Make it compile and run (fairly quick) on Apple M1/M2.

benchmarks/crand.c

@@ -1,11 +1,12 @@
+#include <math.h>
 #include <stdio.h>
 #include <stdlib.h>
 
 int main(void)
 {
-    printf("RAND_MAX = %d\n", RAND_MAX);
+    printf("RAND_MAX = %d (%.1lf bit)\n", RAND_MAX, log2(RAND_MAX));
-    for (uint64_t i = 0; i < 1000000000; ++i)
+    const uint64_t rounds = 1000000000ull * 8 / (unsigned)log2(RAND_MAX);
-    {
+    for (uint64_t i = rounds; i-->0;) {
         rand();
     }
 }

randquik/__init__.py

@@ -1,3 +1,3 @@
-from _version import __version__
+# from _version import __version__
 
-__version__  # Public API
+# __version__  # Public API

randquik/cha.py

@@ -1,3 +1,4 @@
+import sys
 from pathlib import Path
 from typing import Any
 
@@ -24,9 +25,12 @@ ffi.cdef(
     int cha_update(cha_ctx* ctx, uint8_t* out, uint64_t outlen);
     """
 )
-lib = ffi.dlopen(
+libname = "librandquik-chacha20.so"
-    (Path(__file__).parent.parent / "build/librandquik-chacha20.so").as_posix()
+if sys.platform == "darwin":
-)
+    libname = "librandquik-chacha20.dylib"
+elif sys.platform == "win32":
+    libname = "randquik-chacha20.dll"
+lib = ffi.dlopen((Path(__file__).parent.parent / f"build/{libname}").as_posix())
 
 
 def _processKeys(key, iv):

src/chacha20.c

@@ -1,10 +1,25 @@
+#include "chacha20.h"
+
+#if defined(__x86_64__)
 #ifdef __GNUC__
 #pragma GCC target("sse2")
 #pragma GCC target("ssse3")
 #pragma GCC target("avx2")
 #endif
 
-#include "chacha20.h"
+#include <emmintrin.h>  // SSE2
+#include "cha4block.h"
+
+#include <immintrin.h>  // AVX2
+#include <tmmintrin.h>  // SSSE3
+#include "cha8block.h"
+
+#elif defined(__aarch64__)
+#include "sse2neon.h"
+#include "cha4block.h"
+#endif
+
+#include "cha1block.h"
 
 #include <assert.h>
 #include <stdbool.h>
@@ -12,17 +27,10 @@
 #include <stdio.h>
 #include <string.h>
 
-#include <emmintrin.h>
-#include <immintrin.h>
 #include <pthread.h>
 #include <time.h>
-#include <tmmintrin.h>
 #include <unistd.h>
 
-#include "cha1block.h"
-#include "cha4block.h"
-#include "cha8block.h"
-
 void cha_init(cha_ctx* ctx, const uint8_t* key, const uint8_t* iv) {
     ctx->state[0] = 0x61707865;
     ctx->state[1] = 0x3320646e;
@@ -43,7 +51,7 @@ int cha_update(cha_ctx* ctx, uint8_t* out, uint64_t outlen) {
     if (ctx->uncount) {
         // Deliver stored bytes first
         uint64_t N = ctx->uncount >= outlen ? outlen : ctx->uncount;
-        fprintf(stderr, "%lu, %i, %lu\n", N, ctx->uncount, outlen);
+        fprintf(stderr, "%llu, %i, %llu\n", N, ctx->uncount, outlen);
         memcpy(c, ctx->unconsumed, N);
         ctx->uncount -= N;
         c += N;
@@ -53,6 +61,7 @@ int cha_update(cha_ctx* ctx, uint8_t* out, uint64_t outlen) {
         if (c == out + outlen)
             return 0;
     }
+#if defined(__x86_64__)
     // TODO: Handle resume if we are not at block boundary
     if (__builtin_cpu_supports("ssse3")) {
         if (__builtin_cpu_supports("avx2")) {
@@ -62,6 +71,9 @@ int cha_update(cha_ctx* ctx, uint8_t* out, uint64_t outlen) {
         c += _cha_4block(ctx, c, end);
         assert(end - c < 256);
     }
+#elif defined(__aarch64__)
+    c += _cha_4block(ctx, c, end);
+#endif
     c += _cha_block(ctx, c, end);
     assert(c == end);
     return 0;