vulkan: fix matmul integer pipeline selection (#23005)

* vulkan: fix matmul integer pipeline selection

* gate pipeline creation with the right bools

diff --git a/ggml/src/ggml-vulkan/ggml-vulkan.cpp b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
index a0a556206d53d22f2297645aa2f593c6e3f455b7..8c4cf9ef1dbe8e971cb05d72060044a051a60cb3 100644 (file)
--- a/ggml/src/ggml-vulkan/ggml-vulkan.cpp
+++ b/ggml/src/ggml-vulkan/ggml-vulkan.cpp
@@ -3954,13 +3954,13 @@ static void ggml_vk_load_shaders(vk_device& device) {
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME ->a_s, #NAMELC #F16ACC "_aligned_s", NAMELC ## _aligned ## F16ACC ## _len, NAMELC ## _aligned ## F16ACC ## _data, "main", PARAMCOUNT, sizeof(PUSHCONST), s_ ## WG_DENOMS, s_ ## WARPTILE, s_align, false, REQSUBGROUPSIZE > 0, REQSUBGROUPSIZE);   \
 
 #define CREATE_MMQ(TYPE, PIPELINE_NAME, NAMELC, WG_DENOMS, WARPTILE, PUSHCONST, PARAMCOUNT, ID, REQSUBGROUPSIZE) \
-        if (device->mul_mat ## ID ## _l[TYPE]) { \
+        if (device->mul_mat ## ID ## _l_int[TYPE]) { \
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME .f32acc->l, #NAMELC        "_l", NAMELC ## _len,        NAMELC ##  _data,        "main", PARAMCOUNT, sizeof(PUSHCONST), l_ ## WG_DENOMS, l_ ## WARPTILE, 1, false, REQSUBGROUPSIZE > 0, REQSUBGROUPSIZE);   \
         } \
-        if (device->mul_mat ## ID ## _m[TYPE]) { \
+        if (device->mul_mat ## ID ## _m_int[TYPE]) { \
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME .f32acc->m, #NAMELC        "_m", NAMELC ## _len,        NAMELC ##  _data,        "main", PARAMCOUNT, sizeof(PUSHCONST), m_ ## WG_DENOMS, m_ ## WARPTILE, 1, false, REQSUBGROUPSIZE > 0, REQSUBGROUPSIZE);   \
         } \
-        if (device->mul_mat ## ID ## _s[TYPE]) { \
+        if (device->mul_mat ## ID ## _s_int[TYPE]) { \
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME .f32acc->s, #NAMELC        "_s", NAMELC ## _len,        NAMELC ##  _data,        "main", PARAMCOUNT, sizeof(PUSHCONST), s_ ## WG_DENOMS, s_ ## WARPTILE, 1, false, REQSUBGROUPSIZE > 0, REQSUBGROUPSIZE);   \
         } \
 
@@ -4131,11 +4131,11 @@ static void ggml_vk_load_shaders(vk_device& device) {
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME ->a_s, #NAMELC #F16ACC "_aligned_s", NAMELC ## _aligned ## F16ACC ## _fp32_len, NAMELC ## _aligned ## F16ACC ## _fp32_data, "main", PARAMCOUNT, sizeof(PUSHCONST), s_ ## WG_DENOMS, s_ ## WARPTILE, s_align, false, REQSUBGROUPSIZE > 0, REQSUBGROUPSIZE);   \
 
 #define CREATE_MMQ(TYPE, PIPELINE_NAME, NAMELC, WG_DENOMS, WARPTILE, PUSHCONST, PARAMCOUNT, ID) \
-        if (device->mul_mat ## ID ## _l[TYPE]) \
+        if (device->mul_mat ## ID ## _l_int[TYPE]) \
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME ->l, #NAMELC "_l", NAMELC ## _fp32_len, NAMELC ## _fp32_data, "main", PARAMCOUNT, sizeof(PUSHCONST), l_ ## WG_DENOMS, l_ ## WARPTILE, 1);   \
-        if (device->mul_mat ## ID ## _m[TYPE]) \
+        if (device->mul_mat ## ID ## _m_int[TYPE]) \
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME ->m, #NAMELC "_m", NAMELC ## _fp32_len, NAMELC ## _fp32_data, "main", PARAMCOUNT, sizeof(PUSHCONST), m_ ## WG_DENOMS, m_ ## WARPTILE, 1);   \
-        if (device->mul_mat ## ID ## _s[TYPE]) \
+        if (device->mul_mat ## ID ## _s_int[TYPE]) \
             ggml_vk_create_pipeline(device, device-> PIPELINE_NAME ->s, #NAMELC "_s", NAMELC ## _fp32_len, NAMELC ## _fp32_data, "main", PARAMCOUNT, sizeof(PUSHCONST), s_ ## WG_DENOMS, s_ ## WARPTILE, 1);   \
 
         CREATE_MM(GGML_TYPE_F32, pipeline_matmul_f32, matmul_f32_f32, , wg_denoms, warptile, vk_mat_mat_push_constants, 3, , 0);
@@ -5716,12 +5716,12 @@ static vk_device ggml_vk_get_device(size_t idx) {
                 break;
             }
 
-            device->mul_mat_l_int[i]    = true;
-            device->mul_mat_m_int[i]    = true;
-            device->mul_mat_s_int[i]    = true;
-            device->mul_mat_id_l_int[i] = true;
-            device->mul_mat_id_m_int[i] = true;
-            device->mul_mat_id_s_int[i] = true;
+            device->mul_mat_l_int[i]    = device->mul_mat_l[i];
+            device->mul_mat_m_int[i]    = device->mul_mat_m[i];
+            device->mul_mat_s_int[i]    = device->mul_mat_s[i];
+            device->mul_mat_id_l_int[i] = device->mul_mat_id_l[i];
+            device->mul_mat_id_m_int[i] = device->mul_mat_id_m[i];
+            device->mul_mat_id_s_int[i] = device->mul_mat_id_s[i];
         }