Regenerate based on CUDA 11.2

cublas-sys/bindgen.sh

@@ -10,6 +10,7 @@ bindgen \
   --with-derive-eq \
   --with-derive-hash \
   --with-derive-ord \
-  /opt/cuda/include/cublas.h \
-  -- -I/opt/cuda/include \
+  --size_t-is-usize \
+  "$CUDA_PATH/include/cublas.h" \
+  -- -I"$CUDA_PATH/include" \
   > src/cublas.rs

cublas-sys/src/cublas.rs

@@ -1,4 +1,4 @@
-/* automatically generated by rust-bindgen */
+/* automatically generated by rust-bindgen 0.55.1 */
 
 #[repr(C)]
 #[repr(align(8))]
@@ -81,18 +81,32 @@ pub type cudaStream_t = *mut CUstream_st;
 pub enum cudaDataType_t {
     CUDA_R_16F = 2,
     CUDA_C_16F = 6,
+    CUDA_R_16BF = 14,
+    CUDA_C_16BF = 15,
     CUDA_R_32F = 0,
     CUDA_C_32F = 4,
     CUDA_R_64F = 1,
     CUDA_C_64F = 5,
+    CUDA_R_4I = 16,
+    CUDA_C_4I = 17,
+    CUDA_R_4U = 18,
+    CUDA_C_4U = 19,
     CUDA_R_8I = 3,
     CUDA_C_8I = 7,
     CUDA_R_8U = 8,
     CUDA_C_8U = 9,
+    CUDA_R_16I = 20,
+    CUDA_C_16I = 21,
+    CUDA_R_16U = 22,
+    CUDA_C_16U = 23,
     CUDA_R_32I = 10,
     CUDA_C_32I = 11,
     CUDA_R_32U = 12,
     CUDA_C_32U = 13,
+    CUDA_R_64I = 24,
+    CUDA_C_64I = 25,
+    CUDA_R_64U = 26,
+    CUDA_C_64U = 27,
 }
 pub use self::cudaDataType_t as cudaDataType;
 #[repr(u32)]
@@ -220,8 +234,26 @@ pub enum cublasGemmAlgo_t {
 pub enum cublasMath_t {
     CUBLAS_DEFAULT_MATH = 0,
     CUBLAS_TENSOR_OP_MATH = 1,
+    CUBLAS_PEDANTIC_MATH = 2,
+    CUBLAS_TF32_TENSOR_OP_MATH = 3,
+    CUBLAS_MATH_DISALLOW_REDUCED_PRECISION_REDUCTION = 16,
 }
 pub use self::cudaDataType as cublasDataType_t;
+#[repr(u32)]
+#[derive(Debug, Copy, Clone, PartialEq, Eq, Hash, PartialOrd, Ord)]
+pub enum cublasComputeType_t {
+    CUBLAS_COMPUTE_16F = 64,
+    CUBLAS_COMPUTE_16F_PEDANTIC = 65,
+    CUBLAS_COMPUTE_32F = 68,
+    CUBLAS_COMPUTE_32F_PEDANTIC = 69,
+    CUBLAS_COMPUTE_32F_FAST_16F = 74,
+    CUBLAS_COMPUTE_32F_FAST_16BF = 75,
+    CUBLAS_COMPUTE_32F_FAST_TF32 = 77,
+    CUBLAS_COMPUTE_64F = 70,
+    CUBLAS_COMPUTE_64F_PEDANTIC = 71,
+    CUBLAS_COMPUTE_32I = 72,
+    CUBLAS_COMPUTE_32I_PEDANTIC = 73,
+}
 #[repr(C)]
 #[derive(Debug, Copy, Clone)]
 pub struct cublasContext {
@@ -249,6 +281,13 @@ extern "C" {
 extern "C" {
     pub fn cublasGetCudartVersion() -> usize;
 }
+extern "C" {
+    pub fn cublasSetWorkspace_v2(
+        handle: cublasHandle_t,
+        workspace: *mut ::std::os::raw::c_void,
+        workspaceSizeInBytes: usize,
+    ) -> cublasStatus_t;
+}
 extern "C" {
     pub fn cublasSetStream_v2(handle: cublasHandle_t, streamId: cudaStream_t) -> cublasStatus_t;
 }
@@ -2280,7 +2319,7 @@ extern "C" {
         C: *mut ::std::os::raw::c_void,
         Ctype: cudaDataType,
         ldc: ::std::os::raw::c_int,
-        computeType: cudaDataType,
+        computeType: cublasComputeType_t,
         algo: cublasGemmAlgo_t,
     ) -> cublasStatus_t;
 }
@@ -3042,7 +3081,7 @@ extern "C" {
         Ctype: cudaDataType,
         ldc: ::std::os::raw::c_int,
         batchCount: ::std::os::raw::c_int,
-        computeType: cudaDataType,
+        computeType: cublasComputeType_t,
         algo: cublasGemmAlgo_t,
     ) -> cublasStatus_t;
 }
@@ -3069,7 +3108,7 @@ extern "C" {
         ldc: ::std::os::raw::c_int,
         strideC: ::std::os::raw::c_longlong,
         batchCount: ::std::os::raw::c_int,
-        computeType: cudaDataType,
+        computeType: cublasComputeType_t,
         algo: cublasGemmAlgo_t,
     ) -> cublasStatus_t;
 }

cuda-driver-sys/bindgen.sh

@@ -16,5 +16,6 @@ bindgen \
   --with-derive-eq \
   --with-derive-hash \
   --with-derive-ord \
-  wrapper.h -- -I/opt/cuda/include \
+  --size_t-is-usize \
+  wrapper.h -- -I"$CUDA_PATH/include" \
   > src/cuda.rs