diff --git a/.github/workflows/loongarch64.yml b/.github/workflows/loongarch64.yml
index b23c81f96..da7f6c9a0 100644
--- a/.github/workflows/loongarch64.yml
+++ b/.github/workflows/loongarch64.yml
@@ -33,10 +33,8 @@ jobs:
 
       - name: Install APT deps
         run: |
-          sudo add-apt-repository ppa:savoury1/virtualisation
           sudo apt-get update
-          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache \
-          qemu-user-static
+          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache
 
       - name: Download and install loongarch64-toolchain
         run: |
@@ -44,6 +42,20 @@ jobs:
           #wget https://github.com/loongson/build-tools/releases/download/2023.08.08/CLFS-loongarch64-8.1-x86_64-cross-tools-gcc-glibc.tar.xz
           tar -xf CLFS-loongarch64-8.1-x86_64-cross-tools-gcc-glibc.tar.xz -C /opt
 
+      - name: Checkout qemu
+        uses: actions/checkout@v3
+        with:
+          repository: qemu/qemu
+          path: qemu
+          ref: master
+
+      - name: Install qemu
+        run: |
+          cd qemu
+          ./configure --prefix=$GITHUB_WORKSPACE/qemu-install --target-list=loongarch64-linux-user --disable-system --static
+          make -j$(nproc)
+          make install
+
       - name: Set env
         run: |
           echo "LD_LIBRARY_PATH=/opt/cross-tools/target/usr/lib64:/opt/cross-tools/loongarch64-unknown-linux-gnu/lib64:$LD_LIBRARY_PATH" >> $GITHUB_ENV
@@ -76,45 +88,46 @@ jobs:
 
       - name: Test
         run: |
-          qemu-loongarch64-static ./utest/openblas_utest
-          qemu-loongarch64-static ./utest/openblas_utest_ext
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xscblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xdcblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xccblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xzcblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xscblat2 < ./ctest/sin2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xdcblat2 < ./ctest/din2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xccblat2 < ./ctest/cin2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xzcblat2 < ./ctest/zin2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xscblat3 < ./ctest/sin3
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xdcblat3 < ./ctest/din3
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xccblat3 < ./ctest/cin3
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xzcblat3 < ./ctest/zin3
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/sblat1
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/dblat1
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/cblat1
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/zblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/sblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/dblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/cblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/zblat1
+          export PATH=$GITHUB_WORKSPACE/qemu-install/bin/:$PATH
+          qemu-loongarch64 ./utest/openblas_utest
+          qemu-loongarch64 ./utest/openblas_utest_ext
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xscblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xdcblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xccblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xzcblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xscblat2 < ./ctest/sin2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xdcblat2 < ./ctest/din2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xccblat2 < ./ctest/cin2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xzcblat2 < ./ctest/zin2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xscblat3 < ./ctest/sin3
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xdcblat3 < ./ctest/din3
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xccblat3 < ./ctest/cin3
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xzcblat3 < ./ctest/zin3
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/sblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/dblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/cblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/zblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/sblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/dblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/cblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/zblat1
           rm -f ./test/?BLAT2.SUMM
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/sblat2 < ./test/sblat2.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/dblat2 < ./test/dblat2.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/cblat2 < ./test/cblat2.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/zblat2 < ./test/zblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/sblat2 < ./test/sblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/dblat2 < ./test/dblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/cblat2 < ./test/cblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/zblat2 < ./test/zblat2.dat
           rm -f ./test/?BLAT2.SUMM
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/sblat2 < ./test/sblat2.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/dblat2 < ./test/dblat2.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/cblat2 < ./test/cblat2.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/zblat2 < ./test/zblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/sblat2 < ./test/sblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/dblat2 < ./test/dblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/cblat2 < ./test/cblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/zblat2 < ./test/zblat2.dat
           rm -f ./test/?BLAT3.SUMM
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/sblat3 < ./test/sblat3.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/dblat3 < ./test/dblat3.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/cblat3 < ./test/cblat3.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/zblat3 < ./test/zblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/sblat3 < ./test/sblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/dblat3 < ./test/dblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/cblat3 < ./test/cblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/zblat3 < ./test/zblat3.dat
           rm -f ./test/?BLAT3.SUMM
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/sblat3 < ./test/sblat3.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/dblat3 < ./test/dblat3.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/cblat3 < ./test/cblat3.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/zblat3 < ./test/zblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/sblat3 < ./test/sblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/dblat3 < ./test/dblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/cblat3 < ./test/cblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/zblat3 < ./test/zblat3.dat
diff --git a/.github/workflows/loongarch64_clang.yml b/.github/workflows/loongarch64_clang.yml
index b32e4d731..d08e56f62 100644
--- a/.github/workflows/loongarch64_clang.yml
+++ b/.github/workflows/loongarch64_clang.yml
@@ -34,18 +34,30 @@ jobs:
 
