diff --git a/.github/workflows/ubuntu18.yml b/.github/workflows/ubuntu18.yml
index 23850cc..bf1f2d0 100644
--- a/.github/workflows/ubuntu18.yml
+++ b/.github/workflows/ubuntu18.yml
@@ -9,9 +9,10 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - {cxx: -DCMAKE_CXX_COMPILER=g++-5}
-          - {cxx: -DCMAKE_CXX_COMPILER=g++-6}
-          - {cxx: } # default compiler 7
+          - {cxx: -DCMAKE_CXX_COMPILER=g++-5, arch:                         }
+          - {cxx: -DCMAKE_CXX_COMPILER=g++-6, arch:                         }
+          - {cxx:                           , arch:                         } # default=gcc7
+          - {cxx:                           , arch: -DCMAKE_CXX_FLAGS="-m32"} # default=gcc7
     steps:
       - uses: actions/checkout@v2
       - name: Setup cmake
@@ -20,12 +21,13 @@ jobs:
           cmake-version: '3.9.x'
       - name: Install older compilers
         run: |
+          sudo -E dpkg --add-architecture i386
           sudo -E apt-get update
-          sudo -E apt-get install -y --force-yes g++-5 g++-6
+          sudo -E apt-get install -y --force-yes g++-5 g++-6 g++-5-multilib g++-6-multilib g++-multilib linux-libc-dev:i386 libc6:i386 libc6-dev:i386 libc6-dbg:i386
       - name: Prepare build dir
         run: mkdir build
       - name: Configure
-        run: cd build && cmake ${{matrix.cxx}} -DFASTFLOAT_TEST=ON ..
+        run: cd build && cmake ${{matrix.cxx}} ${{matrix.arch}} -DFASTFLOAT_TEST=ON ..
       - name: Build
         run: cmake --build build
       - name: Run basic tests
diff --git a/.github/workflows/ubuntu20.yml b/.github/workflows/ubuntu20.yml
index 68aa1ee..451c9b4 100644
--- a/.github/workflows/ubuntu20.yml
+++ b/.github/workflows/ubuntu20.yml
@@ -9,8 +9,9 @@ jobs:
       fail-fast: false
       matrix:
         include:
-          - {cxx: -DCMAKE_CXX_COMPILER=g++-8}
-          - {cxx: } # default compiler 9
+          - {cxx: -DCMAKE_CXX_COMPILER=g++-8, arch:                         }
+          - {cxx:                           , arch:                         } # default=gcc9
+          - {cxx:                           , arch: -DCMAKE_CXX_FLAGS="-m32"} # default=gcc9
     steps:
       - uses: actions/checkout@v2
       - name: Setup cmake
@@ -19,12 +20,13 @@ jobs:
           cmake-version: '3.9.x'
       - name: install older compilers
         run: |
+          sudo -E dpkg --add-architecture i386
           sudo -E apt-get update
-          sudo -E apt-get install -y --force-yes g++-8 g++-7
-      - name: Prepare build dir
+          sudo -E apt-get install -y g++-8 g++-8-multilib g++-multilib linux-libc-dev:i386 libc6:i386 libc6-dev:i386 libc6-dbg:i386
+    - name: Prepare build dir
         run: mkdir build
       - name: Configure
-        run: cd build && cmake ${{matrix.cxx}} -DFASTFLOAT_TEST=ON ..
+        run: cd build && cmake ${{matrix.cxx}} ${{matrix.arch}} -DFASTFLOAT_TEST=ON ..
       - name: Build
         run: cmake --build build
       - name: Run basic tests
diff --git a/.github/workflows/vs16-ci.yml b/.github/workflows/vs16-ci.yml
index 5a7071a..4f237fd 100644
--- a/.github/workflows/vs16-ci.yml
+++ b/.github/workflows/vs16-ci.yml
@@ -6,6 +6,12 @@ jobs:
   ci:
     name: windows-vs16
     runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - vs: VS16Win64
+          - vs: VS16Win32
     steps:
     - uses: actions/checkout@v2
     - name: 'Run CMake with VS16'
@@ -16,9 +22,9 @@ jobs:
         buildDirectory: "${{ github.workspace }}/../../_temp/windows"
         cmakeBuildType: Release
         buildWithCMake: true
-        cmakeGenerator: VS16Win64
+        cmakeGenerator: ${{matrix.vs}}
         cmakeAppendedArgs: -DFASTFLOAT_TEST=ON
-        buildWithCMakeArgs:  --config Release  
+        buildWithCMakeArgs:  --config Release
 
