whispercpp 1.3.3 → 1.3.4

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 (586) hide show
  1. checksums.yaml +4 -4
  2. data/ext/ruby_whisper_params.c +55 -25
  3. data/ext/sources/CMakeLists.txt +1 -1
  4. data/ext/sources/bindings/javascript/package.json +1 -1
  5. data/ext/sources/build-xcframework.sh +24 -0
  6. data/ext/sources/examples/CMakeLists.txt +1 -0
  7. data/ext/sources/examples/addon.node/addon.cpp +19 -19
  8. data/ext/sources/examples/addon.node/index.js +7 -5
  9. data/ext/sources/examples/bench/bench.cpp +26 -16
  10. data/ext/sources/examples/bench.wasm/index-tmpl.html +10 -9
  11. data/ext/sources/examples/cli/cli.cpp +4 -2
  12. data/ext/sources/examples/command/command.cpp +26 -24
  13. data/ext/sources/examples/command.wasm/index-tmpl.html +5 -4
  14. data/ext/sources/examples/common-ggml.cpp +2 -0
  15. data/ext/sources/examples/lsp/lsp.cpp +19 -17
  16. data/ext/sources/examples/server/server.cpp +24 -13
  17. data/ext/sources/examples/server.py +6 -1
  18. data/ext/sources/examples/stream/stream.cpp +4 -2
  19. data/ext/sources/examples/stream.wasm/emscripten.cpp +6 -6
  20. data/ext/sources/examples/stream.wasm/index-tmpl.html +82 -5
  21. data/ext/sources/examples/talk-llama/CMakeLists.txt +2 -2
  22. data/ext/sources/examples/talk-llama/llama-adapter.cpp +101 -4
  23. data/ext/sources/examples/talk-llama/llama-adapter.h +6 -0
  24. data/ext/sources/examples/talk-llama/llama-arch.cpp +588 -15
  25. data/ext/sources/examples/talk-llama/llama-arch.h +58 -1
  26. data/ext/sources/examples/talk-llama/llama-batch.cpp +103 -71
  27. data/ext/sources/examples/talk-llama/llama-batch.h +31 -18
  28. data/ext/sources/examples/talk-llama/llama-chat.cpp +120 -5
  29. data/ext/sources/examples/talk-llama/llama-chat.h +7 -0
  30. data/ext/sources/examples/talk-llama/llama-context.cpp +460 -357
  31. data/ext/sources/examples/talk-llama/llama-context.h +44 -29
  32. data/ext/sources/examples/talk-llama/llama-cparams.h +4 -4
  33. data/ext/sources/examples/talk-llama/llama-graph.cpp +543 -271
  34. data/ext/sources/examples/talk-llama/llama-graph.h +278 -168
  35. data/ext/sources/examples/talk-llama/llama-hparams.cpp +118 -4
  36. data/ext/sources/examples/talk-llama/llama-hparams.h +61 -15
  37. data/ext/sources/examples/talk-llama/llama-impl.h +2 -0
  38. data/ext/sources/examples/talk-llama/llama-kv-cache-iswa.cpp +326 -0
  39. data/ext/sources/examples/talk-llama/{llama-kv-cache-unified-iswa.h → llama-kv-cache-iswa.h} +38 -29
  40. data/ext/sources/examples/talk-llama/llama-kv-cache.cpp +2020 -0
  41. data/ext/sources/examples/talk-llama/llama-kv-cache.h +358 -27
  42. data/ext/sources/examples/talk-llama/llama-kv-cells.h +80 -28
  43. data/ext/sources/examples/talk-llama/llama-memory-hybrid.cpp +56 -36
  44. data/ext/sources/examples/talk-llama/llama-memory-hybrid.h +30 -29
  45. data/ext/sources/examples/talk-llama/llama-memory-recurrent.cpp +48 -19
  46. data/ext/sources/examples/talk-llama/llama-memory-recurrent.h +13 -14
  47. data/ext/sources/examples/talk-llama/llama-memory.h +16 -10
  48. data/ext/sources/examples/talk-llama/llama-model-loader.cpp +2 -0
  49. data/ext/sources/examples/talk-llama/llama-model-loader.h +3 -2
  50. data/ext/sources/examples/talk-llama/llama-model.cpp +7165 -2336
  51. data/ext/sources/examples/talk-llama/llama-model.h +60 -9
  52. data/ext/sources/examples/talk-llama/llama-quant.cpp +48 -10
  53. data/ext/sources/examples/talk-llama/llama-sampling.cpp +226 -126
  54. data/ext/sources/examples/talk-llama/llama-vocab.cpp +440 -13
  55. data/ext/sources/examples/talk-llama/llama-vocab.h +45 -0
  56. data/ext/sources/examples/talk-llama/llama.cpp +65 -10
  57. data/ext/sources/examples/talk-llama/llama.h +95 -177
  58. data/ext/sources/examples/talk-llama/talk-llama.cpp +9 -6
  59. data/ext/sources/examples/talk-llama/unicode.cpp +207 -0
  60. data/ext/sources/examples/talk-llama/unicode.h +45 -0
  61. data/ext/sources/examples/wchess/wchess.cmd/wchess.cmd.cpp +4 -2
  62. data/ext/sources/examples/whisper.wasm/index-tmpl.html +17 -16
  63. data/ext/sources/ggml/CMakeLists.txt +59 -31
  64. data/ext/sources/ggml/cmake/ggml-config.cmake.in +132 -93
  65. data/ext/sources/ggml/include/ggml-backend.h +17 -1
  66. data/ext/sources/ggml/include/ggml-cpu.h +1 -1
  67. data/ext/sources/ggml/include/ggml-metal.h +1 -6
  68. data/ext/sources/ggml/include/ggml-opt.h +25 -6
  69. data/ext/sources/ggml/include/ggml-webgpu.h +19 -0
  70. data/ext/sources/ggml/include/ggml-zdnn.h +17 -0
  71. data/ext/sources/ggml/include/ggml.h +221 -16
  72. data/ext/sources/ggml/src/CMakeLists.txt +17 -2
  73. data/ext/sources/ggml/src/ggml-alloc.c +265 -141
  74. data/ext/sources/ggml/src/ggml-backend-impl.h +4 -1
  75. data/ext/sources/ggml/src/ggml-backend-reg.cpp +30 -13
  76. data/ext/sources/ggml/src/ggml-backend.cpp +221 -38
  77. data/ext/sources/ggml/src/ggml-blas/CMakeLists.txt +1 -1
  78. data/ext/sources/ggml/src/ggml-blas/ggml-blas.cpp +5 -4
  79. data/ext/sources/ggml/src/ggml-cann/CMakeLists.txt +14 -0
  80. data/ext/sources/ggml/src/ggml-cann/acl_tensor.cpp +3 -1
  81. data/ext/sources/ggml/src/ggml-cann/aclnn_ops.cpp +903 -717
  82. data/ext/sources/ggml/src/ggml-cann/aclnn_ops.h +143 -25
  83. data/ext/sources/ggml/src/ggml-cann/common.h +143 -1
  84. data/ext/sources/ggml/src/ggml-cann/ggml-cann.cpp +488 -69
  85. data/ext/sources/ggml/src/ggml-common.h +17 -0
  86. data/ext/sources/ggml/src/ggml-cpu/CMakeLists.txt +40 -18
  87. data/ext/sources/ggml/src/ggml-cpu/amx/amx.cpp +4 -2
  88. data/ext/sources/ggml/src/ggml-cpu/arch/arm/quants.c +132 -596
  89. data/ext/sources/ggml/src/ggml-cpu/arch/arm/repack.cpp +14 -286
  90. data/ext/sources/ggml/src/ggml-cpu/arch/loongarch/quants.c +103 -582
  91. data/ext/sources/ggml/src/ggml-cpu/arch/powerpc/quants.c +162 -589
  92. data/ext/sources/ggml/src/ggml-cpu/arch/riscv/quants.c +265 -437
  93. data/ext/sources/ggml/src/ggml-cpu/arch/riscv/repack.cpp +3 -58
  94. data/ext/sources/ggml/src/ggml-cpu/arch/s390/quants.c +521 -353
  95. data/ext/sources/ggml/src/ggml-cpu/arch/wasm/quants.c +54 -314
  96. data/ext/sources/ggml/src/ggml-cpu/arch/x86/quants.c +184 -675
  97. data/ext/sources/ggml/src/ggml-cpu/arch/x86/repack.cpp +4679 -1657
  98. data/ext/sources/ggml/src/ggml-cpu/arch-fallback.h +32 -2
  99. data/ext/sources/ggml/src/ggml-cpu/common.h +14 -0
  100. data/ext/sources/ggml/src/ggml-cpu/ggml-cpu-impl.h +13 -6
  101. data/ext/sources/ggml/src/ggml-cpu/ggml-cpu.c +70 -42
  102. data/ext/sources/ggml/src/ggml-cpu/ggml-cpu.cpp +35 -28
  103. data/ext/sources/ggml/src/ggml-cpu/kleidiai/kernels.cpp +152 -18
  104. data/ext/sources/ggml/src/ggml-cpu/kleidiai/kernels.h +7 -1
  105. data/ext/sources/ggml/src/ggml-cpu/kleidiai/kleidiai.cpp +227 -97
  106. data/ext/sources/ggml/src/ggml-cpu/llamafile/sgemm.cpp +474 -1116
  107. data/ext/sources/ggml/src/ggml-cpu/ops.cpp +1587 -1177
  108. data/ext/sources/ggml/src/ggml-cpu/ops.h +5 -8
  109. data/ext/sources/ggml/src/ggml-cpu/quants.c +35 -0
  110. data/ext/sources/ggml/src/ggml-cpu/quants.h +8 -0
  111. data/ext/sources/ggml/src/ggml-cpu/repack.cpp +458 -47
  112. data/ext/sources/ggml/src/ggml-cpu/repack.h +22 -0
  113. data/ext/sources/ggml/src/ggml-cpu/simd-mappings.h +89 -60
  114. data/ext/sources/ggml/src/ggml-cpu/spacemit/ime.cpp +1024 -0
  115. data/ext/sources/ggml/src/ggml-cpu/spacemit/ime.h +13 -0
  116. data/ext/sources/ggml/src/ggml-cpu/spacemit/ime1_kernels.cpp +3196 -0
  117. data/ext/sources/ggml/src/ggml-cpu/spacemit/ime_kernels.h +26 -0
  118. data/ext/sources/ggml/src/ggml-cpu/traits.cpp +2 -2
  119. data/ext/sources/ggml/src/ggml-cpu/traits.h +1 -1
  120. data/ext/sources/ggml/src/ggml-cpu/vec.cpp +170 -26
  121. data/ext/sources/ggml/src/ggml-cpu/vec.h +506 -63
  122. data/ext/sources/ggml/src/ggml-cuda/CMakeLists.txt +20 -16
  123. data/ext/sources/ggml/src/ggml-cuda/add-id.cu +58 -0
  124. data/ext/sources/ggml/src/ggml-cuda/add-id.cuh +3 -0
  125. data/ext/sources/ggml/src/ggml-cuda/binbcast.cu +330 -191
  126. data/ext/sources/ggml/src/ggml-cuda/binbcast.cuh +2 -0
  127. data/ext/sources/ggml/src/ggml-cuda/common.cuh +250 -63
  128. data/ext/sources/ggml/src/ggml-cuda/conv-transpose-1d.cu +1 -4
  129. data/ext/sources/ggml/src/ggml-cuda/conv2d.cu +166 -0
  130. data/ext/sources/ggml/src/ggml-cuda/conv2d.cuh +5 -0
  131. data/ext/sources/ggml/src/ggml-cuda/convert.cu +95 -22
  132. data/ext/sources/ggml/src/ggml-cuda/convert.cuh +15 -0
  133. data/ext/sources/ggml/src/ggml-cuda/cpy-utils.cuh +217 -0
  134. data/ext/sources/ggml/src/ggml-cuda/cpy.cu +64 -307
  135. data/ext/sources/ggml/src/ggml-cuda/cross-entropy-loss.cu +2 -14
  136. data/ext/sources/ggml/src/ggml-cuda/dequantize.cuh +14 -40
  137. data/ext/sources/ggml/src/ggml-cuda/fattn-common.cuh +498 -367
  138. data/ext/sources/ggml/src/ggml-cuda/fattn-mma-f16.cuh +137 -91
  139. data/ext/sources/ggml/src/ggml-cuda/fattn-tile.cu +755 -0
  140. data/ext/sources/ggml/src/ggml-cuda/fattn-tile.cuh +3 -0
  141. data/ext/sources/ggml/src/ggml-cuda/fattn-vec.cuh +593 -0
  142. data/ext/sources/ggml/src/ggml-cuda/fattn-wmma-f16.cu +86 -50
  143. data/ext/sources/ggml/src/ggml-cuda/fattn.cu +185 -198
  144. data/ext/sources/ggml/src/ggml-cuda/fattn.cuh +2 -0
  145. data/ext/sources/ggml/src/ggml-cuda/getrows.cu +50 -39
  146. data/ext/sources/ggml/src/ggml-cuda/ggml-cuda.cu +379 -107
  147. data/ext/sources/ggml/src/ggml-cuda/im2col.cu +196 -35
  148. data/ext/sources/ggml/src/ggml-cuda/im2col.cuh +1 -0
  149. data/ext/sources/ggml/src/ggml-cuda/mean.cu +56 -2
  150. data/ext/sources/ggml/src/ggml-cuda/mma.cuh +198 -45
  151. data/ext/sources/ggml/src/ggml-cuda/mmf.cu +123 -0