       - name: Install APT deps
         run: |
-          sudo add-apt-repository ppa:savoury1/virtualisation
           sudo apt-get update
-          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache \
-          qemu-user-static
+          sudo apt-get install autoconf automake autotools-dev ninja-build make ccache
 
       - name: Download and install loongarch64-toolchain
         run: |
-          wget http://ftp.loongnix.cn/toolchain/llvm/llvm8/clang+llvm_8.0.1-6_amd64-linux-gnu_debian-10.tar.gz
-          wget http://ftp.loongnix.cn/toolchain/gcc/release/loongarch/gcc8/loongson-gnu-toolchain-8.3-x86_64-loongarch64-linux-gnu-rc1.3.tar.xz
+          wget https://github.com/XiWeiGu/loongarch64_toolchain/releases/download/V0.1/clang+llvm_8.0.1-6_amd64-linux-gnu_debian-10.tar.gz
+          wget https://github.com/XiWeiGu/loongarch64_toolchain/releases/download/V0.1/loongson-gnu-toolchain-8.3-x86_64-loongarch64-linux-gnu-rc1.3.tar.xz
           tar -xf clang+llvm_8.0.1-6_amd64-linux-gnu_debian-10.tar.gz -C /opt
           tar -xf loongson-gnu-toolchain-8.3-x86_64-loongarch64-linux-gnu-rc1.3.tar.xz -C /opt
 
+      - name: Checkout qemu
+        uses: actions/checkout@v3
+        with:
+          repository: qemu/qemu
+          path: qemu
+          ref: master
+
+      - name: Install qemu
+        run: |
+          cd qemu
+          ./configure --prefix=$GITHUB_WORKSPACE/qemu-install --target-list=loongarch64-linux-user --disable-system --static
+          make -j$(nproc)
+          make install
+
       - name: Set env
         run: |
           echo "PATH=$GITHUB_WORKSPACE:/opt/clang+llvm_8.0.1-6_amd64-linux-gnu_debian-10/bin:/opt/loongson-gnu-toolchain-8.3-x86_64-loongarch64-linux-gnu-rc1.3/bin:$PATH" >> $GITHUB_ENV
@@ -77,46 +89,47 @@ jobs:
 
