Merge pull request #13 from PedramAlizadeh/documentation

Added documentation, pthread linking, and Hex value printing for better debugging

CHANGELOG.md

 # Changelog for TransferBench
 
+## v1.14
+### Added
+- Added documentation
+- Added pthread linking in src/Makefile and CMakeLists.txt
+- Added printing off the hex value of the floats for output and reference
+ 
 ## v1.13
 ### Added
 - Added support for cmake

CMakeLists.txt

@@ -6,9 +6,9 @@ else()
 endif()
 cmake_minimum_required(VERSION 3.5)
 project(TransferBench CXX)
-set( CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O3 -L${ROCM_PATH}/hsa/lib")
-include_directories(${ROCM_PATH}/hsa/include)
-link_libraries(numa hsa-runtime64)
+set( CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O3 -L${ROCM_PATH}/lib")
+include_directories(${ROCM_PATH}/include)
+link_libraries(numa hsa-runtime64 pthread)
 set (CMAKE_RUNTIME_OUTPUT_DIRECTORY ..)
 add_executable(TransferBench src/TransferBench.cpp)
 target_include_directories(TransferBench PRIVATE src/include)

docs/introduction.rst

+*******************************************
+Welcome to TransferBench's documentation!
+*******************************************
+TransferBench is a simple utility capable of benchmarking simultaneous transfers between user-specified devices (CPUs/GPUs).
+A Transfer is defined as a single operation where an executor reads and adds together values from source (SRC) memory locations, then writes the sum to destination (DST) memory locations. This simplifies to a simple copy operation when dealing with single SRC/DST.
+
+The user has control over the SRC and DST memory locations by indicating memory type followed by the device index. TransferBench supports coarse-grained pinned host memory, unpinned host memory, fine-grained host memory, coarse-grained global device memory, fine-grained global device memory, and null memory (for an empty transfer). In addition, the user can determine the size of the transfer (number of bytes to copy) for their tests.
+
+The executor of the transfer can also be specified by the user. The options are CPU, kernel-based GPU, and SDMA-based GPU (DMA) executors. TransferBench also provides the option to choose the number of sub-executors. In case of a CPU executor this argument specifies the number of CPU threads, while for a GPU executor it defines the number of compute units (CU). If DMA is specified as the executor, the sub-executor argument determines the number of streams to be used.
+
+For more examples, please refer to the example.cfg file in the examples folder.

src/Makefile

@@ -3,12 +3,12 @@ ROCM_PATH ?= /opt/rocm
 HIPCC=$(ROCM_PATH)/bin/hipcc
 
 EXE=TransferBench
-CXXFLAGS = -O3 -I. -Iinclude -I$(ROCM_PATH)/hsa/include -lnuma -L$(ROCM_PATH)/hsa/lib -lhsa-runtime64
-
+CXXFLAGS = -O3 -I. -Iinclude -I$(ROCM_PATH)/include -lnuma -L$(ROCM_PATH)/lib -lhsa-runtime64
+LDFLAGS    += -lpthread
 all: $(EXE)
 
 $(EXE): $(EXE).cpp $(shell find -regex ".*\.\hpp")
-	$(HIPCC) $(CXXFLAGS) $< -o ../$@
+	$(HIPCC) $(CXXFLAGS) $< -o ../$@ $(LDFLAGS)
 
 clean:
 	rm -f *.o ../$(EXE)

src/TransferBench.cpp

@@ -1452,8 +1452,7 @@ void Transfer::ValidateDst(EnvVars const& ev)
     {
       if (reference[i] != output[i])
       {
-        printf("\n[ERROR] Destination array %d value at index %lu (%.3f) does not match expected value (%.3f)\n",
-               dstIdx, i, output[i], reference[i]);
+	printf("\n[ERROR] Destination array %d value at index %lu (%.3f) [%X] does not match expected value (%.3f) [%X]\n", dstIdx, i, output[i], *(unsigned int*)&output[i], reference[i], *(unsigned int*)&reference[i]);
         printf("[ERROR] Failed Transfer details: #%d: %s -> [%c%d:%d] -> %s\n",
                this->transferIndex,
                this->SrcToStr().c_str(),

src/include/EnvVars.hpp

@@ -28,7 +28,7 @@ THE SOFTWARE.
 #include <time.h>
 #include "Kernels.hpp"
 
-#define TB_VERSION "1.13"
+#define TB_VERSION "1.14"
 
 extern char const MemTypeStr[];
 extern char const ExeTypeStr[];