pytorch-1.8-jetson.patch

diff --git a/aten/src/ATen/cpu/vec256/vec256_float_neon.h b/aten/src/ATen/cpu/vec256/vec256_float_neon.h
index 58a4afa..8c53933 100644
--- a/aten/src/ATen/cpu/vec256/vec256_float_neon.h
+++ b/aten/src/ATen/cpu/vec256/vec256_float_neon.h
@@ -25,6 +25,7 @@ namespace {
 //    https://bugs.llvm.org/show_bug.cgi?id=45824
 // Most likely we will do aarch32 support with inline asm.
 #if defined(__aarch64__)
+#if defined(__clang__) || (__GNUC__ > 8 || (__GNUC__ == 8 && __GNUC_MINOR__ > 3))
 
 #ifdef __BIG_ENDIAN__
 #error "Big endian is not supported."
@@ -696,4 +697,5 @@ Vec256<float> inline fmadd(const Vec256<float>& a, const Vec256<float>& b, const
 
 #endif /* defined(aarch64) */
 
+#endif
 }}}
diff --git a/aten/src/ATen/cuda/CUDAContext.cpp b/aten/src/ATen/cuda/CUDAContext.cpp
index 1751128..a090e70 100644
--- a/aten/src/ATen/cuda/CUDAContext.cpp
+++ b/aten/src/ATen/cuda/CUDAContext.cpp
@@ -24,6 +24,7 @@ void initCUDAContextVectors() {
 void initDeviceProperty(DeviceIndex device_index) {
   cudaDeviceProp device_prop;
   AT_CUDA_CHECK(cudaGetDeviceProperties(&device_prop, device_index));
+  device_prop.maxThreadsPerBlock = device_prop.maxThreadsPerBlock / 2;
   device_properties[device_index] = device_prop;
 }
 
diff --git a/aten/src/ATen/cuda/detail/KernelUtils.h b/aten/src/ATen/cuda/detail/KernelUtils.h
index 45056ab..bb2b100 100644
--- a/aten/src/ATen/cuda/detail/KernelUtils.h
+++ b/aten/src/ATen/cuda/detail/KernelUtils.h
@@ -22,7 +22,7 @@ namespace at { namespace cuda { namespace detail {
 
 
 // Use 1024 threads per block, which requires cuda sm_2x or above
-constexpr int CUDA_NUM_THREADS = 1024;
+constexpr int CUDA_NUM_THREADS = 512;
 
 // CUDA: number of blocks for threads.
 inline int GET_BLOCKS(const int64_t N) {
diff --git a/aten/src/THCUNN/common.h b/aten/src/THCUNN/common.h
index 69b7f3a..6e57b77 100644
--- a/aten/src/THCUNN/common.h
+++ b/aten/src/THCUNN/common.h
@@ -5,7 +5,7 @@
   "Some of weight/gradient/input tensors are located on different GPUs. Please move them to a single one.")
 
 // Use 1024 threads per block, which requires cuda sm_2x or above
-const int CUDA_NUM_THREADS = 1024;
+const int CUDA_NUM_THREADS = 512;
 
 // CUDA: number of blocks for threads.
 inline int GET_BLOCKS(const int64_t N)