Support CudaIpc connection within a single process (#593)

* Allow CudaIpc connection between GPUs in a single process
* Added an example of connection in a single process
* Minor interface updates

---------

Co-authored-by: Binyang Li <binyli@microsoft.com>

examples/tutorials/01-basic-concepts/Makefile

@@ -0,0 +1,22 @@
+CUDA_HOME ?= /usr/local/cuda
+ROCM_HOME ?= /opt/rocm
+
+# Check if nvcc exists, otherwise use hipcc
+ifeq ($(shell which $(CUDA_HOME)/bin/nvcc 2>/dev/null),)
+    COMPILER := $(ROCM_HOME)/bin/hipcc
+    ARCH_FLAG := -D__HIP_PLATFORM_AMD__=1
+else
+    COMPILER := $(CUDA_HOME)/bin/nvcc
+    ARCH_FLAG := -arch=native
+endif
+
+TARGET = gpu_ping_pong
+SRC = gpu_ping_pong.cu
+
+all: $(TARGET)
+
+$(TARGET): $(SRC)
+	$(COMPILER) $(ARCH_FLAG) -o $@ $< -lmscclpp
+
+clean:
+	rm -f $(TARGET)