  152. data/ext/sources/ggml/src/ggml-cuda/mmf.cuh +496 -0
  153. data/ext/sources/ggml/src/ggml-cuda/mmq.cu +206 -57
  154. data/ext/sources/ggml/src/ggml-cuda/mmq.cuh +1262 -721
  155. data/ext/sources/ggml/src/ggml-cuda/{mmv.cu → mmvf.cu} +53 -53
  156. data/ext/sources/ggml/src/ggml-cuda/{mmv.cuh → mmvf.cuh} +3 -3
  157. data/ext/sources/ggml/src/ggml-cuda/mmvq.cu +64 -73
  158. data/ext/sources/ggml/src/ggml-cuda/norm.cu +284 -12
  159. data/ext/sources/ggml/src/ggml-cuda/norm.cuh +7 -0
  160. data/ext/sources/ggml/src/ggml-cuda/opt-step-sgd.cu +49 -0
  161. data/ext/sources/ggml/src/ggml-cuda/opt-step-sgd.cuh +5 -0
  162. data/ext/sources/ggml/src/ggml-cuda/pad.cu +46 -23
  163. data/ext/sources/ggml/src/ggml-cuda/pad_reflect_1d.cu +91 -0
  164. data/ext/sources/ggml/src/ggml-cuda/pad_reflect_1d.cuh +5 -0
  165. data/ext/sources/ggml/src/ggml-cuda/quantize.cu +12 -10
  166. data/ext/sources/ggml/src/ggml-cuda/reduce_rows.cuh +53 -0
  167. data/ext/sources/ggml/src/ggml-cuda/roll.cu +67 -0
  168. data/ext/sources/ggml/src/ggml-cuda/roll.cuh +5 -0
  169. data/ext/sources/ggml/src/ggml-cuda/rope.cu +21 -27
  170. data/ext/sources/ggml/src/ggml-cuda/scale.cu +14 -11
  171. data/ext/sources/ggml/src/ggml-cuda/set-rows.cu +276 -0
  172. data/ext/sources/ggml/src/ggml-cuda/set-rows.cuh +7 -0
  173. data/ext/sources/ggml/src/ggml-cuda/softcap.cu +34 -0
  174. data/ext/sources/ggml/src/ggml-cuda/softcap.cuh +5 -0
  175. data/ext/sources/ggml/src/ggml-cuda/softmax.cu +126 -59
  176. data/ext/sources/ggml/src/ggml-cuda/ssm-conv.cu +10 -2
  177. data/ext/sources/ggml/src/ggml-cuda/ssm-scan.cu +322 -100
  178. data/ext/sources/ggml/src/ggml-cuda/sum.cu +6 -10
  179. data/ext/sources/ggml/src/ggml-cuda/sumrows.cu +21 -4
  180. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu +7 -0
  181. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_0.cu +7 -0
  182. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_1.cu +7 -0
  183. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q5_0.cu +7 -0
  184. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q5_1.cu +7 -0
  185. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q8_0.cu +7 -0
  186. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-f16.cu +7 -0
  187. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu +7 -0
  188. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_1.cu +7 -0
  189. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q5_0.cu +7 -0
  190. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q5_1.cu +7 -0
  191. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q8_0.cu +7 -0
  192. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-f16.cu +7 -0
  193. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q4_0.cu +7 -0
  194. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q4_1.cu +7 -0
  195. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q5_0.cu +7 -0
  196. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q5_1.cu +7 -0
  197. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q8_0.cu +7 -0
  198. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-f16.cu +7 -0
  199. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q4_0.cu +7 -0
  200. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q4_1.cu +7 -0
  201. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q5_0.cu +7 -0
  202. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q5_1.cu +7 -0
  203. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q8_0.cu +7 -0
  204. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-f16.cu +7 -0
  205. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q4_0.cu +7 -0
  206. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q4_1.cu +7 -0
  207. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q5_0.cu +7 -0
  208. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q5_1.cu +7 -0
  209. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q8_0.cu +7 -0
  210. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-f16.cu +7 -0
  211. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q4_0.cu +7 -0
  212. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q4_1.cu +7 -0
  213. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q5_0.cu +7 -0
  214. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q5_1.cu +7 -0
  215. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu +7 -0
  216. data/ext/sources/ggml/src/ggml-cuda/template-instances/generate_cu_files.py +21 -18
  217. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu +5 -0
  218. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu +5 -0
  219. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu +5 -0
  220. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu +5 -0
  221. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu +5 -0
  222. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu +5 -0
  223. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu +5 -0
  224. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu +5 -0
  225. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu +5 -0
  226. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu +5 -0
  227. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu +5 -0
  228. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu +5 -0
  229. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu +5 -0
  230. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu +5 -0
  231. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu +5 -0
  232. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu +5 -0
  233. data/ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu +5 -0
  234. data/ext/sources/ggml/src/ggml-cuda/topk-moe.cu +259 -0
  235. data/ext/sources/ggml/src/ggml-cuda/topk-moe.cuh +14 -0
  236. data/ext/sources/ggml/src/ggml-cuda/tsembd.cu +3 -3
  237. data/ext/sources/ggml/src/ggml-cuda/unary.cu +90 -0
  238. data/ext/sources/ggml/src/ggml-cuda/unary.cuh +8 -0
  239. data/ext/sources/ggml/src/ggml-cuda/upscale.cu +92 -6
  240. data/ext/sources/ggml/src/ggml-cuda/vecdotq.cuh +110 -22
  241. data/ext/sources/ggml/src/ggml-cuda/vendors/cuda.h +4 -0
  242. data/ext/sources/ggml/src/ggml-cuda/vendors/hip.h +58 -36
  243. data/ext/sources/ggml/src/ggml-cuda/vendors/musa.h +4 -3
  244. data/ext/sources/ggml/src/ggml-hip/CMakeLists.txt +10 -2
  245. data/ext/sources/ggml/src/ggml-impl.h +119 -9
  246. data/ext/sources/ggml/src/ggml-metal/CMakeLists.txt +10 -7
  247. data/ext/sources/ggml/src/ggml-metal/ggml-metal-common.cpp +446 -0
  248. data/ext/sources/ggml/src/ggml-metal/ggml-metal-common.h +52 -0
  249. data/ext/sources/ggml/src/ggml-metal/ggml-metal-context.h +33 -0
  250. data/ext/sources/ggml/src/ggml-metal/ggml-metal-context.m +600 -0
  251. data/ext/sources/ggml/src/ggml-metal/ggml-metal-device.cpp +1376 -0
  252. data/ext/sources/ggml/src/ggml-metal/ggml-metal-device.h +226 -0
  253. data/ext/sources/ggml/src/ggml-metal/ggml-metal-device.m +1308 -0
  254. data/ext/sources/ggml/src/ggml-metal/ggml-metal-impl.h +136 -63
  255. data/ext/sources/ggml/src/ggml-metal/ggml-metal-ops.cpp +3158 -0
  256. data/ext/sources/ggml/src/ggml-metal/ggml-metal-ops.h +82 -0
  257. data/ext/sources/ggml/src/ggml-metal/ggml-metal.cpp +718 -0
  258. data/ext/sources/ggml/src/ggml-metal/ggml-metal.metal +2854 -1503
  259. data/ext/sources/ggml/src/ggml-musa/CMakeLists.txt +18 -8
  260. data/ext/sources/ggml/src/ggml-opencl/CMakeLists.txt +18 -0
  261. data/ext/sources/ggml/src/ggml-opencl/ggml-opencl.cpp +2510 -242
  262. data/ext/sources/ggml/src/ggml-opencl/kernels/add.cl +107 -0
  263. data/ext/sources/ggml/src/ggml-opencl/kernels/add_id.cl +42 -0
  264. data/ext/sources/ggml/src/ggml-opencl/kernels/conv2d.cl +185 -0
  265. data/ext/sources/ggml/src/ggml-opencl/kernels/conv2d_f16_f32.cl +176 -0
  266. data/ext/sources/ggml/src/ggml-opencl/kernels/cvt.cl +84 -0
  267. data/ext/sources/ggml/src/ggml-opencl/kernels/div.cl +66 -0
  268. data/ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f16.cl +370 -0
  269. data/ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32.cl +370 -0
  270. data/ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32_f16.cl +373 -0
  271. data/ext/sources/ggml/src/ggml-opencl/kernels/gelu.cl +27 -0
  272. data/ext/sources/ggml/src/ggml-opencl/kernels/glu.cl +177 -0
  273. data/ext/sources/ggml/src/ggml-opencl/kernels/group_norm.cl +49 -0
  274. data/ext/sources/ggml/src/ggml-opencl/kernels/im2col_f16.cl +1 -1
  275. data/ext/sources/ggml/src/ggml-opencl/kernels/im2col_f32.cl +1 -1
  276. data/ext/sources/ggml/src/ggml-opencl/kernels/mul.cl +73 -0
  277. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mat_f16_f32.cl +130 -0
  278. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f16_f32_l4_lm.cl +132 -0
  279. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f32_f32_l4_lm.cl +133 -0
  280. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_mxfp4_f32.cl +189 -0
  281. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_mxfp4_f32_flat.cl +176 -0
  282. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_q8_0_f32.cl +140 -0
  283. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_q8_0_f32_flat.cl +222 -0
  284. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_mxfp4_f32.cl +144 -0
  285. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_mxfp4_f32_flat.cl +167 -0
  286. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q8_0_f32.cl +125 -0
  287. data/ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q8_0_f32_flat.cl +202 -0
  288. data/ext/sources/ggml/src/ggml-opencl/kernels/norm.cl +80 -0
  289. data/ext/sources/ggml/src/ggml-opencl/kernels/rms_norm.cl +79 -0
