
  rdoc_options:
867
1117
  - "--main"
868
1118
  - README.md
@@ -879,8 +1129,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
879
1129
  - !ruby/object:Gem::Version
880
1130
  version: '0'
881
1131
  requirements: []
882
- rubygems_version: 3.5.22
883
- signing_key:
1132
+ rubygems_version: 3.6.9
884
1133
  specification_version: 4
885
1134
  summary: Ruby whisper.cpp bindings
886
1135
  test_files:
@@ -894,6 +1143,10 @@ test_files:
894
1143
  - test/test_package.rb
895
1144
  - test/test_params.rb
896
1145
  - test/test_segment.rb
1146
+ - test/test_token.rb
897
1147
  - test/test_vad.rb
1148
+ - test/test_vad_context.rb
898
1149
  - test/test_vad_params.rb
1150
+ - test/test_vad_segment.rb
1151
+ - test/test_vad_segments.rb
899
1152
  - test/test_whisper.rb