diff --git a/ggml/src/ggml-vulkan/ggml-vulkan.cpp b/ggml/src/ggml-vulkan/ggml-vulkan.cpp index ae6e76c15..c5368bfe8 100644 --- a/ggml/src/ggml-vulkan/ggml-vulkan.cpp +++ b/ggml/src/ggml-vulkan/ggml-vulkan.cpp @@ -1404,10 +1404,10 @@ static void ggml_vk_load_shaders(vk_device& device) { // spec constants and tile sizes for non-quant matmul/matmul_id l_warptile = { 256, 128, 256, 64 }; m_warptile = { 256, 128, 128, 64 }; - s_warptile = { 128, 32, 16, 64 }; + s_warptile = { 128, 64, 64, 64 }; l_wg_denoms = {128, 256, 1 }; m_wg_denoms = {128, 128, 1 }; - s_wg_denoms = { 32, 16, 1 }; + s_wg_denoms = { 64, 64, 1 }; // spec constants and tile sizes for quant matmul (non-Qi_K) l_warptile_mmq = { 256, 128, 256, 64 };