  290. data/ext/sources/ggml/src/ggml-opencl/kernels/scale.cl +3 -2
  291. data/ext/sources/ggml/src/ggml-opencl/kernels/set_rows.cl +189 -0
  292. data/ext/sources/ggml/src/ggml-opencl/kernels/softmax_4_f16.cl +34 -13
  293. data/ext/sources/ggml/src/ggml-opencl/kernels/softmax_4_f32.cl +34 -13
  294. data/ext/sources/ggml/src/ggml-opencl/kernels/softmax_f16.cl +34 -13
  295. data/ext/sources/ggml/src/ggml-opencl/kernels/softmax_f32.cl +34 -13
  296. data/ext/sources/ggml/src/ggml-opencl/kernels/sub.cl +66 -0
  297. data/ext/sources/ggml/src/ggml-opencl/kernels/transpose.cl +20 -0
  298. data/ext/sources/ggml/src/ggml-opencl/kernels/tsembd.cl +2 -2
  299. data/ext/sources/ggml/src/ggml-opencl/kernels/upscale.cl +2 -3
  300. data/ext/sources/ggml/src/ggml-opt.cpp +97 -41
  301. data/ext/sources/ggml/src/ggml-quants.c +111 -16
  302. data/ext/sources/ggml/src/ggml-quants.h +6 -0
  303. data/ext/sources/ggml/src/ggml-rpc/ggml-rpc.cpp +67 -47
  304. data/ext/sources/ggml/src/ggml-sycl/backend.hpp +2 -0
  305. data/ext/sources/ggml/src/ggml-sycl/binbcast.cpp +15 -5
  306. data/ext/sources/ggml/src/ggml-sycl/binbcast.hpp +6 -0
  307. data/ext/sources/ggml/src/ggml-sycl/concat.cpp +25 -16
  308. data/ext/sources/ggml/src/ggml-sycl/conv.cpp +10 -4
  309. data/ext/sources/ggml/src/ggml-sycl/convert.cpp +166 -99
  310. data/ext/sources/ggml/src/ggml-sycl/cpy.cpp +72 -306
  311. data/ext/sources/ggml/src/ggml-sycl/cpy.hpp +213 -1
  312. data/ext/sources/ggml/src/ggml-sycl/dmmv.cpp +67 -49
  313. data/ext/sources/ggml/src/ggml-sycl/dpct/helper.hpp +1 -31
  314. data/ext/sources/ggml/src/ggml-sycl/element_wise.cpp +79 -29
  315. data/ext/sources/ggml/src/ggml-sycl/element_wise.hpp +2 -0
  316. data/ext/sources/ggml/src/ggml-sycl/gemm.hpp +14 -26
  317. data/ext/sources/ggml/src/ggml-sycl/getrows.cpp +9 -6
  318. data/ext/sources/ggml/src/ggml-sycl/ggml-sycl.cpp +328 -323
  319. data/ext/sources/ggml/src/ggml-sycl/gla.cpp +2 -2
  320. data/ext/sources/ggml/src/ggml-sycl/im2col.cpp +2 -2
  321. data/ext/sources/ggml/src/ggml-sycl/mmq.cpp +80 -60
  322. data/ext/sources/ggml/src/ggml-sycl/mmvq.cpp +201 -132
  323. data/ext/sources/ggml/src/ggml-sycl/norm.cpp +74 -55
  324. data/ext/sources/ggml/src/ggml-sycl/quantize.hpp +133 -0
  325. data/ext/sources/ggml/src/ggml-sycl/quants.hpp +8 -9
  326. data/ext/sources/ggml/src/ggml-sycl/rope.cpp +35 -42
  327. data/ext/sources/ggml/src/ggml-sycl/set_rows.cpp +234 -0
  328. data/ext/sources/ggml/src/ggml-sycl/set_rows.hpp +8 -0
  329. data/ext/sources/ggml/src/ggml-sycl/softmax.cpp +3 -3
  330. data/ext/sources/ggml/src/ggml-sycl/tsembd.cpp +12 -6
  331. data/ext/sources/ggml/src/ggml-sycl/vecdotq.hpp +2 -6
  332. data/ext/sources/ggml/src/ggml-sycl/wkv.cpp +16 -12
  333. data/ext/sources/ggml/src/ggml-vulkan/ggml-vulkan.cpp +3492 -883
  334. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add.comp +41 -1
  335. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add_id.comp +42 -0
  336. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argmax.comp +13 -4
  337. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argsort.comp +39 -29
  338. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv2d_mm.comp +349 -0
  339. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_from_quant.comp +2 -2
  340. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_to_quant.comp +66 -12
  341. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs.comp +154 -0
  342. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs_cm2.comp +21 -0
  343. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq2_s.comp +1 -1
  344. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq2_xxs.comp +2 -1
  345. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq3_s.comp +6 -5
  346. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq3_xxs.comp +4 -2
  347. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_mxfp4.comp +32 -0
  348. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q2_k.comp +1 -1
  349. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q3_k.comp +1 -1
  350. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q4_k.comp +1 -1
  351. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q5_k.comp +1 -1
  352. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q6_k.comp +1 -1
  353. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/exp.comp +21 -0
  354. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp +69 -24
  355. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.comp +60 -20
  356. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp +98 -42
  357. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm2.comp +64 -27
  358. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_split_k_reduce.comp +74 -13
  359. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_erf.comp +27 -0
  360. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_quick.comp +11 -0
  361. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_erf.comp +39 -0
  362. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_binary_head.comp +4 -17
  363. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/get_rows.comp +19 -10
  364. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/get_rows_quant.comp +25 -15
  365. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_head.comp +4 -0
  366. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardsigmoid.comp +22 -0
  367. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardswish.comp +22 -0
  368. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col.comp +18 -14
  369. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col_3d.comp +126 -0
  370. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_base.comp +65 -1
  371. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_nc.comp +11 -7
  372. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq.comp +140 -0
  373. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp +144 -531
  374. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_cm2.comp +206 -38
  375. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm_funcs.comp +556 -0
  376. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq.comp +12 -5
  377. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_funcs.comp +15 -9
  378. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/multi_add.comp +111 -0
  379. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_sgd.comp +22 -0
  380. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/pad.comp +24 -3
  381. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/quantize_q8_1.comp +53 -3
  382. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm.comp +55 -11
  383. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_partials.comp +65 -0
  384. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/roll.comp +46 -0
  385. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_head.comp +1 -4
  386. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_multi.comp +7 -9
  387. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_neox.comp +7 -9
  388. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_norm.comp +7 -9
  389. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rte.comp +5 -0
  390. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/scale.comp +1 -1
  391. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max.comp +29 -7
  392. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_back.comp +4 -0
  393. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sqrt.comp +17 -0
  394. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sum_rows.comp +38 -5
  395. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu_oai.comp +14 -0
  396. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/timestep_embedding.comp +4 -3
  397. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/types.comp +101 -9
  398. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/upscale.comp +69 -5
  399. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/utils.comp +25 -0
  400. data/ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp +335 -77
  401. data/ext/sources/ggml/src/ggml-webgpu/CMakeLists.txt +54 -0
  402. data/ext/sources/ggml/src/ggml-webgpu/ggml-webgpu.cpp +1558 -0
  403. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/add.tmpl.wgsl +44 -0
  404. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/add_in_place.tmpl.wgsl +41 -0
  405. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/binary_head.tmpl +45 -0
  406. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl +930 -0
  407. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/cpy.wgsl +60 -0
  408. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/embed_wgsl.py +124 -0
  409. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/get_rows.tmpl.wgsl +874 -0
  410. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/memset.wgsl +40 -0
  411. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul.tmpl.wgsl +44 -0
  412. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_in_place.tmpl.wgsl +41 -0
  413. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat.tmpl.wgsl +907 -0
  414. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rms_norm.wgsl +57 -0
  415. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/rms_norm_in_place.wgsl +48 -0
  416. data/ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/set_rows.wgsl +81 -0
  417. data/ext/sources/ggml/src/ggml-zdnn/CMakeLists.txt +36 -0
  418. data/ext/sources/ggml/src/ggml-zdnn/common.hpp +59 -0
  419. data/ext/sources/ggml/src/ggml-zdnn/ggml-zdnn.cpp +628 -0
  420. data/ext/sources/ggml/src/ggml-zdnn/mmf.cpp +80 -0
  421. data/ext/sources/ggml/src/ggml-zdnn/mmf.hpp +12 -0
  422. data/ext/sources/ggml/src/ggml-zdnn/utils.cpp +79 -0
