diff --git a/ggml/src/ggml-vulkan/ggml-vulkan.cpp b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
index ebbb412e55f..ee6ce37db39 100644
--- a/ggml/src/ggml-vulkan/ggml-vulkan.cpp
+++ b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
@@ -3114,13 +3114,15 @@ static void ggml_vk_load_shaders(vk_device& device) {
         rm_stdq = 2;
     uint32_t rm_iq = 2 * rm_kq;
 
-    const bool use_subgroups = device->subgroup_arithmetic && device->architecture != vk_device_architecture::AMD_GCN;
+    const bool use_subgroups_float = device->subgroup_arithmetic;
+    const bool use_subgroups = use_subgroups_float && device->architecture != vk_device_architecture::AMD_GCN;
     // Ensure a subgroup size >= 16 is available
     const bool use_subgroups16 = use_subgroups && subgroup_min_size_16;
 
     const uint32_t subgroup_size = (device->vendor_id == VK_VENDOR_ID_INTEL && device->subgroup_size_control && device->subgroup_min_size <= 16 && device->subgroup_max_size >= 16) ? 16 : device->subgroup_size;
     const uint32_t subgroup_size16 = std::max(subgroup_size, 16u);
 
+    const uint32_t force_subgroup_size_float = use_subgroups_float ? subgroup_size : 0;
     const uint32_t force_subgroup_size = use_subgroups ? subgroup_size : 0;
     const uint32_t force_subgroup_size16 = use_subgroups16 ? subgroup_size16 : 0;
 
@@ -3128,6 +3130,10 @@ static void ggml_vk_load_shaders(vk_device& device) {
         const uint32_t wg_size_subgroup   = (w == DMMV_WG_SIZE_SUBGROUP) ? subgroup_size : (subgroup_size * 4);
         const uint32_t wg_size_subgroup16 = (w == DMMV_WG_SIZE_SUBGROUP) ? subgroup_size16 : (subgroup_size16 * 4);
 
+        const shader_reduction_mode reduc_float = (use_subgroups_float && w == DMMV_WG_SIZE_SUBGROUP) ? SHADER_REDUCTION_MODE_SUBGROUP :
+                                                 (use_subgroups_float && w == DMMV_WG_SIZE_LARGE) ? SHADER_REDUCTION_MODE_HYBRID :
+                                                 SHADER_REDUCTION_MODE_SHMEM;
+
         const shader_reduction_mode reduc = (use_subgroups && w == DMMV_WG_SIZE_SUBGROUP) ? SHADER_REDUCTION_MODE_SUBGROUP :
                                             (use_subgroups && w == DMMV_WG_SIZE_LARGE) ? SHADER_REDUCTION_MODE_HYBRID :
                                             SHADER_REDUCTION_MODE_SHMEM;
@@ -3137,9 +3143,9 @@ static void ggml_vk_load_shaders(vk_device& device) {
                                               SHADER_REDUCTION_MODE_SHMEM;
 
