Forsaken ProjectX C++ library
- support platform independent frameworks
- support forsaken file formats
- support checksum algorithm
- support aligned linalg simd types
Iterating in 3 byte steps over 4096 byte blocks and shifting the i64 zero extended 3bytes by a (0..31) clamped multiple of 7.
Writing out all the shift amount and the results I came up with the sequence
(2,3,5|7|14,21,28|31|35|37) that alternates between prime numbers and multiples of 7 and the term LFSL
Use the sequence to generate Mersenne primes as the loop shift step.
constexpr i64 compute_block( u8 buf[4096], u64 len, u32 shl = 0)
{
i64 ret = 0;
for(; len>=((1<<2)-1); buf+=(1<<2)-1, len-=(1<<2)-1, shl=(shl + ((1<<3)-1)) % (1<<5))
ret += ((i64)le32toh(*(u32*)buf & 0x00FFFFFF)) << shl;
if(len > 0)
ret += ((i64)le16toh(*(u16*)buf & (len == 2 ? 0xFFFF : 0x00FF))) << shl;
return ret;
}prjxsum -r data/levels
0033edcdbe539ba0 *data/levels/accworld/accworld.bsp
63f04c40700994f6 *data/levels/alpha/alpha.bsp
5abe6fdc6c8805a7 *data/levels/arena/arena.bsp
6666a6ef564aa3a5 *data/levels/asteroid/asteroid.bsp
4055f871a6c94a95 *data/levels/astro/astro.bsp
5b41245d017b4439 *data/levels/asubchb/asubchb.bsp
7fbb19a1d8dabd3d *data/levels/azchb/azchb.bsp
b0dccc483a3accd6 *data/levels/azt-sp01/azt-sp01.bsp
7fbb19a1d8dabd3d *data/levels/aztec/aztec.bspusing namespace prjx::math;
using namespace prjx::checksum;
using namespace prjx::file;