  423. data/ext/sources/ggml/src/ggml-zdnn/utils.hpp +19 -0
  424. data/ext/sources/ggml/src/ggml.c +478 -98
  425. data/ext/sources/ggml/src/gguf.cpp +8 -1
  426. data/ext/sources/src/whisper.cpp +23 -46
  427. data/ext/sources/tests/CMakeLists.txt +8 -1
  428. data/ext/sources/tests/test-vad-full.cpp +3 -3
  429. data/ext/sources/tests/test-vad.cpp +2 -2
  430. data/lib/whisper/model/uri.rb +1 -1
  431. data/sig/whisper.rbs +7 -0
  432. data/test/test_params.rb +8 -0
  433. data/test/test_whisper.rb +1 -1
  434. data/whispercpp.gemspec +1 -1
  435. metadata +164 -157
  436. data/ext/sources/examples/talk-llama/llama-kv-cache-unified-iswa.cpp +0 -279
  437. data/ext/sources/examples/talk-llama/llama-kv-cache-unified.cpp +0 -1841
  438. data/ext/sources/examples/talk-llama/llama-kv-cache-unified.h +0 -303
  439. data/ext/sources/ggml/include/ggml-kompute.h +0 -50
  440. data/ext/sources/ggml/src/ggml-amx/CMakeLists.txt +0 -107
  441. data/ext/sources/ggml/src/ggml-amx/common.h +0 -94
  442. data/ext/sources/ggml/src/ggml-amx/ggml-amx.cpp +0 -446
  443. data/ext/sources/ggml/src/ggml-amx/mmq.cpp +0 -2510
  444. data/ext/sources/ggml/src/ggml-amx/mmq.h +0 -17
  445. data/ext/sources/ggml/src/ggml-cann/kernels/CMakeLists.txt +0 -30
  446. data/ext/sources/ggml/src/ggml-cann/kernels/ascendc_kernels.h +0 -19
  447. data/ext/sources/ggml/src/ggml-cann/kernels/dup.cpp +0 -234
  448. data/ext/sources/ggml/src/ggml-cann/kernels/get_row_f16.cpp +0 -197
  449. data/ext/sources/ggml/src/ggml-cann/kernels/get_row_f32.cpp +0 -190
  450. data/ext/sources/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp +0 -204
  451. data/ext/sources/ggml/src/ggml-cann/kernels/get_row_q8_0.cpp +0 -191
  452. data/ext/sources/ggml/src/ggml-cann/kernels/quantize_f16_q8_0.cpp +0 -218
  453. data/ext/sources/ggml/src/ggml-cann/kernels/quantize_f32_q8_0.cpp +0 -216
  454. data/ext/sources/ggml/src/ggml-cann/kernels/quantize_float_to_q4_0.cpp +0 -295
  455. data/ext/sources/ggml/src/ggml-cuda/fattn-tile-f16.cu +0 -357
  456. data/ext/sources/ggml/src/ggml-cuda/fattn-tile-f16.cuh +0 -3
  457. data/ext/sources/ggml/src/ggml-cuda/fattn-tile-f32.cu +0 -365
  458. data/ext/sources/ggml/src/ggml-cuda/fattn-tile-f32.cuh +0 -3
  459. data/ext/sources/ggml/src/ggml-cuda/fattn-vec-f16.cuh +0 -482
  460. data/ext/sources/ggml/src/ggml-cuda/fattn-vec-f32.cuh +0 -472
  461. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu +0 -5
  462. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q4_0.cu +0 -5
  463. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q4_1.cu +0 -5
  464. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q5_0.cu +0 -5
  465. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q5_1.cu +0 -5
  466. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q8_0.cu +0 -5
  467. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-f16.cu +0 -5
  468. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu +0 -5
  469. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_1.cu +0 -5
  470. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q5_0.cu +0 -5
  471. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q5_1.cu +0 -5
  472. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q8_0.cu +0 -5
  473. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-f16.cu +0 -5
  474. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q4_0.cu +0 -5
  475. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q4_1.cu +0 -5
  476. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q5_0.cu +0 -5
  477. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q5_1.cu +0 -5
  478. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q8_0.cu +0 -5
  479. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-f16.cu +0 -5
  480. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q4_0.cu +0 -5
  481. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q4_1.cu +0 -5
  482. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q5_0.cu +0 -5
  483. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q5_1.cu +0 -5
  484. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q8_0.cu +0 -5
  485. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-f16.cu +0 -5
  486. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q4_0.cu +0 -5
  487. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q4_1.cu +0 -5
  488. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q5_0.cu +0 -5
  489. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q5_1.cu +0 -5
  490. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q8_0.cu +0 -5
  491. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-f16.cu +0 -5
  492. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q4_0.cu +0 -5
  493. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q4_1.cu +0 -5
  494. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q5_0.cu +0 -5
  495. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q5_1.cu +0 -5
  496. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu +0 -5
  497. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu +0 -5
  498. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu +0 -5
  499. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q4_0.cu +0 -5
  500. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q4_1.cu +0 -5
  501. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q5_0.cu +0 -5
  502. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q5_1.cu +0 -5
  503. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q8_0.cu +0 -5
  504. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu +0 -5
  505. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q4_0.cu +0 -5
  506. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q4_1.cu +0 -5
  507. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q5_0.cu +0 -5
  508. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q5_1.cu +0 -5
  509. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q8_0.cu +0 -5
  510. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-f16.cu +0 -5
  511. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu +0 -5
  512. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_1.cu +0 -5
  513. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q5_0.cu +0 -5
  514. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q5_1.cu +0 -5
  515. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q8_0.cu +0 -5
  516. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-f16.cu +0 -5
  517. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q4_0.cu +0 -5
  518. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q4_1.cu +0 -5
  519. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q5_0.cu +0 -5
  520. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q5_1.cu +0 -5
  521. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q8_0.cu +0 -5
  522. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-f16.cu +0 -5
  523. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q4_0.cu +0 -5
  524. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q4_1.cu +0 -5
  525. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q5_0.cu +0 -5
  526. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q5_1.cu +0 -5
  527. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q8_0.cu +0 -5
  528. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-f16.cu +0 -5
  529. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q4_0.cu +0 -5
  530. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q4_1.cu +0 -5
  531. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q5_0.cu +0 -5
  532. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q5_1.cu +0 -5
  533. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q8_0.cu +0 -5
  534. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-f16.cu +0 -5
  535. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q4_0.cu +0 -5
  536. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q4_1.cu +0 -5
  537. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q5_0.cu +0 -5
  538. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q5_1.cu +0 -5
  539. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu +0 -5
  540. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu +0 -5
  541. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu +0 -5
  542. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q4_0.cu +0 -5
  543. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q4_1.cu +0 -5
  544. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q5_0.cu +0 -5
  545. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q5_1.cu +0 -5
  546. data/ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q8_0.cu +0 -5
  547. data/ext/sources/ggml/src/ggml-kompute/CMakeLists.txt +0 -166
  548. data/ext/sources/ggml/src/ggml-kompute/ggml-kompute.cpp +0 -2251
  549. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/common.comp +0 -112
  550. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_add.comp +0 -58
  551. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_addrow.comp +0 -25
  552. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f16_f16.comp +0 -52
  553. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f16_f32.comp +0 -52
  554. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f32_f16.comp +0 -52
  555. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f32_f32.comp +0 -52
  556. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_diagmask.comp +0 -30
  557. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_gelu.comp +0 -22
  558. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows.comp +0 -17
  559. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_f16.comp +0 -31
  560. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_f32.comp +0 -31
