SHA256C3c, SHA256C3d, SHA256C3e, SHA256C3f,
};
-#ifdef IS_AMD
+#if 1
void sha256_transform (const u32 w[16], u32 digest[8])
{
#else
+// this is basically a much cleaner version, but apparently drops speeds by over 100% :(
+
#define PUTCHAR32_BE(a,p,c) ((u8 *)(a))[(p) ^ 3] = (u8) (c)
#define GETCHAR32_BE(a,p) ((u8 *)(a))[(p) ^ 3]