Remove usage of __uint128_t and gcc compiler

Use double trick instead: https://codeforces.com/blog/entry/96759

.github/workflows/wheels.yml

@@ -96,18 +96,6 @@ jobs:
     runs-on: windows-2019
     steps:
     - uses: actions/checkout@v2
-    # https://github.com/egor-tensin/setup-mingw/issues/17
-    - name: Set up MinGW
-      uses: egor-tensin/[email protected]
-      with:
-        version: 12.2.0
-    - name: Set environment variables for compiler
-      env:
-        CC: x86_64-w64-mingw32-gcc
-        CXX: x86_64-w64-mingw32-g++
-      run: |
-        echo "CC=${CC}" >> $GITHUB_ENV
-        echo "CXX=${CXX}" >> $GITHUB_ENV
     - name: set up python 3.8
       uses: actions/setup-python@v2
       with:

cryptomite/__init__.py

@@ -16,4 +16,4 @@
 from cryptomite.trevisan import Trevisan
 from cryptomite.utils import von_neumann
 
-__version__ = '0.1.1'
+__version__ = '0.1.2'

src/bigntt.cpp

@@ -21,17 +21,21 @@ uint64_t add(uint64_t a, uint64_t b) {
  * @pre a,b < P
  */
 static uint64_t sub(uint64_t a, uint64_t b) {
-    __int128_t c = a;
-    c -= b;
-    __int128_t d = c + P;
-    __int128_t e = c >> 64;
-    // return c if c >= 0 else d
-    return (c&~e) | (d&e);
+    uint64_t c = a - b;
+    if (a < b) {
+        c += P;
+    }
+
+    return c;
 }
 
 uint64_t mul(uint64_t a, uint64_t b) {
-    __int128_t n = a; n*= b;
-    return n % P;
+    // correct if a,b,P < 2^57
+    uint64_t c = (double)a * b / P;
+    int64_t ans = int64_t(a * b - c * P) % int64_t(P);
+    if (ans < 0)
+        ans += P;
+    return ans;
 }
 
 /**