  561. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q4_0.comp +0 -38
  562. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q4_1.comp +0 -39
  563. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q6_k.comp +0 -44
  564. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul.comp +0 -52
  565. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_f16.comp +0 -69
  566. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_mat_f32.comp +0 -51
  567. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_0.comp +0 -33
  568. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_1.comp +0 -35
  569. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_k.comp +0 -140
  570. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q6_k.comp +0 -106
  571. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q8_0.comp +0 -73
  572. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n.comp +0 -52
  573. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n_pre.comp +0 -28
  574. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_norm.comp +0 -84
  575. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_relu.comp +0 -21
  576. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rmsnorm.comp +0 -53
  577. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f16.comp +0 -52
  578. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f32.comp +0 -52
  579. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f16.comp +0 -52
  580. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f32.comp +0 -52
  581. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_scale.comp +0 -19
  582. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_scale_8.comp +0 -23
  583. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_silu.comp +0 -22
  584. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_softmax.comp +0 -72
  585. data/ext/sources/ggml/src/ggml-kompute/kompute-shaders/rope_common.comp +0 -71
  586. data/ext/sources/ggml/src/ggml-metal/ggml-metal.m +0 -6280
metadata CHANGED
@@ -1,7 +1,7 @@
1
1
  --- !ruby/object:Gem::Specification
2
2
  name: whispercpp
3
3
  version: !ruby/object:Gem::Version
4
- version: 1.3.3
4
+ version: 1.3.4
5
5
  platform: ruby
6
6
  authors:
7
7
  - Georgi Gerganov
@@ -9,7 +9,7 @@ authors:
9
9
  autorequire:
10
10
  bindir: bin
11
11
  cert_chain: []
12
- date: 2025-07-12 00:00:00.000000000 Z
12
+ date: 2025-10-08 00:00:00.000000000 Z
13
13
  dependencies: []
14
14
  description: High-performance inference of OpenAI's Whisper automatic speech recognition
15
15
  (ASR) model via Ruby
@@ -138,10 +138,9 @@ files:
138
138
  - ext/sources/examples/talk-llama/llama-impl.h
139
139
  - ext/sources/examples/talk-llama/llama-io.cpp
140
140
  - ext/sources/examples/talk-llama/llama-io.h
141
- - ext/sources/examples/talk-llama/llama-kv-cache-unified-iswa.cpp
142
- - ext/sources/examples/talk-llama/llama-kv-cache-unified-iswa.h
143
- - ext/sources/examples/talk-llama/llama-kv-cache-unified.cpp
144
- - ext/sources/examples/talk-llama/llama-kv-cache-unified.h
141
+ - ext/sources/examples/talk-llama/llama-kv-cache-iswa.cpp
142
+ - ext/sources/examples/talk-llama/llama-kv-cache-iswa.h
143
+ - ext/sources/examples/talk-llama/llama-kv-cache.cpp
145
144
  - ext/sources/examples/talk-llama/llama-kv-cache.h
146
145
  - ext/sources/examples/talk-llama/llama-kv-cells.h
147
146
  - ext/sources/examples/talk-llama/llama-memory-hybrid.cpp
@@ -201,22 +200,18 @@ files:
201
200
  - ext/sources/ggml/include/ggml-cpp.h
202
201
  - ext/sources/ggml/include/ggml-cpu.h
203
202
  - ext/sources/ggml/include/ggml-cuda.h
204
- - ext/sources/ggml/include/ggml-kompute.h
205
203
  - ext/sources/ggml/include/ggml-metal.h
206
204
  - ext/sources/ggml/include/ggml-opencl.h
207
205
  - ext/sources/ggml/include/ggml-opt.h
208
206
  - ext/sources/ggml/include/ggml-rpc.h
209
207
  - ext/sources/ggml/include/ggml-sycl.h
210
208
  - ext/sources/ggml/include/ggml-vulkan.h
209
+ - ext/sources/ggml/include/ggml-webgpu.h
210
+ - ext/sources/ggml/include/ggml-zdnn.h
211
211
  - ext/sources/ggml/include/ggml.h
212
212
  - ext/sources/ggml/include/gguf.h
213
213
  - ext/sources/ggml/src/CMakeLists.txt
214
214
  - ext/sources/ggml/src/ggml-alloc.c
215
- - ext/sources/ggml/src/ggml-amx/CMakeLists.txt
216
- - ext/sources/ggml/src/ggml-amx/common.h
217
- - ext/sources/ggml/src/ggml-amx/ggml-amx.cpp
218
- - ext/sources/ggml/src/ggml-amx/mmq.cpp
219
- - ext/sources/ggml/src/ggml-amx/mmq.h
220
215
  - ext/sources/ggml/src/ggml-backend-impl.h
221
216
  - ext/sources/ggml/src/ggml-backend-reg.cpp
222
217
  - ext/sources/ggml/src/ggml-backend.cpp
@@ -230,16 +225,6 @@ files:
230
225
  - ext/sources/ggml/src/ggml-cann/aclnn_ops.h
231
226
  - ext/sources/ggml/src/ggml-cann/common.h
232
227
  - ext/sources/ggml/src/ggml-cann/ggml-cann.cpp
233
- - ext/sources/ggml/src/ggml-cann/kernels/CMakeLists.txt
234
- - ext/sources/ggml/src/ggml-cann/kernels/ascendc_kernels.h
235
- - ext/sources/ggml/src/ggml-cann/kernels/dup.cpp
236
- - ext/sources/ggml/src/ggml-cann/kernels/get_row_f16.cpp
237
- - ext/sources/ggml/src/ggml-cann/kernels/get_row_f32.cpp
238
- - ext/sources/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp
239
- - ext/sources/ggml/src/ggml-cann/kernels/get_row_q8_0.cpp
240
- - ext/sources/ggml/src/ggml-cann/kernels/quantize_f16_q8_0.cpp
241
- - ext/sources/ggml/src/ggml-cann/kernels/quantize_f32_q8_0.cpp
242
- - ext/sources/ggml/src/ggml-cann/kernels/quantize_float_to_q4_0.cpp
243
228
  - ext/sources/ggml/src/ggml-common.h
244
229
  - ext/sources/ggml/src/ggml-cpu/CMakeLists.txt
245
230
  - ext/sources/ggml/src/ggml-cpu/amx/amx.cpp
@@ -283,6 +268,10 @@ files:
283
268
  - ext/sources/ggml/src/ggml-cpu/repack.cpp
284
269
  - ext/sources/ggml/src/ggml-cpu/repack.h
285
270
  - ext/sources/ggml/src/ggml-cpu/simd-mappings.h
271
+ - ext/sources/ggml/src/ggml-cpu/spacemit/ime.cpp
272
+ - ext/sources/ggml/src/ggml-cpu/spacemit/ime.h
273
+ - ext/sources/ggml/src/ggml-cpu/spacemit/ime1_kernels.cpp
274
+ - ext/sources/ggml/src/ggml-cpu/spacemit/ime_kernels.h
286
275
  - ext/sources/ggml/src/ggml-cpu/traits.cpp
287
276
  - ext/sources/ggml/src/ggml-cpu/traits.h
288
277
  - ext/sources/ggml/src/ggml-cpu/unary-ops.cpp
@@ -292,6 +281,8 @@ files:
292
281
  - ext/sources/ggml/src/ggml-cuda/CMakeLists.txt
293
282
  - ext/sources/ggml/src/ggml-cuda/acc.cu
294
283
  - ext/sources/ggml/src/ggml-cuda/acc.cuh
284
+ - ext/sources/ggml/src/ggml-cuda/add-id.cu
285
+ - ext/sources/ggml/src/ggml-cuda/add-id.cuh
295
286
  - ext/sources/ggml/src/ggml-cuda/arange.cu
296
287
  - ext/sources/ggml/src/ggml-cuda/arange.cuh
297
288
  - ext/sources/ggml/src/ggml-cuda/argmax.cu
@@ -311,11 +302,14 @@ files:
311
302
  - ext/sources/ggml/src/ggml-cuda/conv2d-dw.cuh
312
303
  - ext/sources/ggml/src/ggml-cuda/conv2d-transpose.cu
313
304
  - ext/sources/ggml/src/ggml-cuda/conv2d-transpose.cuh
305
+ - ext/sources/ggml/src/ggml-cuda/conv2d.cu
306
+ - ext/sources/ggml/src/ggml-cuda/conv2d.cuh
314
307
  - ext/sources/ggml/src/ggml-cuda/convert.cu
315
308
  - ext/sources/ggml/src/ggml-cuda/convert.cuh
316
309
  - ext/sources/ggml/src/ggml-cuda/count-equal.cu
317
310
  - ext/sources/ggml/src/ggml-cuda/count-equal.cuh
318
311
  - ext/sources/ggml/src/ggml-cuda/cp-async.cuh
312
+ - ext/sources/ggml/src/ggml-cuda/cpy-utils.cuh
319
313
  - ext/sources/ggml/src/ggml-cuda/cpy.cu
320
314
  - ext/sources/ggml/src/ggml-cuda/cpy.cuh
321
315
  - ext/sources/ggml/src/ggml-cuda/cross-entropy-loss.cu
@@ -325,12 +319,9 @@ files:
325
319
  - ext/sources/ggml/src/ggml-cuda/diagmask.cuh
326
320
  - ext/sources/ggml/src/ggml-cuda/fattn-common.cuh
327
321
  - ext/sources/ggml/src/ggml-cuda/fattn-mma-f16.cuh
328
- - ext/sources/ggml/src/ggml-cuda/fattn-tile-f16.cu
329
- - ext/sources/ggml/src/ggml-cuda/fattn-tile-f16.cuh
330
- - ext/sources/ggml/src/ggml-cuda/fattn-tile-f32.cu
331
- - ext/sources/ggml/src/ggml-cuda/fattn-tile-f32.cuh
332
- - ext/sources/ggml/src/ggml-cuda/fattn-vec-f16.cuh
333
- - ext/sources/ggml/src/ggml-cuda/fattn-vec-f32.cuh
322
+ - ext/sources/ggml/src/ggml-cuda/fattn-tile.cu
323
+ - ext/sources/ggml/src/ggml-cuda/fattn-tile.cuh
324
+ - ext/sources/ggml/src/ggml-cuda/fattn-vec.cuh
334
325
  - ext/sources/ggml/src/ggml-cuda/fattn-wmma-f16.cu
335
326
  - ext/sources/ggml/src/ggml-cuda/fattn-wmma-f16.cuh
336
327
  - ext/sources/ggml/src/ggml-cuda/fattn.cu
@@ -345,28 +336,41 @@ files:
345
336
  - ext/sources/ggml/src/ggml-cuda/mean.cu
346
337
  - ext/sources/ggml/src/ggml-cuda/mean.cuh
347
338
  - ext/sources/ggml/src/ggml-cuda/mma.cuh
339
+ - ext/sources/ggml/src/ggml-cuda/mmf.cu
340
+ - ext/sources/ggml/src/ggml-cuda/mmf.cuh
348
341
  - ext/sources/ggml/src/ggml-cuda/mmq.cu
349
342
  - ext/sources/ggml/src/ggml-cuda/mmq.cuh
350
- - ext/sources/ggml/src/ggml-cuda/mmv.cu
351
- - ext/sources/ggml/src/ggml-cuda/mmv.cuh
343
+ - ext/sources/ggml/src/ggml-cuda/mmvf.cu
344
+ - ext/sources/ggml/src/ggml-cuda/mmvf.cuh
352
345
  - ext/sources/ggml/src/ggml-cuda/mmvq.cu
353
346
  - ext/sources/ggml/src/ggml-cuda/mmvq.cuh
354
347
  - ext/sources/ggml/src/ggml-cuda/norm.cu
355
348
  - ext/sources/ggml/src/ggml-cuda/norm.cuh
356
349
  - ext/sources/ggml/src/ggml-cuda/opt-step-adamw.cu
357
350
  - ext/sources/ggml/src/ggml-cuda/opt-step-adamw.cuh
351
+ - ext/sources/ggml/src/ggml-cuda/opt-step-sgd.cu
352
+ - ext/sources/ggml/src/ggml-cuda/opt-step-sgd.cuh
358
353
  - ext/sources/ggml/src/ggml-cuda/out-prod.cu
359
354
  - ext/sources/ggml/src/ggml-cuda/out-prod.cuh
360
355
  - ext/sources/ggml/src/ggml-cuda/pad.cu
361
356
  - ext/sources/ggml/src/ggml-cuda/pad.cuh
357
+ - ext/sources/ggml/src/ggml-cuda/pad_reflect_1d.cu
358
+ - ext/sources/ggml/src/ggml-cuda/pad_reflect_1d.cuh