       - name: Test
         run: |
-          qemu-loongarch64-static ./utest/openblas_utest
-          qemu-loongarch64-static ./utest/openblas_utest_ext
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xscblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xdcblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xccblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xzcblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xscblat2 < ./ctest/sin2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xdcblat2 < ./ctest/din2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xccblat2 < ./ctest/cin2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xzcblat2 < ./ctest/zin2
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xscblat3 < ./ctest/sin3
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xdcblat3 < ./ctest/din3
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xccblat3 < ./ctest/cin3
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./ctest/xzcblat3 < ./ctest/zin3
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/sblat1
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/dblat1
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/cblat1
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/zblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/sblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/dblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/cblat1
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/zblat1
+          export PATH=$GITHUB_WORKSPACE/qemu-install/bin/:$PATH
+          qemu-loongarch64 ./utest/openblas_utest
+          qemu-loongarch64 ./utest/openblas_utest_ext
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xscblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xdcblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xccblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xzcblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xscblat2 < ./ctest/sin2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xdcblat2 < ./ctest/din2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xccblat2 < ./ctest/cin2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xzcblat2 < ./ctest/zin2
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xscblat3 < ./ctest/sin3
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xdcblat3 < ./ctest/din3
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xccblat3 < ./ctest/cin3
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./ctest/xzcblat3 < ./ctest/zin3
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/sblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/dblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/cblat1
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/zblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/sblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/dblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/cblat1
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/zblat1
           rm -f ./test/?BLAT2.SUMM
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/sblat2 < ./test/sblat2.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/dblat2 < ./test/dblat2.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/cblat2 < ./test/cblat2.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/zblat2 < ./test/zblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/sblat2 < ./test/sblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/dblat2 < ./test/dblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/cblat2 < ./test/cblat2.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/zblat2 < ./test/zblat2.dat
           rm -f ./test/?BLAT2.SUMM
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/sblat2 < ./test/sblat2.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/dblat2 < ./test/dblat2.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/cblat2 < ./test/cblat2.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/zblat2 < ./test/zblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/sblat2 < ./test/sblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/dblat2 < ./test/dblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/cblat2 < ./test/cblat2.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/zblat2 < ./test/zblat2.dat
           rm -f ./test/?BLAT3.SUMM
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/sblat3 < ./test/sblat3.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/dblat3 < ./test/dblat3.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/cblat3 < ./test/cblat3.dat
-          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64-static ./test/zblat3 < ./test/zblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/sblat3 < ./test/sblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/dblat3 < ./test/dblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/cblat3 < ./test/cblat3.dat
+          OPENBLAS_NUM_THREADS=1 OMP_NUM_THREADS=1 qemu-loongarch64 ./test/zblat3 < ./test/zblat3.dat
           rm -f ./test/?BLAT3.SUMM
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/sblat3 < ./test/sblat3.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/dblat3 < ./test/dblat3.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/cblat3 < ./test/cblat3.dat
-          OPENBLAS_NUM_THREADS=2 qemu-loongarch64-static ./test/zblat3 < ./test/zblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/sblat3 < ./test/sblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/dblat3 < ./test/dblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/cblat3 < ./test/cblat3.dat
+          OPENBLAS_NUM_THREADS=2 qemu-loongarch64 ./test/zblat3 < ./test/zblat3.dat
 
diff --git a/cmake/cc.cmake b/cmake/cc.cmake
index 9dce9ea35..775239e1c 100644
--- a/cmake/cc.cmake
+++ b/cmake/cc.cmake
@@ -6,6 +6,7 @@ include(CheckCCompilerFlag)
 
 if (${CMAKE_C_COMPILER_ID} MATCHES "IntelLLVM")
   set(CCOMMON_OPT "${CCOMMON_OPT} -fp-model=consistent")
+  set(GCC_VERSION 100)
 endif ()
 
 if (${CMAKE_C_COMPILER_ID} STREQUAL "GNU" OR ${CMAKE_C_COMPILER_ID} STREQUAL "LSB" OR ${CMAKE_C_COMPILER_ID} MATCHES "Clang")
diff --git a/cmake/fc.cmake b/cmake/fc.cmake
index 69246385c..8798ce8b4 100644
--- a/cmake/fc.cmake
+++ b/cmake/fc.cmake
@@ -117,12 +117,12 @@ if (${F_COMPILER} STREQUAL "GFORTRAN" OR ${F_COMPILER} STREQUAL "F95" OR CMAKE_F
   endif ()
 endif ()
 
-if (${F_COMPILER} STREQUAL "INTEL")
+if (${F_COMPILER} STREQUAL "INTEL" OR CMAKE_Fortran_COMPILER_ID MATCHES "Intel")
   set(CCOMMON_OPT "${CCOMMON_OPT} -DF_INTERFACE_INTEL")
   if (INTERFACE64)
     set(FCOMMON_OPT "${FCOMMON_OPT} -i8")
   endif ()
-  set(FCOMMON_OPT "${FCOMMON_OPT} -recursive")
+  set(FCOMMON_OPT "${FCOMMON_OPT} -recursive -fp-model=consistent")
   if (USE_OPENMP)
     set(FCOMMON_OPT "${FCOMMON_OPT} -openmp")
   endif ()
diff --git a/cpuid_x86.c b/cpuid_x86.c
index 40735c563..e157232cf 100644
--- a/cpuid_x86.c
+++ b/cpuid_x86.c
@@ -2525,6 +2525,7 @@ int get_coretype(void){
       case 0x7:
         switch (exmodel) {
         case 5:
+	case 6:
           if (support_avx2())
             return CORE_ZEN;
           else
diff --git a/driver/level3/level3_thread.c b/driver/level3/level3_thread.c
index 9fec7afca..ddb39abd6 100644
--- a/driver/level3/level3_thread.c
+++ b/driver/level3/level3_thread.c
@@ -570,6 +570,8 @@ static int gemm_driver(blas_arg_t *args, BLASLONG *range_m, BLASLONG
   InitializeCriticalSection((PCRITICAL_SECTION)&level3_lock);
 #else
   static pthread_mutex_t  level3_lock    = PTHREAD_MUTEX_INITIALIZER;
+  static pthread_cond_t  level3_wakeup    = PTHREAD_COND_INITIALIZER;
+  volatile static BLASLONG CPU_AVAILABLE = MAX_CPU_NUMBER;
 #endif
 
