aboutsummaryrefslogtreecommitdiff
path: root/ggml-cuda.h
diff options
context:
space:
mode:
Diffstat (limited to 'ggml-cuda.h')
-rw-r--r--ggml-cuda.h10
1 files changed, 9 insertions, 1 deletions
diff --git a/ggml-cuda.h b/ggml-cuda.h
index 1fd67eb..36782d9 100644
--- a/ggml-cuda.h
+++ b/ggml-cuda.h
@@ -26,9 +26,14 @@ extern "C" {
} while (0)
extern cublasHandle_t g_cublasH;
-extern cudaStream_t g_cudaStream;
+extern cudaStream_t g_cudaStream;
+extern cudaStream_t g_cudaStream2;
+extern cudaEvent_t g_cudaEvent;
void ggml_init_cublas(void);
+void * ggml_cuda_host_malloc(size_t size);
+void ggml_cuda_host_free(void * ptr);
+
void * ggml_cuda_pool_malloc(size_t size, size_t * actual_size);
void ggml_cuda_pool_free(void * ptr, size_t size);
@@ -41,6 +46,9 @@ void dequantize_row_q8_0_cuda(const void * vx, float * y, int k, cudaStream_t st
cudaError_t ggml_cuda_h2d_tensor_2d(void * dst, const struct ggml_tensor * src, uint64_t i3, uint64_t i2, cudaStream_t stream);
+typedef void (*dequantize_row_q_cuda_t)(const void * x, float * y, int k, cudaStream_t stream);
+dequantize_row_q_cuda_t ggml_get_dequantize_row_q_cuda(enum ggml_type type);
+
#ifdef __cplusplus
}
#endif