362
359
  - ext/sources/ggml/src/ggml-cuda/pool2d.cu
363
360
  - ext/sources/ggml/src/ggml-cuda/pool2d.cuh
364
361
  - ext/sources/ggml/src/ggml-cuda/quantize.cu
365
362
  - ext/sources/ggml/src/ggml-cuda/quantize.cuh
363
+ - ext/sources/ggml/src/ggml-cuda/reduce_rows.cuh
364
+ - ext/sources/ggml/src/ggml-cuda/roll.cu
365
+ - ext/sources/ggml/src/ggml-cuda/roll.cuh
366
366
  - ext/sources/ggml/src/ggml-cuda/rope.cu
367
367
  - ext/sources/ggml/src/ggml-cuda/rope.cuh
368
368
  - ext/sources/ggml/src/ggml-cuda/scale.cu
369
369
  - ext/sources/ggml/src/ggml-cuda/scale.cuh
370
+ - ext/sources/ggml/src/ggml-cuda/set-rows.cu
371
+ - ext/sources/ggml/src/ggml-cuda/set-rows.cuh
372
+ - ext/sources/ggml/src/ggml-cuda/softcap.cu
373
+ - ext/sources/ggml/src/ggml-cuda/softcap.cuh
370
374
  - ext/sources/ggml/src/ggml-cuda/softmax.cu
371
375
  - ext/sources/ggml/src/ggml-cuda/softmax.cuh
372
376
  - ext/sources/ggml/src/ggml-cuda/ssm-conv.cu
@@ -396,93 +400,59 @@ files:
396
400
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_2.cu
397
401
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_4.cu
398
402
  - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-mma-f16-instance-ncols1_8-ncols2_8.cu
399
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-f16.cu
400
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q4_0.cu
401
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q4_1.cu
402
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q5_0.cu
403
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q5_1.cu
404
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-f16-q8_0.cu
405
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-f16.cu
406
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_0.cu
407
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q4_1.cu
408
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q5_0.cu
409
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q5_1.cu
410
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_0-q8_0.cu
411
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-f16.cu
412
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q4_0.cu
413
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q4_1.cu
414
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q5_0.cu
415
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q5_1.cu
416
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q4_1-q8_0.cu
417
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-f16.cu
418
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q4_0.cu
419
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q4_1.cu
420
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q5_0.cu
421
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q5_1.cu
422
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_0-q8_0.cu
423
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-f16.cu
424
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q4_0.cu
425
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q4_1.cu
426
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q5_0.cu
427
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q5_1.cu
428
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q5_1-q8_0.cu
429
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-f16.cu
430
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q4_0.cu
431
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q4_1.cu
432
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q5_0.cu
433
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q5_1.cu
434
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs128-q8_0-q8_0.cu
435
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs256-f16-f16.cu
436
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-f16.cu
437
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q4_0.cu
438
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q4_1.cu
439
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q5_0.cu
440
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q5_1.cu
441
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f16-instance-hs64-f16-q8_0.cu
442
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-f16.cu
443
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q4_0.cu
444
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q4_1.cu
445
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q5_0.cu
446
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q5_1.cu
447
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-f16-q8_0.cu
448
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-f16.cu
449
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_0.cu
450
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q4_1.cu
451
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q5_0.cu
452
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q5_1.cu
453
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_0-q8_0.cu
454
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-f16.cu
455
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q4_0.cu
456
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q4_1.cu
457
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q5_0.cu
458
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q5_1.cu
459
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q4_1-q8_0.cu
460
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-f16.cu
461
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q4_0.cu
462
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q4_1.cu
463
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q5_0.cu
464
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q5_1.cu
465
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_0-q8_0.cu
466
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-f16.cu
467
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q4_0.cu
468
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q4_1.cu
469
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q5_0.cu
470
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q5_1.cu
471
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q5_1-q8_0.cu
472
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-f16.cu
473
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q4_0.cu
474
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q4_1.cu
475
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q5_0.cu
476
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q5_1.cu
477
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs128-q8_0-q8_0.cu
478
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs256-f16-f16.cu
479
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-f16.cu
480
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q4_0.cu
481
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q4_1.cu
482
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q5_0.cu
483
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q5_1.cu
484
- - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-f32-instance-hs64-f16-q8_0.cu
403
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-f16.cu
404
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_0.cu
405
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q4_1.cu
406
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q5_0.cu
407
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q5_1.cu
408
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-f16-q8_0.cu
409
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-f16.cu
410
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_0.cu
411
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q4_1.cu
412
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q5_0.cu
413
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q5_1.cu
414
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_0-q8_0.cu
415
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-f16.cu
416
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q4_0.cu
417
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q4_1.cu
418
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q5_0.cu
419
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q5_1.cu
420
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q4_1-q8_0.cu
421
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-f16.cu
422
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q4_0.cu
423
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q4_1.cu
424
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q5_0.cu
425
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q5_1.cu
426
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_0-q8_0.cu
427
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-f16.cu
428
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q4_0.cu
429
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q4_1.cu
430
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q5_0.cu
431
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q5_1.cu
432
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q5_1-q8_0.cu
433
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-f16.cu
434
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q4_0.cu
435
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q4_1.cu
436
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q5_0.cu
437
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q5_1.cu
438
+ - ext/sources/ggml/src/ggml-cuda/template-instances/fattn-vec-instance-q8_0-q8_0.cu
485
439
  - ext/sources/ggml/src/ggml-cuda/template-instances/generate_cu_files.py
440
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_1.cu
441
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_10.cu
442
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_11.cu
443
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_12.cu
