diff --git a/setup.py b/setup.py
index 13bf9d044..93d53639c 100644
--- a/setup.py
+++ b/setup.py
@@ -215,8 +215,8 @@ def run(self):
             ],
             'ort': [
                 'onnx>=1.10.2',
-                'onnxruntime-gpu==1.10.0; python_version<"3.10" and platform_machine != "aarch64"',
-                'onnxruntime-gpu; python_version>="3.10" and platform_machine != "aarch64"',
+                'onnxruntime-gpu==1.10.0; python_version<"3.10" and platform_machine == "x86_64"',
+                'onnxruntime-gpu; python_version>="3.10" and platform_machine == "x86_64"',
             ],
             'nvidia': ['py3nvml>=0.2.6'],
             'amd': ['amdsmi'],
diff --git a/superbench/benchmarks/micro_benchmarks/cuda_decode_performance/CMakeLists.txt b/superbench/benchmarks/micro_benchmarks/cuda_decode_performance/CMakeLists.txt
index 907f616fa..1022aed3d 100644
--- a/superbench/benchmarks/micro_benchmarks/cuda_decode_performance/CMakeLists.txt
+++ b/superbench/benchmarks/micro_benchmarks/cuda_decode_performance/CMakeLists.txt
@@ -4,121 +4,120 @@
 cmake_minimum_required(VERSION 3.18)
 project(cuda_decode_performance)
 
-
- # Check architecture
- if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|arm64")
-    message(WARNING "Skipping Cuda decode Performance build. This build only supports x86_64 arch.")
- else()
-    find_package(CUDA QUIET)
-    if(CUDA_FOUND)
-    set(CMAKE_CXX_STANDARD 17)
-    set(CMAKE_CXX_STANDARD_REQUIRED ON)
-
-    set(THIRD_PARTY_SAMPLE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../../../../third_party/Video_Codec_SDK/Samples)
-    set(NVCODEC_PUBLIC_INTERFACE_DIR ${THIRD_PARTY_SAMPLE_DIR}/../Interface)
-    set(NVCODEC_UTILS_DIR ${THIRD_PARTY_SAMPLE_DIR}/Utils)
-    set(NV_CODEC_DIR ${THIRD_PARTY_SAMPLE_DIR}/NvCodec)
-    set(NV_DEC_DIR ${THIRD_PARTY_SAMPLE_DIR}/NvCodec/NvDecoder)
-
-    if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
-        find_package(PkgConfig REQUIRED)
-        pkg_check_modules(PC_AVCODEC REQUIRED IMPORTED_TARGET libavcodec)
-        pkg_check_modules(PC_AVFORMAT REQUIRED IMPORTED_TARGET libavformat)
-        pkg_check_modules(PC_AVUTIL REQUIRED IMPORTED_TARGET libavutil)
-        pkg_check_modules(PC_SWRESAMPLE REQUIRED IMPORTED_TARGET libswresample)
-
-        set(NV_FFMPEG_HDRS ${PC_AVCODEC_INCLUDE_DIRS})
-        find_library(AVCODEC_LIBRARY NAMES avcodec
-        HINTS
-        ${PC_AVCODEC_LIBDIR}
-        ${PC_AVCODEC_LIBRARY_DIRS}
-        )
-        find_library(AVFORMAT_LIBRARY NAMES avformat
-        HINTS
-        ${PC_AVFORMAT_LIBDIR}
-        ${PC_AVFORMAT_LIBRARY_DIRS}
-        )
-        find_library(AVUTIL_LIBRARY NAMES avutil
-        HINTS
-        ${PC_AVUTIL_LIBDIR}
-        ${PC_AVUTIL_LIBRARY_DIRS}
-        )
-        find_library(SWRESAMPLE_LIBRARY NAMES swresample
-        HINTS
-        ${PC_SWRESAMPLE_LIBDIR}
-        ${PC_SWRESAMPLE_LIBRARY_DIRS}
-        )