   blas_arg_t newarg;
@@ -639,6 +641,12 @@ static int gemm_driver(blas_arg_t *args, BLASLONG *range_m, BLASLONG
   EnterCriticalSection((PCRITICAL_SECTION)&level3_lock);
 #else
   pthread_mutex_lock(&level3_lock);
+  while(CPU_AVAILABLE < nthreads) { 
+    pthread_cond_wait(&level3_wakeup, &level3_lock);
+  } 
+  CPU_AVAILABLE -= nthreads;
+  WMB;
+  pthread_mutex_unlock(&level3_lock);
 #endif
 
 #ifdef USE_ALLOC_HEAP
@@ -783,6 +791,10 @@ static int gemm_driver(blas_arg_t *args, BLASLONG *range_m, BLASLONG
 #elif defined(OS_WINDOWS)
   LeaveCriticalSection((PCRITICAL_SECTION)&level3_lock);
 #else
+  pthread_mutex_lock(&level3_lock);
+  CPU_AVAILABLE += nthreads;
+  WMB;
+  pthread_cond_signal(&level3_wakeup);
   pthread_mutex_unlock(&level3_lock);
 #endif
 
diff --git a/driver/others/dynamic.c b/driver/others/dynamic.c
index e3f905265..1f7142009 100644
--- a/driver/others/dynamic.c
+++ b/driver/others/dynamic.c
@@ -927,6 +927,7 @@ static gotoblas_t *get_coretype(void){
 	case 0x7:
       switch (exmodel) {
       case 5:
+      case 6:
         if (support_avx2())
           return &gotoblas_ZEN;
         else
diff --git a/interface/CMakeLists.txt b/interface/CMakeLists.txt
index 05b893b8c..449072bae 100644
--- a/interface/CMakeLists.txt
+++ b/interface/CMakeLists.txt
@@ -137,7 +137,7 @@ endif ()
 foreach (float_type ${FLOAT_TYPES})
 
   if (${float_type} STREQUAL "COMPLEX" OR ${float_type} STREQUAL "ZCOMPLEX")
-    GenerateNamedObjects("zaxpy.c" "" "axpyc" ${CBLAS_FLAG} "" "" false ${float_type})
+    GenerateNamedObjects("zaxpy.c" "CONJ" "axpyc" ${CBLAS_FLAG} "" "" false ${float_type})
 
     GenerateNamedObjects("zger.c" "" "geru" ${CBLAS_FLAG} "" "" false ${float_type})
     GenerateNamedObjects("zger.c" "CONJ" "gerc" ${CBLAS_FLAG} "" "" false ${float_type})
diff --git a/kernel/arm64/dot_kernel_sve.c b/kernel/arm64/dot_kernel_sve.c
index 16f4cd537..bc9975214 100644
--- a/kernel/arm64/dot_kernel_sve.c
+++ b/kernel/arm64/dot_kernel_sve.c
@@ -108,7 +108,12 @@ dot_kernel_sve(BLASLONG n, FLOAT* x, FLOAT* y)
         [N_] "r" (n),
         [X_] "r" (x),
         [Y_] "r" (y)
-      :);
+      : "cc",
+        "memory",
+        "x0", "x1", "x2", "x3", "x4", "x5", "x6", "x7",
+        "x8", "x9", "x10", "x11", "x12", "x13", "d1", 
+        "z0", "z1"
+  );
 