         for (uint32_t i = 0; i < mul_mat_vec_max_cols; ++i) {
-            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_F32 ][i], "mul_mat_vec_f32_f32_f32",  arr_dmmv_f32_f32_f32_len[reduc],  arr_dmmv_f32_f32_f32_data[reduc],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups, force_subgroup_size);
-            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_F16 ][i], "mul_mat_vec_f16_f32_f32",  arr_dmmv_f16_f32_f32_len[reduc],  arr_dmmv_f16_f32_f32_data[reduc],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups, force_subgroup_size);
-            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_BF16][i], "mul_mat_vec_bf16_f32_f32", arr_dmmv_bf16_f32_f32_len[reduc], arr_dmmv_bf16_f32_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups, force_subgroup_size);
+            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_F32 ][i], "mul_mat_vec_f32_f32_f32",  arr_dmmv_f32_f32_f32_len[reduc_float],  arr_dmmv_f32_f32_f32_data[reduc_float],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups_float, force_subgroup_size_float);
+            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_F16 ][i], "mul_mat_vec_f16_f32_f32",  arr_dmmv_f16_f32_f32_len[reduc_float],  arr_dmmv_f16_f32_f32_data[reduc_float],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups_float, force_subgroup_size_float);
+            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_BF16][i], "mul_mat_vec_bf16_f32_f32", arr_dmmv_bf16_f32_f32_len[reduc_float], arr_dmmv_bf16_f32_f32_data[reduc_float], "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups_float, force_subgroup_size_float);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_Q4_0][i], "mul_mat_vec_q4_0_f32_f32", arr_dmmv_q4_0_f32_f32_len[reduc], arr_dmmv_q4_0_f32_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2*rm_stdq, 1, 1}, {wg_size_subgroup, 2*rm_stdq, i+1}, 1, true, use_subgroups, force_subgroup_size);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_Q4_1][i], "mul_mat_vec_q4_1_f32_f32", arr_dmmv_q4_1_f32_f32_len[reduc], arr_dmmv_q4_1_f32_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2*rm_stdq, 1, 1}, {wg_size_subgroup, 2*rm_stdq, i+1}, 1, true, use_subgroups, force_subgroup_size);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_Q5_0][i], "mul_mat_vec_q5_0_f32_f32", arr_dmmv_q5_0_f32_f32_len[reduc], arr_dmmv_q5_0_f32_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2*rm_stdq, 1, 1}, {wg_size_subgroup, 2*rm_stdq, i+1}, 1, true, use_subgroups, force_subgroup_size);
@@ -3161,9 +3167,9 @@ static void ggml_vk_load_shaders(vk_device& device) {
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_IQ4_NL][i],  "mul_mat_vec_iq4_nl_f32_f32",  arr_dmmv_iq4_nl_f32_f32_len[reduc16],  arr_dmmv_iq4_nl_f32_f32_data[reduc16],  "main", 3, sizeof(vk_mat_vec_push_constants), {rm_iq, 1, 1}, {wg_size_subgroup16, rm_iq, i+1}, 1, true, use_subgroups16, force_subgroup_size16);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f32_f32[w][GGML_TYPE_MXFP4][i],   "mul_mat_vec_mxfp4_f32_f32",   arr_dmmv_mxfp4_f32_f32_len[reduc16],   arr_dmmv_mxfp4_f32_f32_data[reduc16],   "main", 3, sizeof(vk_mat_vec_push_constants), {rm_iq, 1, 1}, {wg_size_subgroup16, rm_iq, i+1}, 1, true, use_subgroups16, force_subgroup_size16);
 
-            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_F32 ][i], "mul_mat_vec_f32_f16_f32",  arr_dmmv_f32_f16_f32_len[reduc],  arr_dmmv_f32_f16_f32_data[reduc],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups, force_subgroup_size);
-            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_F16 ][i], "mul_mat_vec_f16_f16_f32",  arr_dmmv_f16_f16_f32_len[reduc],  arr_dmmv_f16_f16_f32_data[reduc],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups, force_subgroup_size);
-            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_BF16][i], "mul_mat_vec_bf16_f16_f32", arr_dmmv_bf16_f16_f32_len[reduc], arr_dmmv_bf16_f16_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups, force_subgroup_size);
+            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_F32 ][i], "mul_mat_vec_f32_f16_f32",  arr_dmmv_f32_f16_f32_len[reduc_float],  arr_dmmv_f32_f16_f32_data[reduc_float],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups_float, force_subgroup_size_float);
+            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_F16 ][i], "mul_mat_vec_f16_f16_f32",  arr_dmmv_f16_f16_f32_len[reduc_float],  arr_dmmv_f16_f16_f32_data[reduc_float],  "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups_float, force_subgroup_size_float);
+            ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_BF16][i], "mul_mat_vec_bf16_f16_f32", arr_dmmv_bf16_f16_f32_len[reduc_float], arr_dmmv_bf16_f16_f32_data[reduc_float], "main", 3, sizeof(vk_mat_vec_push_constants), {2, 1, 1}, {wg_size_subgroup, 2, i+1}, 1, false, use_subgroups_float, force_subgroup_size_float);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_Q4_0][i], "mul_mat_vec_q4_0_f16_f32", arr_dmmv_q4_0_f16_f32_len[reduc], arr_dmmv_q4_0_f16_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2*rm_stdq, 1, 1}, {wg_size_subgroup, 2*rm_stdq, i+1}, 1, true, use_subgroups, force_subgroup_size);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_Q4_1][i], "mul_mat_vec_q4_1_f16_f32", arr_dmmv_q4_1_f16_f32_len[reduc], arr_dmmv_q4_1_f16_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2*rm_stdq, 1, 1}, {wg_size_subgroup, 2*rm_stdq, i+1}, 1, true, use_subgroups, force_subgroup_size);
             ggml_vk_create_pipeline(device, device->pipeline_dequant_mul_mat_vec_f16_f32[w][GGML_TYPE_Q5_0][i], "mul_mat_vec_q5_0_f16_f32", arr_dmmv_q5_0_f16_f32_len[reduc], arr_dmmv_q5_0_f16_f32_data[reduc], "main", 3, sizeof(vk_mat_vec_push_constants), {2*rm_stdq, 1, 1}, {wg_size_subgroup, 2*rm_stdq, i+1}, 1, true, use_subgroups, force_subgroup_size);