     - name: 'Run CTest'
       run: ctest -C Release    --output-on-failure  -R basictest
diff --git a/.github/workflows/vs16-clang-ci.yml b/.github/workflows/vs16-clang-ci.yml
index 08afa25..2bf8e2b 100644
--- a/.github/workflows/vs16-clang-ci.yml
+++ b/.github/workflows/vs16-clang-ci.yml
@@ -6,6 +6,12 @@ jobs:
   ci:
     name: windows-vs16
     runs-on: windows-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        include:
+          - vs: VS16Win64
+          - vs: VS16Win32
     steps:
     - uses: actions/checkout@v2
     - name: 'Run CMake with VS16'
@@ -16,7 +22,7 @@ jobs:
         buildDirectory: "${{ github.workspace }}/../../_temp/windows"
         cmakeBuildType: Release   
         buildWithCMake: true
-        cmakeGenerator: VS16Win64 
+        cmakeGenerator: ${{matrix.vs}}
         cmakeAppendedArgs:  -T ClangCL -DFASTFLOAT_TEST=ON
         buildWithCMakeArgs: --config Release  
         
diff --git a/.gitignore b/.gitignore
new file mode 100644
index 0000000..a5a3ead
--- /dev/null
+++ b/.gitignore
@@ -0,0 +1,2 @@
+build/*
+Testing/*
\ No newline at end of file
diff --git a/include/fast_float/float_common.h b/include/fast_float/float_common.h
index 60e5f6e..a98a54b 100644
--- a/include/fast_float/float_common.h
+++ b/include/fast_float/float_common.h
@@ -139,15 +139,31 @@ fastfloat_really_inline value128 full_multiplication(uint64_t value1,
   return answer;
 }
 
-#else
+#else // gcc
 
 // compute value1 * value2
-fastfloat_really_inline value128 full_multiplication(uint64_t value1,
-                                                     uint64_t value2) {
+fastfloat_really_inline value128 full_multiplication(uint64_t a,
+                                                     uint64_t b) {
   value128 answer;
-  __uint128_t r = ((__uint128_t)value1) * value2;
+#if  defined(__i386) || defined(__i386__) || defined(_M_IX86) || defined(__arm__)
+  static constexpr const uint64_t lo32 = 0xffffffffu;
+  // https://stackoverflow.com/questions/28868367/getting-the-high-part-of-64-bit-integer-multiplication
+  uint64_t a_lo = a & lo32;
+  uint64_t a_hi = a >> 32;
+  uint64_t b_lo = b & lo32;
+  uint64_t b_hi = b >> 32;
+  uint64_t ab_hi =  a_hi * b_hi;
+  uint64_t ab_mid = a_hi * b_lo;
+  uint64_t ba_mid = b_hi * a_lo;
+  uint64_t ab_lo =  a_lo * b_lo;
+  uint64_t carry_bit = ((ab_mid & lo32) + (ba_mid & lo32) + (ab_lo >> 32)) >> 32;
+  answer.high = ab_hi + (ab_mid >> 32) + (ba_mid >> 32) + carry_bit;
+  answer.low = ab_lo + (ab_mid & lo32) + (ba_mid & lo32);
+#else // if defined(__x86_64) || defined(__x86_64__) || defined(__amd64) || defined(_M_X64) || defined(__aarch64__) || defined(_M_ARM64)
+  __uint128_t r = ((__uint128_t)a) * b;
   answer.low = uint64_t(r);
   answer.high = uint64_t(r >> 64);
+#endif
   return answer;
 }
 
diff --git a/tests/long_random64.cpp b/tests/long_random64.cpp
index b8a6b05..47c4d0c 100644
--- a/tests/long_random64.cpp
+++ b/tests/long_random64.cpp
@@ -90,7 +90,7 @@ void random_values(size_t N) {
 
 int main() {
   errors = 0;
-  size_t N = size_t(1) << 32;
+  size_t N = size_t(1) << (sizeof(size_t) * 4); // shift: 32 for 64bit, 16 for 32bit
   random_values(N);
   if (errors == 0) {
     std::cout << std::endl;
diff --git a/tests/random64.cpp b/tests/random64.cpp
index 3f37975..d169fbb 100644
--- a/tests/random64.cpp
+++ b/tests/random64.cpp
@@ -92,7 +92,7 @@ void random_values(size_t N) {
 
 int main() {
   errors = 0;
-  size_t N = size_t(1) << 32;
+  size_t N = size_t(1) << (sizeof(size_t) * 4); // shift: 32 for 64bit, 16 for 32bit
   random_values(N);
   if (errors == 0) {
     std::cout << std::endl;