444
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_13.cu
445
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_14.cu
446
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_15.cu
447
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_16.cu
448
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_2.cu
449
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_3.cu
450
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_4.cu
451
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_5.cu
452
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_6.cu
453
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_7.cu
454
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_8.cu
455
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmf-instance-ncols_9.cu
486
456
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-iq1_s.cu
487
457
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_s.cu
488
458
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-iq2_xs.cu
@@ -491,6 +461,7 @@ files:
491
461
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-iq3_xxs.cu
492
462
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_nl.cu
493
463
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-iq4_xs.cu
464
+ - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-mxfp4.cu
494
465
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q2_k.cu
495
466
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q3_k.cu
496
467
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q4_0.cu
@@ -501,6 +472,8 @@ files:
501
472
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q5_k.cu
502
473
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q6_k.cu
503
474
  - ext/sources/ggml/src/ggml-cuda/template-instances/mmq-instance-q8_0.cu
475
+ - ext/sources/ggml/src/ggml-cuda/topk-moe.cu
476
+ - ext/sources/ggml/src/ggml-cuda/topk-moe.cuh
504
477
  - ext/sources/ggml/src/ggml-cuda/tsembd.cu
505
478
  - ext/sources/ggml/src/ggml-cuda/tsembd.cuh
506
479
  - ext/sources/ggml/src/ggml-cuda/unary.cu
@@ -515,48 +488,18 @@ files:
515
488
  - ext/sources/ggml/src/ggml-cuda/wkv.cuh
516
489
  - ext/sources/ggml/src/ggml-hip/CMakeLists.txt
517
490
  - ext/sources/ggml/src/ggml-impl.h
518
- - ext/sources/ggml/src/ggml-kompute/CMakeLists.txt
519
- - ext/sources/ggml/src/ggml-kompute/ggml-kompute.cpp
520
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/common.comp
521
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_add.comp
522
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_addrow.comp
523
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f16_f16.comp
524
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f16_f32.comp
525
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f32_f16.comp
526
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_cpy_f32_f32.comp
527
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_diagmask.comp
528
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_gelu.comp
529
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows.comp
530
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_f16.comp
531
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_f32.comp
532
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q4_0.comp
533
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q4_1.comp
534
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_getrows_q6_k.comp
535
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul.comp
536
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_f16.comp
537
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_mat_f32.comp
538
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_0.comp
539
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_1.comp
540
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q4_k.comp
541
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q6_k.comp
542
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mat_q8_0.comp
543
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n.comp
544
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_mul_mv_q_n_pre.comp
545
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_norm.comp
546
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_relu.comp
547
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rmsnorm.comp
548
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f16.comp
549
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_neox_f32.comp
550
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f16.comp
551
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_rope_norm_f32.comp
552
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_scale.comp
553
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_scale_8.comp
554
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_silu.comp
555
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/op_softmax.comp
556
- - ext/sources/ggml/src/ggml-kompute/kompute-shaders/rope_common.comp
557
491
  - ext/sources/ggml/src/ggml-metal/CMakeLists.txt
492
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-common.cpp
493
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-common.h
494
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-context.h
495
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-context.m
496
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-device.cpp
497
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-device.h
498
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-device.m
558
499
  - ext/sources/ggml/src/ggml-metal/ggml-metal-impl.h
559
- - ext/sources/ggml/src/ggml-metal/ggml-metal.m
500
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-ops.cpp
501
+ - ext/sources/ggml/src/ggml-metal/ggml-metal-ops.h
502
+ - ext/sources/ggml/src/ggml-metal/ggml-metal.cpp
560
503
  - ext/sources/ggml/src/ggml-metal/ggml-metal.metal
561
504
  - ext/sources/ggml/src/ggml-musa/CMakeLists.txt
562
505
  - ext/sources/ggml/src/ggml-musa/mudnn.cu
@@ -564,14 +507,20 @@ files:
564
507
  - ext/sources/ggml/src/ggml-opencl/CMakeLists.txt
565
508
  - ext/sources/ggml/src/ggml-opencl/ggml-opencl.cpp
566
509
  - ext/sources/ggml/src/ggml-opencl/kernels/add.cl
510
+ - ext/sources/ggml/src/ggml-opencl/kernels/add_id.cl
567
511
  - ext/sources/ggml/src/ggml-opencl/kernels/argsort.cl
568
512
  - ext/sources/ggml/src/ggml-opencl/kernels/clamp.cl
569
513
  - ext/sources/ggml/src/ggml-opencl/kernels/concat.cl
514
+ - ext/sources/ggml/src/ggml-opencl/kernels/conv2d.cl
515
+ - ext/sources/ggml/src/ggml-opencl/kernels/conv2d_f16_f32.cl
570
516
  - ext/sources/ggml/src/ggml-opencl/kernels/cpy.cl
571
517
  - ext/sources/ggml/src/ggml-opencl/kernels/cvt.cl
572
518
  - ext/sources/ggml/src/ggml-opencl/kernels/diag_mask_inf.cl
573
519
  - ext/sources/ggml/src/ggml-opencl/kernels/div.cl
574
520
  - ext/sources/ggml/src/ggml-opencl/kernels/embed_kernel.py
521
+ - ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f16.cl
522
+ - ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32.cl
523
+ - ext/sources/ggml/src/ggml-opencl/kernels/flash_attn_f32_f16.cl
575
524
  - ext/sources/ggml/src/ggml-opencl/kernels/gelu.cl
576
525
  - ext/sources/ggml/src/ggml-opencl/kernels/gemv_noshuffle.cl
577
526
  - ext/sources/ggml/src/ggml-opencl/kernels/gemv_noshuffle_general.cl
@@ -582,18 +531,29 @@ files:
582
531
  - ext/sources/ggml/src/ggml-opencl/kernels/im2col_f32.cl
583
532
  - ext/sources/ggml/src/ggml-opencl/kernels/mul.cl
584
533
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mat_Ab_Bi_8x4.cl
534
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mat_f16_f32.cl
535
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f16_f32_l4_lm.cl
536
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mm_f32_f32_l4_lm.cl
585
537
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f16.cl
586
538
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f32.cl
587
539
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f32_1row.cl
588
540
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f16_f32_l4.cl
589
541
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_f32_f32.cl
542
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_mxfp4_f32.cl
543
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_mxfp4_f32_flat.cl
590
544
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_q4_0_f32_8x_flat.cl
545
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_q8_0_f32.cl
546
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_id_q8_0_f32_flat.cl
547
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_mxfp4_f32.cl
548
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_mxfp4_f32_flat.cl
591
549
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q4_0_f32.cl
592
550
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q4_0_f32_1d_16x_flat.cl
593
551
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q4_0_f32_1d_8x_flat.cl
594
552
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q4_0_f32_8x_flat.cl
595
553
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q4_0_f32_v.cl
596
554
  - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q6_k.cl
555
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q8_0_f32.cl
556
+ - ext/sources/ggml/src/ggml-opencl/kernels/mul_mv_q8_0_f32_flat.cl
597
557
  - ext/sources/ggml/src/ggml-opencl/kernels/norm.cl
598
558
  - ext/sources/ggml/src/ggml-opencl/kernels/pad.cl
599
559
  - ext/sources/ggml/src/ggml-opencl/kernels/relu.cl
@@ -601,6 +561,7 @@ files:
601
561
  - ext/sources/ggml/src/ggml-opencl/kernels/rms_norm.cl
602
562
  - ext/sources/ggml/src/ggml-opencl/kernels/rope.cl
603
563
  - ext/sources/ggml/src/ggml-opencl/kernels/scale.cl
564
+ - ext/sources/ggml/src/ggml-opencl/kernels/set_rows.cl
604
565
  - ext/sources/ggml/src/ggml-opencl/kernels/sigmoid.cl
