From cddf92ae35c83d05951d83e833798575e7cb2d3b Mon Sep 17 00:00:00 2001 From: deano Date: Thu, 2 Jul 2026 13:11:12 +0300 Subject: [PATCH] deps: point llama.cpp at GeometricAGI fork with RoPE correctness fix Bumps server/deps/llama.cpp from ac06e543 (Luce-Org perf/luce-verify-kernels) to GeometricAGI/lucebox-ggml @ 0fe65d93 = ac06e543 + one commit: 'ggml-cuda: defer RoPE mod-2pi reduction until after angle scaling'. rope_theta_fp64() reduced the rotation angle mod-2pi before rope_yarn's freq_scale multiply and the callers' /freq_factor; the reduction does not commute with scaling, so every RoPE config with freq_scale!=1 or freq_factor!=1 was numerically wrong (freq_factors, YaRN, mrope/vision/imrope, ROPE_BACK, ROPE_SET_ROWS). rope.cu is byte-identical between ac06e543 and 30c9d7d, where this exact change was verified via test-backend-ops on gfx1201 + gfx1100: ROPE 268/268, ROPE_BACK 228/228, ROPE_SET_ROWS 48/48 (was 356 failures), SOFT_MAX unchanged 212/212. Upstream to Luce-Org when ready. --- .gitmodules | 4 ++-- server/deps/llama.cpp | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/.gitmodules b/.gitmodules index c3b57efdc..f8245a147 100644 --- a/.gitmodules +++ b/.gitmodules @@ -1,7 +1,7 @@ [submodule "dflash/deps/llama.cpp"] path = server/deps/llama.cpp - url = https://github.com/Luce-Org/llama.cpp-dflash-ggml.git - branch = luce-dflash + url = https://github.com/GeometricAGI/lucebox-ggml.git + branch = fix/rope-fp64-deferred-reduction [submodule "dflash/deps/Block-Sparse-Attention"] path = server/deps/Block-Sparse-Attention url = https://github.com/mit-han-lab/Block-Sparse-Attention.git diff --git a/server/deps/llama.cpp b/server/deps/llama.cpp index ac06e5431..0fe65d935 160000 --- a/server/deps/llama.cpp +++ b/server/deps/llama.cpp @@ -1 +1 @@ -Subproject commit ac06e5431da1768e6838a5616d60c6053f6af19d +Subproject commit 0fe65d9354b7c5da52a7741d2e37ba85f0d0c925