-        set(AVCODEC_LIB ${AVCODEC_LIBRARY})
-        set(AVFORMAT_LIB ${AVFORMAT_LIBRARY})
-        set(AVUTIL_LIB ${AVUTIL_LIBRARY})
-        set(SWRESAMPLE_LIB ${SWRESAMPLE_LIBRARY})
-    endif()
-
-    set(APP_SOURCES
-    ${CMAKE_CURRENT_SOURCE_DIR}/AppDecPerf.cpp
-    )
-
-    set(NV_DEC_SOURCES
-    ${NV_DEC_DIR}/NvDecoder.cpp
-    ${CMAKE_CURRENT_SOURCE_DIR}/OptimizedNvDecoder.cpp
-    )
-
-    set(NV_DEC_HDRS
-    ${NV_DEC_DIR}/NvDecoder.h
-    ${NVCODEC_PUBLIC_INTERFACE_DIR}/cuviddec.h
-    ${NVCODEC_PUBLIC_INTERFACE_DIR}/nvcuvid.h
-    ${NVCODEC_UTILS_DIR}/NvCodecUtils.h
-    ${NVCODEC_UTILS_DIR}/FFmpegDemuxer.h
-    ${CMAKE_CURRENT_SOURCE_DIR}/ThreadPoolUtils.h
-    ${CMAKE_CURRENT_SOURCE_DIR}/OptimizedNvDecoder.h
-    )
-
-    source_group( "headers" FILES ${NV_DEC_HDRS} )
-    source_group( "sources" FILES ${APP_SOURCES} ${NV_DEC_SOURCES})
-    set(CMAKE_LIBRARY_PATH "${CUDA_TOOLKIT_ROOT_DIR}/lib64/stubs;${CUDA_TOOLKIT_ROOT_DIR}/lib/stubs;${CUDA_TOOLKIT_ROOT_DIR}/lib64;${CUDA_TOOLKIT_ROOT_DIR}/lib;${CMAKE_LIBRARY_PATH}")
-    find_package(CUDA)
-    set(CUDA_HOST_COMPILER ${CMAKE_CXX_COMPILER})
-    set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS};-gencode arch=compute_50,code=\"sm_50,compute_50\")
-    if ( CMAKE_COMPILER_IS_GNUCC )
-        if(NOT "${CUDA_NVCC_FLAGS}" MATCHES "-std=c\\+\\+11" )
-        list(APPEND CUDA_NVCC_FLAGS -std=c++11)
-        endif()
-    endif()
-
-    # Check if the file exists
-    if (NOT EXISTS "/usr/local/lib/libnvcuvid.so" )
-        execute_process(
-            COMMAND sudo ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1 /usr/local/lib/libnvcuvid.so
-            RESULT_VARIABLE result
-        )
-        if(result)
-            message(FATAL_ERROR "Failed to create symbolic link for nvcuvid lib: ${result}")
-        endif()
-    endif ()
-
-    find_library(CUVID_LIB nvcuvid
-    HINTS
-    "/usr/local/lib/"
-    "${CMAKE_CURRENT_SOURCE_DIR}/../../../../third_party/Video_Codec_SDK/Lib/linux/stubs/x86_64/"
-    )
-
-    cuda_add_executable(${PROJECT_NAME} ${APP_SOURCES} ${NV_DEC_SOURCES} ${NV_DEC_HDRS})
-
-    set_target_properties(${PROJECT_NAME} PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
-
-    target_include_directories(${PROJECT_NAME} PUBLIC ${CUDA_INCLUDE_DIRS}
-    ${NVCODEC_PUBLIC_INTERFACE_DIR}
-    ${NVCODEC_UTILS_DIR}
-    ${NV_CODEC_DIR}
-    ${NV_APPDEC_COMMON_DIR}
-    ${NV_FFMPEG_HDRS}
-    ${THIRD_PARTY_SAMPLE_DIR}
-    )
-
-    target_link_libraries(${PROJECT_NAME} ${CUDA_CUDA_LIBRARY} ${CMAKE_DL_LIBS} ${CUVID_LIB} ${AVCODEC_LIB}
-    ${AVFORMAT_LIB} ${AVUTIL_LIB} ${SWRESAMPLE_LIB})
-
-    install(TARGETS ${PROJECT_NAME} RUNTIME DESTINATION bin LIBRARY DESTINATION lib)
-    endif()
-
- endif()
+# Check architecture
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|arm64")
+   message(WARNING "Skipping Cuda decode Performance build. This build only supports x86_64 arch.")