   return ret;
 }
diff --git a/kernel/arm64/zdot_thunderx2t99.c b/kernel/arm64/zdot_thunderx2t99.c
index 6f65e5cfd..d48392412 100644
--- a/kernel/arm64/zdot_thunderx2t99.c
+++ b/kernel/arm64/zdot_thunderx2t99.c
@@ -292,7 +292,10 @@ static void zdot_compute(BLASLONG n, FLOAT *x, BLASLONG inc_x, FLOAT *y, BLASLON
 	: "cc",
 	  "memory",
 	  "x0", "x1", "x2", "x3", "x4", "x5",
-	  "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7"
+	  "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7",
+	  "v16", "v17", "v18", "v19", "v20", "v21", "v22",
+	  "v23", "v24", "v25", "v26", "v27", "v28", "v29",
+	  "v30", "v31"
 	);
 
 	cf=OPENBLAS_MAKE_COMPLEX_FLOAT(dotr, doti);
diff --git a/kernel/x86_64/tobf16.c b/kernel/x86_64/tobf16.c
index a88fdcc2e..e8e228955 100644
--- a/kernel/x86_64/tobf16.c
+++ b/kernel/x86_64/tobf16.c
@@ -144,10 +144,11 @@ void CNAME(BLASLONG n, FLOAT_TYPE * in, BLASLONG inc_in, bfloat16 * out, BLASLON
     if (inc_in == 0 || inc_out == 0 || n <= 100000) {
         nthreads = 1;
     } else {
+        nthreads = num_cpu_avail(1);
         if (n/100000 < 100) {
-            nthreads = 4;
-        } else {
-            nthreads = 16;
+            nthreads = MAX(nthreads,4);
+//        } else {
+//            nthreads = MAX(nthreads,16);
         }
     }
 
diff --git a/lapack-netlib/TESTING/EIG/ddrvst.f b/lapack-netlib/TESTING/EIG/ddrvst.f
index 805fd8271..311684238 100644
--- a/lapack-netlib/TESTING/EIG/ddrvst.f
+++ b/lapack-netlib/TESTING/EIG/ddrvst.f
@@ -2772,7 +2772,7 @@ c           LIWEDC = 12
                      RESULT( NTEST ) = ULPINV
                      RESULT( NTEST+1 ) = ULPINV
                      RESULT( NTEST+2 ) = ULPINV
-                     GO TO 700
+                     GO TO 1750
                   END IF
                END IF
 *
@@ -2797,13 +2797,13 @@ c           LIWEDC = 12
                      RETURN
                   ELSE
                      RESULT( NTEST ) = ULPINV
-                     GO TO 700
+                     GO TO 1750
                   END IF
                END IF
 *
                IF( M3.EQ.0 .AND. N.GT.0 ) THEN
                   RESULT( NTEST ) = ULPINV
-                  GO TO 700
+                  GO TO 1750
                END IF
 *
 *              Do test 78 (or +54)
@@ -2819,6 +2819,8 @@ c           LIWEDC = 12
      $                           MAX( UNFL, TEMP3*ULP )
 *
                CALL DLACPY( ' ', N, N, V, LDU, A, LDA )
+*
+ 1750       CONTINUE
 *
  1720       CONTINUE
 *
diff --git a/lapack-netlib/TESTING/EIG/dlahd2.f b/lapack-netlib/TESTING/EIG/dlahd2.f
index a6c65635f..cccbd2537 100644
--- a/lapack-netlib/TESTING/EIG/dlahd2.f
+++ b/lapack-netlib/TESTING/EIG/dlahd2.f
@@ -534,8 +534,8 @@
      $      / ' 2: norm( I - Q'' Q )   / ( m ulp )',
      $      / ' 3: norm( I - PT PT'' )   / ( n ulp )',
      $      / ' 4: norm( Y - Q'' C )   / ( norm(Y) max(m,nrhs) ulp )' )
- 9968 FORMAT( / ' Tests performed:  See sdrvst.f' )
- 9967 FORMAT( / ' Tests performed:  See cdrvst.f' )
+ 9968 FORMAT( / ' Tests performed:  See ddrvst.f' )
+ 9967 FORMAT( / ' Tests performed:  See zdrvst.f' )
 *
 *     End of DLAHD2
 *
diff --git a/lapack-netlib/TESTING/EIG/sdrvst.f b/lapack-netlib/TESTING/EIG/sdrvst.f
index be6d33cee..2d02f54ed 100644
--- a/lapack-netlib/TESTING/EIG/sdrvst.f
+++ b/lapack-netlib/TESTING/EIG/sdrvst.f
@@ -2772,7 +2772,7 @@ c           LIWEDC = 12
                      RESULT( NTEST ) = ULPINV
                      RESULT( NTEST+1 ) = ULPINV
                      RESULT( NTEST+2 ) = ULPINV
-                     GO TO 700
+                     GO TO 1750
                   END IF
                END IF
 *
@@ -2797,13 +2797,13 @@ c           LIWEDC = 12
                      RETURN
                   ELSE
                      RESULT( NTEST ) = ULPINV
-                     GO TO 700
+                     GO TO 1750
                   END IF
                END IF
 *
                IF( M3.EQ.0 .AND. N.GT.0 ) THEN
                   RESULT( NTEST ) = ULPINV
-                  GO TO 700
+                  GO TO 1750
                END IF
 *
 *              Do test 78 (or +54)
@@ -2819,6 +2819,8 @@ c           LIWEDC = 12
      $                           MAX( UNFL, TEMP3*ULP )
 *
                CALL SLACPY( ' ', N, N, V, LDU, A, LDA )
+*
+ 1750       CONTINUE
 *
  1720       CONTINUE
 *
diff --git a/utest/test_fork.c b/utest/test_fork.c
index 558026031..44a619349 100644
--- a/utest/test_fork.c
+++ b/utest/test_fork.c
@@ -33,6 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #include <sys/types.h>
 #include <sys/wait.h>
+#include <errno.h>
 #include <cblas.h>
 #include "openblas_utest.h"
 
