try CI fix

Author: JohannesGaessler

ggml/src/ggml-cuda/mmf.cu

@@ -14,6 +14,7 @@ static __global__ void mul_mat_f(
         const int ncols, const int nchannels_y, const int stride_row, const int stride_col_y, const int stride_col_dst,
         const int channel_ratio, const int stride_channel_x, const int stride_channel_y, const int stride_channel_dst,
         const int sample_ratio, const int stride_sample_x, const int stride_sample_y, const int stride_sample_dst) {
+#if !defined(GGML_USE_HIP) && !defined(GGML_USE_MUSA)
     typedef tile<16, 8, T>     tile_A;
     typedef tile< 8, 8, T>     tile_B;
     typedef tile<16, 8, float> tile_C;
@@ -130,6 +131,13 @@ static __global__ void mul_mat_f(
         }
         dst[j*stride_col_dst + row0 + threadIdx.x] = sum;
     }
+#else
+    NO_DEVICE_CODE;
+    GGML_UNUSED(x); GGML_UNUSED(y); GGML_UNUSED(ids); GGML_UNUSED(dst);
+    GGML_UNUSED(ncols); GGML_UNUSED(nchannels_y); GGML_UNUSED(stride_row); GGML_UNUSED(stride_col_y); GGML_UNUSED(stride_col_dst);
+    GGML_UNUSED(channel_ratio); GGML_UNUSED(stride_channel_x); GGML_UNUSED(stride_channel_y); GGML_UNUSED(stride_channel_dst);
+    GGML_UNUSED(sample_ratio); GGML_UNUSED(stride_sample_x); GGML_UNUSED(stride_sample_y); GGML_UNUSED(stride_sample_dst);
+#endif // !defined(GGML_USE_HIP) && !defined(GGML_USE_MUSA)
 }
 
 template <typename T, int cols_per_block>

ggml/src/ggml-cuda/vendors/hip.h

@@ -4,6 +4,7 @@
 #include <hip/hip_runtime.h>
 #include <hipblas/hipblas.h>
 #include <hip/hip_fp16.h>
+#include <hip/hip_bf16.h>
 #include <hip/hip_bfloat16.h>
 // for rocblas_initialize()
 #include "rocblas/rocblas.h"
@@ -200,6 +201,7 @@
 #endif
 
 typedef hip_bfloat16 nv_bfloat16;
+typedef __hip_bfloat162 nv_bfloat162;
 
 typedef int8_t int8x4_t __attribute__((ext_vector_type(4)));
 typedef uint8_t uint8x4_t __attribute__((ext_vector_type(4)));

ggml/src/ggml-cuda/vendors/musa.h

@@ -137,4 +137,5 @@
 #define cudaStreamEndCapture musaStreamEndCapture
 #define cudaOccupancyMaxActiveBlocksPerMultiprocessor musaOccupancyMaxActiveBlocksPerMultiprocessor
 
-typedef mt_bfloat16 nv_bfloat16;
+typedef __mt_bfloat16 nv_bfloat16;
+typedef __mt_bfloat162 nv_bfloat162;