+else()
+   find_package(CUDA QUIET)
+   if(CUDA_FOUND)
+   set(CMAKE_CXX_STANDARD 17)
+   set(CMAKE_CXX_STANDARD_REQUIRED ON)
+
+   set(THIRD_PARTY_SAMPLE_DIR ${CMAKE_CURRENT_SOURCE_DIR}/../../../../third_party/Video_Codec_SDK/Samples)
+   set(NVCODEC_PUBLIC_INTERFACE_DIR ${THIRD_PARTY_SAMPLE_DIR}/../Interface)
+   set(NVCODEC_UTILS_DIR ${THIRD_PARTY_SAMPLE_DIR}/Utils)
+   set(NV_CODEC_DIR ${THIRD_PARTY_SAMPLE_DIR}/NvCodec)
+   set(NV_DEC_DIR ${THIRD_PARTY_SAMPLE_DIR}/NvCodec/NvDecoder)
+
+   if(CMAKE_SYSTEM_NAME STREQUAL "Linux")
+       find_package(PkgConfig REQUIRED)
+       pkg_check_modules(PC_AVCODEC REQUIRED IMPORTED_TARGET libavcodec)
+       pkg_check_modules(PC_AVFORMAT REQUIRED IMPORTED_TARGET libavformat)
+       pkg_check_modules(PC_AVUTIL REQUIRED IMPORTED_TARGET libavutil)
+       pkg_check_modules(PC_SWRESAMPLE REQUIRED IMPORTED_TARGET libswresample)
+
+       set(NV_FFMPEG_HDRS ${PC_AVCODEC_INCLUDE_DIRS})
+       find_library(AVCODEC_LIBRARY NAMES avcodec
+       HINTS
+       ${PC_AVCODEC_LIBDIR}
+       ${PC_AVCODEC_LIBRARY_DIRS}
+       )
+       find_library(AVFORMAT_LIBRARY NAMES avformat
+       HINTS
+       ${PC_AVFORMAT_LIBDIR}
+       ${PC_AVFORMAT_LIBRARY_DIRS}
+       )
+       find_library(AVUTIL_LIBRARY NAMES avutil
+       HINTS
+       ${PC_AVUTIL_LIBDIR}
+       ${PC_AVUTIL_LIBRARY_DIRS}
+       )
+       find_library(SWRESAMPLE_LIBRARY NAMES swresample
+       HINTS
+       ${PC_SWRESAMPLE_LIBDIR}
+       ${PC_SWRESAMPLE_LIBRARY_DIRS}
+       )
+       set(AVCODEC_LIB ${AVCODEC_LIBRARY})
+       set(AVFORMAT_LIB ${AVFORMAT_LIBRARY})
+       set(AVUTIL_LIB ${AVUTIL_LIBRARY})
+       set(SWRESAMPLE_LIB ${SWRESAMPLE_LIBRARY})
+   endif()
+
+   set(APP_SOURCES
+   ${CMAKE_CURRENT_SOURCE_DIR}/AppDecPerf.cpp
+   )
+
+   set(NV_DEC_SOURCES
+   ${NV_DEC_DIR}/NvDecoder.cpp
+   ${CMAKE_CURRENT_SOURCE_DIR}/OptimizedNvDecoder.cpp
+   )
+
+   set(NV_DEC_HDRS
+   ${NV_DEC_DIR}/NvDecoder.h
+   ${NVCODEC_PUBLIC_INTERFACE_DIR}/cuviddec.h
+   ${NVCODEC_PUBLIC_INTERFACE_DIR}/nvcuvid.h
+   ${NVCODEC_UTILS_DIR}/NvCodecUtils.h
+   ${NVCODEC_UTILS_DIR}/FFmpegDemuxer.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/ThreadPoolUtils.h
+   ${CMAKE_CURRENT_SOURCE_DIR}/OptimizedNvDecoder.h
+   )
+
+   source_group( "headers" FILES ${NV_DEC_HDRS} )
+   source_group( "sources" FILES ${APP_SOURCES} ${NV_DEC_SOURCES})
