shishantbiswas/c3c
1
1
Fork 0
mirror of https://github.com/c3lang/c3c.git synced 2026-02-27 03:51:18 +00:00
Code Issues Packages Projects Releases Wiki Activity

optimize test_ct_intlog2 test and whirlpool hash (#2938)

Browse Source 
* optimize `test_ct_intlog2` while still covering all 128 bit positions

* refactor whirlpool to reduce code bloat

replaced the fully unrolled round loop with a runtime loop, reducing
instruction count by 80k in `process_block` and yielding aprox 30%
performance boost due to improved cache locality.

* use compile-time arrays for `test_ct_intlog2`
This commit is contained in:
Manu Linares
2026-02-15 21:54:46 -03:00
committed by GitHub
parent 5d0c41da6b
commit 4b03a84b00
3 changed files with 107 additions and 30 deletions
Download Patch File Download Diff File Expand all files Collapse all files

47
lib/std/hash/whirlpool/whirlpool.c3
View File

@@ -128,15 +128,15 @@ fn char[HASH_SIZE] Whirlpool.final(&self)
}
macro ulong @w_op(#src, $shift) @private
=> S_BOX[(0 * 256) + (int)(#src[($shift + 0) & 7] >> 56) ]
^ S_BOX[(1 * 256) + (int)(#src[($shift + 7) & 7] >> 48) & 0xFF]
^ S_BOX[(2 * 256) + (int)(#src[($shift + 6) & 7] >> 40) & 0xFF]
^ S_BOX[(3 * 256) + (int)(#src[($shift + 5) & 7] >> 32) & 0xFF]
^ S_BOX[(4 * 256) + (int)(#src[($shift + 4) & 7] >> 24) & 0xFF]
^ S_BOX[(5 * 256) + (int)(#src[($shift + 3) & 7] >> 16) & 0xFF]
^ S_BOX[(6 * 256) + (int)(#src[($shift + 2) & 7] >> 8) & 0xFF]
^ S_BOX[(7 * 256) + (int)(#src[($shift + 1) & 7] >> 0) & 0xFF];
macro ulong @w_op(#src, shift) @private
=> S_BOX[(0 * 256) + (int)(#src[(shift + 0) & 7] >> 56) ]
^ S_BOX[(1 * 256) + (int)(#src[(shift + 7) & 7] >> 48) & 0xFF]
^ S_BOX[(2 * 256) + (int)(#src[(shift + 6) & 7] >> 40) & 0xFF]
^ S_BOX[(3 * 256) + (int)(#src[(shift + 5) & 7] >> 32) & 0xFF]
^ S_BOX[(4 * 256) + (int)(#src[(shift + 4) & 7] >> 24) & 0xFF]
^ S_BOX[(5 * 256) + (int)(#src[(shift + 3) & 7] >> 16) & 0xFF]
^ S_BOX[(6 * 256) + (int)(#src[(shift + 2) & 7] >> 8) & 0xFF]
^ S_BOX[(7 * 256) + (int)(#src[(shift + 1) & 7] >> 0) & 0xFF];
const ulong[10] RC @private = {
@@ -158,26 +158,33 @@ fn void Whirlpool.process_block(&self, char* block) @local
ulong[2 * 8] k; // key
ulong[2 * 8] state; // state
// NOTE: These loops are unrolled with C3's Chad-tier compile-time evaluation.
$for var $round = 0; $round < 8; $round++:
k[$round] = self.hash[$round];
state[$round] = $$bswap(mem::load((ulong*)block + $round, 1)) ^ self.hash[$round];
self.hash[$round] = state[$round];
// NOTE: These loops are kept as $for to ensure initial setup is unrolled.
$for var $i = 0; $i < 8; $i++:
k[$i] = self.hash[$i];
state[$i] = $$bswap(mem::load((ulong*)block + $i, 1)) ^ self.hash[$i];
self.hash[$i] = state[$i];
$endfor
$for var $round = 0; $round < ROUNDS; ++$round :
var $m = $round % 2;
// Use regular for loops for the rounds to avoid massive code bloat. 80K less instructions.
for (int round = 0; round < ROUNDS; ++round)
{
int m = round % 2;
int next_m = m ^ 1;
ulong* pk = &k[m * 8];
ulong* nk = &k[next_m * 8];
ulong* ps = &state[m * 8];
ulong* ns = &state[next_m * 8];
k[(($m ^ 1) * 8) + 0] = @w_op((&k[$m * 8]), 0) ^ RC[$round];
nk[0] = @w_op(pk, 0) ^ RC[round];
$for var $i = 1; $i < 8; $i++ :
k[(($m ^ 1) * 8) + $i] = @w_op((&k[$m * 8]), $i);
nk[$i] = @w_op(pk, $i);
$endfor
$for var $i = 0; $i < 8; $i++ :
state[(($m ^ 1) * 8) + $i] = @w_op(&(state[$m * 8]), $i) ^ k[(($m ^ 1) * 8) + $i];
ns[$i] = @w_op(ps, $i) ^ nk[$i];
$endfor
$endfor
}
$for var $x = 0; $x < 8; $x++:
self.hash[$x] ^= state[$x];