@@ -41,7 +42,7 @@ static void* xmalloc(size_t n)
     void* tmp;
     tmp = malloc(n);
     if (tmp == NULL) {
-        fprintf(stderr, "You are about to die\n");
+        fprintf(stderr, "Failed to allocate memory for the testcase.\n");
         exit(1);
     } else {
         return tmp;
@@ -103,6 +104,7 @@ exit(0);
 
     fork_pid = fork();
     if (fork_pid == -1) {
+        perror("fork");
         CTEST_ERR("Failed to fork process.");
     } else if (fork_pid == 0) {
         // Compute a DGEMM product in the child process to check that the
@@ -113,7 +115,8 @@ exit(0);
         // recursively
         fork_pid_nested = fork();
         if (fork_pid_nested == -1) {
-            CTEST_ERR("Failed to fork process.");
+            perror("fork");
+            CTEST_ERR("Failed to fork nested process.");
             exit(1);
         } else if (fork_pid_nested == 0) {
             check_dgemm(a, b, d, c, n);
diff --git a/utest/test_post_fork.c b/utest/test_post_fork.c
index 6d640aebb..d6e87f2ba 100644
--- a/utest/test_post_fork.c
+++ b/utest/test_post_fork.c
@@ -33,6 +33,7 @@ USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 
 #include <sys/types.h>
 #include <sys/wait.h>
+#include <errno.h>
 #include <cblas.h>
 #ifdef USE_OPENMP
 #include <omp.h>
@@ -44,7 +45,7 @@ static void* xmalloc(size_t n)
     void* tmp;
     tmp = malloc(n);
     if (tmp == NULL) {
-        fprintf(stderr, "You are about to die\n");
+        fprintf(stderr, "Failed to allocate memory for the test payload.\n");
         exit(1);
     } else {
         return tmp;
@@ -114,7 +115,11 @@ exit(0);
 
         fork_pid = fork();
         if (fork_pid == -1) {
-            CTEST_ERR("Failed to fork process.");
+            perror("fork");
+            CTEST_ERR("Failed to fork subprocesses in a loop.");
+#ifdef USE_OPENMP
+            CTEST_ERR("Number of OpenMP threads was %d in this attempt.",i);
+#endif
         } else if (fork_pid == 0) {
             // Just pretend to do something, e.g. call `uname`, then exit
             exit(0);