+   set(CMAKE_LIBRARY_PATH "${CUDA_TOOLKIT_ROOT_DIR}/lib64/stubs;${CUDA_TOOLKIT_ROOT_DIR}/lib/stubs;${CUDA_TOOLKIT_ROOT_DIR}/lib64;${CUDA_TOOLKIT_ROOT_DIR}/lib;${CMAKE_LIBRARY_PATH}")
+   find_package(CUDA)
+   set(CUDA_HOST_COMPILER ${CMAKE_CXX_COMPILER})
+   set(CUDA_NVCC_FLAGS ${CUDA_NVCC_FLAGS};-gencode arch=compute_50,code=\"sm_50,compute_50\")
+   if ( CMAKE_COMPILER_IS_GNUCC )
+       if(NOT "${CUDA_NVCC_FLAGS}" MATCHES "-std=c\\+\\+11" )
+       list(APPEND CUDA_NVCC_FLAGS -std=c++11)
+       endif()
+   endif()
+
+   # Check if the file exists
+   if (NOT EXISTS "/usr/local/lib/libnvcuvid.so" )
+       execute_process(
+           COMMAND sudo ln -s /usr/lib/x86_64-linux-gnu/libnvcuvid.so.1 /usr/local/lib/libnvcuvid.so
+           RESULT_VARIABLE result
+       )
+       if(result)
+           message(FATAL_ERROR "Failed to create symbolic link for nvcuvid lib: ${result}")
+       endif()
+   endif ()
+
+   find_library(CUVID_LIB nvcuvid
+   HINTS
+   "/usr/local/lib/"
+   "${CMAKE_CURRENT_SOURCE_DIR}/../../../../third_party/Video_Codec_SDK/Lib/linux/stubs/x86_64/"
+   )
+
+   cuda_add_executable(${PROJECT_NAME} ${APP_SOURCES} ${NV_DEC_SOURCES} ${NV_DEC_HDRS})
+
+   set_target_properties(${PROJECT_NAME} PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
+
+   target_include_directories(${PROJECT_NAME} PUBLIC ${CUDA_INCLUDE_DIRS}
+   ${NVCODEC_PUBLIC_INTERFACE_DIR}
+   ${NVCODEC_UTILS_DIR}
+   ${NV_CODEC_DIR}
+   ${NV_APPDEC_COMMON_DIR}
+   ${NV_FFMPEG_HDRS}
+   ${THIRD_PARTY_SAMPLE_DIR}
+   )
+
+   target_link_libraries(${PROJECT_NAME} ${CUDA_CUDA_LIBRARY} ${CMAKE_DL_LIBS} ${CUVID_LIB} ${AVCODEC_LIB}
+   ${AVFORMAT_LIB} ${AVUTIL_LIB} ${SWRESAMPLE_LIB})
+
+   install(TARGETS ${PROJECT_NAME} RUNTIME DESTINATION bin LIBRARY DESTINATION lib)
+   endif()
+
+endif()
diff --git a/third_party/Makefile b/third_party/Makefile
index 383a206dd..63ca48f36 100755
--- a/third_party/Makefile
+++ b/third_party/Makefile
@@ -19,19 +19,16 @@ NUM_MAKE_JOBS ?= $(shell nproc --ignore=2)
 .PHONY: all cuda_with_msccl cuda rocm common cuda_cutlass cuda_bandwidthTest cuda_nccl_tests cuda_perftest cuda_msccl rocm_perftest fio rocm_rccl_tests rocm_rocblas rocm_bandwidthTest gpcnet cuda_gpuburn cpu_stream cpu_hpl directx_amf_encoding_latency directx_amd rocm_hipblaslt megatron_lm megatron_deepspeed apex_rocm
 