605
566
  - ext/sources/ggml/src/ggml-opencl/kernels/silu.cl
606
567
  - ext/sources/ggml/src/ggml-opencl/kernels/softmax_4_f16.cl
@@ -655,9 +616,12 @@ files:
655
616
  - ext/sources/ggml/src/ggml-sycl/outprod.cpp
656
617
  - ext/sources/ggml/src/ggml-sycl/outprod.hpp
657
618
  - ext/sources/ggml/src/ggml-sycl/presets.hpp
619
+ - ext/sources/ggml/src/ggml-sycl/quantize.hpp
658
620
  - ext/sources/ggml/src/ggml-sycl/quants.hpp
659
621
  - ext/sources/ggml/src/ggml-sycl/rope.cpp
660
622
  - ext/sources/ggml/src/ggml-sycl/rope.hpp
623
+ - ext/sources/ggml/src/ggml-sycl/set_rows.cpp
624
+ - ext/sources/ggml/src/ggml-sycl/set_rows.hpp
661
625
  - ext/sources/ggml/src/ggml-sycl/softmax.cpp
662
626
  - ext/sources/ggml/src/ggml-sycl/softmax.hpp
663
627
  - ext/sources/ggml/src/ggml-sycl/sycl_hw.cpp
@@ -675,12 +639,14 @@ files:
675
639
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/CMakeLists.txt
676
640
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/acc.comp
677
641
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add.comp
642
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/add_id.comp
678
643
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argmax.comp
679
644
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/argsort.comp
680
645
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/clamp.comp
681
646
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/concat.comp
682
647
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/contig_copy.comp
683
648
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv2d_dw.comp
649
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv2d_mm.comp
684
650
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/conv_transpose_1d.comp
685
651
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy.comp
686
652
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/copy_from_quant.comp
@@ -700,6 +666,7 @@ files:
700
666
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq3_xxs.comp
701
667
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq4_nl.comp
702
668
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_iq4_xs.comp
669
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_mxfp4.comp
703
670
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q2_k.comp
704
671
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q3_k.comp
705
672
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q4_0.comp
@@ -712,13 +679,17 @@ files:
712
679
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/dequant_q8_0.comp
713
680
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/diag_mask_inf.comp
714
681
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/div.comp
682
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/exp.comp
715
683
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn.comp
716
684
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_base.comp
717
685
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm1.comp
718
686
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_cm2.comp
719
687
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/flash_attn_split_k_reduce.comp
720
688
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu.comp
689
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_erf.comp
690
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/geglu_quick.comp
721
691
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu.comp
692
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_erf.comp
722
693
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/gelu_quick.comp
723
694
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_binary_head.comp
724
695
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/generic_head.comp
@@ -728,7 +699,10 @@ files:
728
699
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_head.comp
729
700
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/glu_main.comp
730
701
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/group_norm.comp
702
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardsigmoid.comp
703
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/hardswish.comp
731
704
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col.comp
705
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/im2col_3d.comp
732
706
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/l2_norm.comp
733
707
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/leaky_relu.comp
734
708
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul.comp
@@ -749,12 +723,16 @@ files:
749
723
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q4_k.comp
750
724
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q5_k.comp
751
725
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vec_q6_k.comp
726
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mat_vecq.comp
752
727
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp
753
728
  - 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
754
730
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq.comp
755
731
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/mul_mmq_funcs.comp
732
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/multi_add.comp
756
733
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/norm.comp
757
734
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_adamw.comp
735
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/opt_step_sgd.comp
758
736
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/pad.comp
759
737
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/pool2d.comp
760
738
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/quantize_q8_1.comp
@@ -764,11 +742,14 @@ files:
764
742
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/repeat_back.comp
765
743
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm.comp
766
744
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_back.comp
745
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rms_norm_partials.comp
746
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/roll.comp
767
747
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_head.comp
768
748
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_multi.comp
769
749
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_neox.comp
770
750
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_norm.comp
771
751
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rope_vision.comp
752
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/rte.comp
772
753
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/scale.comp
773
754
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sigmoid.comp
774
755
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/silu.comp
@@ -776,10 +757,12 @@ files:
776
757
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sin.comp
777
758
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max.comp
778
759
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/soft_max_back.comp
760
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sqrt.comp
779
761
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/square.comp
780
762
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sub.comp
781
763
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/sum_rows.comp
782
764
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu.comp
765
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/swiglu_oai.comp
783
766
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/tanh.comp
784
767
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/test_bfloat16_support.comp
785
768
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/test_coopmat2_support.comp
@@ -788,9 +771,33 @@ files:
788
771
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/timestep_embedding.comp
789
772
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/types.comp
790
773
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/upscale.comp
774
+ - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/utils.comp
791
775
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp
792
776
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/wkv6.comp
793
777
  - ext/sources/ggml/src/ggml-vulkan/vulkan-shaders/wkv7.comp
778
+ - ext/sources/ggml/src/ggml-webgpu/CMakeLists.txt
779
+ - 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
782
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/binary_head.tmpl
783
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/common_decls.tmpl
784
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/cpy.wgsl
785
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/embed_wgsl.py
786
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/get_rows.tmpl.wgsl
787
+ - 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
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/mul_mat.tmpl.wgsl
791
+ - 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
793
+ - ext/sources/ggml/src/ggml-webgpu/wgsl-shaders/set_rows.wgsl
794
+ - ext/sources/ggml/src/ggml-zdnn/CMakeLists.txt
795
+ - ext/sources/ggml/src/ggml-zdnn/common.hpp
796
+ - ext/sources/ggml/src/ggml-zdnn/ggml-zdnn.cpp
797
+ - ext/sources/ggml/src/ggml-zdnn/mmf.cpp
798
+ - ext/sources/ggml/src/ggml-zdnn/mmf.hpp
799
+ - ext/sources/ggml/src/ggml-zdnn/utils.cpp
800
+ - ext/sources/ggml/src/ggml-zdnn/utils.hpp
794
801
  - ext/sources/ggml/src/ggml.c
795
802
  - ext/sources/ggml/src/ggml.cpp
796
803
  - ext/sources/ggml/src/gguf.cpp
@@ -872,7 +879,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
872
879
  - !ruby/object:Gem::Version
873
880
  version: '0'
874
881
  requirements: []
875
- rubygems_version: 3.1.2
882
+ rubygems_version: 3.5.22
876
883
  signing_key:
877
884
  specification_version: 4
878
885
  summary: Ruby whisper.cpp bindings