From: Jeff Bolz Date: Tue, 25 Feb 2025 15:30:21 +0000 (-0600) Subject: vulkan: fix assertion when qy_needs_dequant (#12068) X-Git-Tag: upstream/0.0.4853~75 X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=a82c9e7c23ef6db48cebfa194dc9cebbc4ac3552;p=pkg%2Fggml%2Fsources%2Fllama.cpp vulkan: fix assertion when qy_needs_dequant (#12068) Looks like a copy/paste bug from qx_needs_dequant. --- diff --git a/ggml/src/ggml-vulkan/ggml-vulkan.cpp b/ggml/src/ggml-vulkan/ggml-vulkan.cpp index 5864e981..abe3e790 100644 --- a/ggml/src/ggml-vulkan/ggml-vulkan.cpp +++ b/ggml/src/ggml-vulkan/ggml-vulkan.cpp @@ -4192,7 +4192,7 @@ static void ggml_vk_mul_mat_q_f16(ggml_backend_vk_context * ctx, vk_context& sub } if (qy_needs_dequant) { d_Y = ctx->prealloc_y; - GGML_ASSERT(d_Y->size >= y_sz * ne02 * ne03); + GGML_ASSERT(d_Y->size >= y_sz * ne12 * ne13); } else { d_Y = d_Qy; y_buf_offset = qy_buf_offset; @@ -4769,7 +4769,7 @@ static void ggml_vk_mul_mat_id_q_f16(ggml_backend_vk_context * ctx, vk_context& } if (qy_needs_dequant) { d_Y = ctx->prealloc_y; - GGML_ASSERT(d_Y->size >= y_sz * ne02 * ne03); + GGML_ASSERT(d_Y->size >= y_sz * ne12 * ne13); } else { d_Y = d_Qy; y_buf_offset = qy_buf_offset;