 # Build targets.
-ifeq ($(shell uname -m), aarch64)
-all: cuda
-cuda_with_msccl: cuda cuda_msccl
-cuda: common cuda_cutlass cuda_bandwidthTest cuda_nccl_tests cuda_perftest cuda_gpuburn megatron_lm megatron_deepspeed
-cpu: common cpu_perftest
-common: cpu_stream fio
-else
 all: cuda rocm
 cuda_with_msccl: cuda cuda_msccl
 cuda: common cuda_cutlass cuda_bandwidthTest cuda_nccl_tests cuda_perftest gpcnet cuda_gpuburn megatron_lm megatron_deepspeed
 rocm: common rocm_perftest rocm_rccl_tests rocm_rocblas rocm_bandwidthTest rocm_hipblaslt megatron_deepspeed apex_rocm
-cpu: common cpu_perftest
-common: cpu_hpl cpu_stream fio
+cpu: common cpu_perftest cpu_stream
+common: fio
+
+# non aarch64 specific targets
+ifneq ($(shell uname -m), aarch64)
+common: fio cpu_hpl
 directx_amd: directx_amf_encoding_latency
 endif
 
diff --git a/third_party/stream-tests/Makefile b/third_party/stream-tests/Makefile
index 8a86c0c59..a652defd9 100644
--- a/third_party/stream-tests/Makefile
+++ b/third_party/stream-tests/Makefile
@@ -1,29 +1,28 @@
 # Copyright (c) Microsoft Corporation.
 # Licensed under the MIT license.
 
-GENFLAGS= -DSTREAM_ARRAY_SIZE=400000000
-ZEN3FLAGS= -DSTREAM_ARRAY_SIZE=400000000 -march=znver3
-ZEN4FLAGS= -DSTREAM_ARRAY_SIZE=800000000 -march=znver4
-NEO2FLAGS= -DSTREAM_ARRAY_SIZE=120000000 -mcpu=neoverse-v2
+GENFLAGS := -DSTREAM_ARRAY_SIZE=400000000
+ZEN3FLAGS := -DSTREAM_ARRAY_SIZE=400000000 -march=znver3
+ZEN4FLAGS := -DSTREAM_ARRAY_SIZE=800000000 -march=znver4
+NEO2FLAGS := -DSTREAM_ARRAY_SIZE=120000000 -mcpu=neoverse-v2
 
-GEN_OUTPUT= streamx86.exe
-ZEN3_OUTPUT= streamZen3.exe
-ZEN4_OUTPUT= streamZen4.exe
-NEO2_OUTPUT= streamNeo2.exe
+GEN_OUTPUT := streamx86.exe
+ZEN3_OUTPUT := streamZen3.exe
+ZEN4_OUTPUT := streamZen4.exe
+NEO2_OUTPUT := streamNeo2.exe
 
 ARCH := $(shell uname -m)
 
 ifeq ($(ARCH), aarch64)
-CFLAGS = -Ofast -fopenmp -DNTIMES=200
-CC=gcc
+CFLAGS := -Ofast -fopenmp -DNTIMES=200
+CC := gcc
 all: NEO2
 else
-CC= /opt/AMD/aocc-compiler-4.0.0/bin/clang
-CFLAGS= -Ofast -mcmodel=large -mavx2 -ffp-contract=fast -lomp -fopenmp -fnt-store=aggressive -DNTIMES=10
+CC := /opt/AMD/aocc-compiler-4.0.0/bin/clang
+CFLAGS := -Ofast -mcmodel=large -mavx2 -ffp-contract=fast -lomp -fopenmp -fnt-store=aggressive -DNTIMES=10
 all: ZEN3 ZEN4 X86
 endif
 
-
 ZEN3: stream.c
 	$(CC) $(CFLAGS) $(ZEN3FLAGS) stream.c -o $(ZEN3_OUTPUT)
 ZEN4: