Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Blake2s optimizations #99

Merged
merged 11 commits into from
Feb 19, 2024
Merged

Blake2s optimizations #99

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
11 commits
Select commit Hold shift + click to select a range
5f31f54
Blake2s opt
fmkra Feb 6, 2024
4135798
Fmt
fmkra Feb 7, 2024
3ce4a9b
Revert invalid blake2s opt
fmkra Feb 7, 2024
e9f8537
Opt blake2s u8
fmkra Feb 7, 2024
00acea1
Opt array_extend
fmkra Feb 7, 2024
c36b61d
Opt from_span
fmkra Feb 8, 2024
96d3434
Merge branch 'main' into feat/blake2s-opts
fmkra Feb 8, 2024
9ba77b7
Remove from_span
fmkra Feb 8, 2024
56e96c8
array_append opt
fmkra Feb 8, 2024
627903d
Merge branch 'main_benches' into feat/blake2s-opts
fmkra Feb 9, 2024
29505da
Merge branch 'main' into feat/blake2s-opts
fmkra Feb 9, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
24 changes: 10 additions & 14 deletions src/common/array_append.cairo
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,25 +50,21 @@ impl ArrayU32AppendFelt of ArrayAppendTrait<u32, felt252> {
}

impl ArrayU32AppendFeltsSpan of ArrayAppendTrait<u32, Span<felt252>> {
fn append_little_endian(ref self: Array<u32>, element: Span<felt252>) {
let mut i = 0;
fn append_little_endian(ref self: Array<u32>, mut element: Span<felt252>) {
loop {
if i == element.len() {
break;
match element.pop_front() {
Option::Some(elem) => self.append_little_endian(*elem),
Option::None => { break; }
}
self.append_little_endian(*element[i]);
i += 1;
};
}
}
fn append_big_endian(ref self: Array<u32>, element: Span<felt252>) {
let mut i = 0;
fn append_big_endian(ref self: Array<u32>, mut element: Span<felt252>) {
loop {
if i == element.len() {
break;
match element.pop_front() {
Option::Some(elem) => self.append_big_endian(*elem),
Option::None => { break; }
}
self.append_big_endian(*element[i]);
i += 1;
};
}
}
}

Expand Down
210 changes: 80 additions & 130 deletions src/common/blake2s.cairo
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,15 +87,7 @@ fn blake2s_init() -> blake2s_state {
0x1F83D9AB,
0x5BE0CD19
];
let mut buf = ArrayTrait::new();
let mut i = 0;
loop {
if i == 16 {
break;
}
buf.append(0);
i += 1;
};
let mut buf = array![0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0];

blake2s_state { h: blake2s_IV, t0: 0, t1: 0, f0: 0, buf: buf, buflen: 0 }
}
Expand Down Expand Up @@ -132,127 +124,85 @@ fn blake2s_compress(mut s: blake2s_state, m: Array<u32>) -> blake2s_state {

// ROUND function begin

let mut a = 0;
let mut b = 0;
let mut c = 0;
let mut d = 0;
let mut i = 0;
loop {
if i == 8 {
break;
}
if i == 0 {
a = v0;
b = v4;
c = v8;
d = v12;
} else if i == 1 {
a = v1;
b = v5;
c = v9;
d = v13;
} else if i == 2 {
a = v2;
b = v6;
c = v10;
d = v14;
} else if i == 3 {
a = v3;
b = v7;
c = v11;
d = v15;
} else if i == 4 {
a = v0;
b = v5;
c = v10;
d = v15;
} else if i == 5 {
a = v1;
b = v6;
c = v11;
d = v12;
} else if i == 6 {
a = v2;
b = v7;
c = v8;
d = v13;
} else if i == 7 {
a = v3;
b = v4;
c = v9;
d = v14;
};

// G function begin

// a = a + b + m[sigma[r][2*i]]
a = u32_wrapping_add(u32_wrapping_add(a, b), *m_span.at(*sigma[2 * i]));

d = rotr16(d ^ a);

// c = c + d
c = u32_wrapping_add(c, d);

b = rotr12(b ^ c);

// a = a + b + m[sigma[r][2*i+1]]
a = u32_wrapping_add(u32_wrapping_add(a, b), *m_span.at(*sigma[2 * i + 1]));

d = rotr8(d ^ a);

// c = c + d
c = u32_wrapping_add(c, d);

b = rotr7(b ^ c);

// G function end

if i == 0 {
v0 = a;
v4 = b;
v8 = c;
v12 = d;
} else if i == 1 {
v1 = a;
v5 = b;
v9 = c;
v13 = d;
} else if i == 2 {
v2 = a;
v6 = b;
v10 = c;
v14 = d;
} else if i == 3 {
v3 = a;
v7 = b;
v11 = c;
v15 = d;
} else if i == 4 {
v0 = a;
v5 = b;
v10 = c;
v15 = d;
} else if i == 5 {
v1 = a;
v6 = b;
v11 = c;
v12 = d;
} else if i == 6 {
v2 = a;
v7 = b;
v8 = c;
v13 = d;
} else if i == 7 {
v3 = a;
v4 = b;
v9 = c;
v14 = d;
};

i += 1;
};

// ROUND function end
// 0 - 0,4,8,12
v0 = u32_wrapping_add(u32_wrapping_add(v0, v4), *m_span.at(*sigma[0]));
v12 = rotr16(v12 ^ v0);
v8 = u32_wrapping_add(v8, v12);
v4 = rotr12(v4 ^ v8);
v0 = u32_wrapping_add(u32_wrapping_add(v0, v4), *m_span.at(*sigma[1]));
v12 = rotr8(v12 ^ v0);
v8 = u32_wrapping_add(v8, v12);
v4 = rotr7(v4 ^ v8);

// 1 - 1,5,9,13
v1 = u32_wrapping_add(u32_wrapping_add(v1, v5), *m_span.at(*sigma[2]));
v13 = rotr16(v13 ^ v1);
v9 = u32_wrapping_add(v9, v13);
v5 = rotr12(v5 ^ v9);
v1 = u32_wrapping_add(u32_wrapping_add(v1, v5), *m_span.at(*sigma[3]));
v13 = rotr8(v13 ^ v1);
v9 = u32_wrapping_add(v9, v13);
v5 = rotr7(v5 ^ v9);

// 2 - 2,6,10,14
v2 = u32_wrapping_add(u32_wrapping_add(v2, v6), *m_span.at(*sigma[4]));
v14 = rotr16(v14 ^ v2);
v10 = u32_wrapping_add(v10, v14);
v6 = rotr12(v6 ^ v10);
v2 = u32_wrapping_add(u32_wrapping_add(v2, v6), *m_span.at(*sigma[5]));
v14 = rotr8(v14 ^ v2);
v10 = u32_wrapping_add(v10, v14);
v6 = rotr7(v6 ^ v10);

// 3 - 3,7,11,15
v3 = u32_wrapping_add(u32_wrapping_add(v3, v7), *m_span.at(*sigma[6]));
v15 = rotr16(v15 ^ v3);
v11 = u32_wrapping_add(v11, v15);
v7 = rotr12(v7 ^ v11);
v3 = u32_wrapping_add(u32_wrapping_add(v3, v7), *m_span.at(*sigma[7]));
v15 = rotr8(v15 ^ v3);
v11 = u32_wrapping_add(v11, v15);
v7 = rotr7(v7 ^ v11);

// 4 - 0,5,10,15
v0 = u32_wrapping_add(u32_wrapping_add(v0, v5), *m_span.at(*sigma[8]));
v15 = rotr16(v15 ^ v0);
v10 = u32_wrapping_add(v10, v15);
v5 = rotr12(v5 ^ v10);
v0 = u32_wrapping_add(u32_wrapping_add(v0, v5), *m_span.at(*sigma[9]));
v15 = rotr8(v15 ^ v0);
v10 = u32_wrapping_add(v10, v15);
v5 = rotr7(v5 ^ v10);

// 5 - 1,6,11,12
v1 = u32_wrapping_add(u32_wrapping_add(v1, v6), *m_span.at(*sigma[10]));
v12 = rotr16(v12 ^ v1);
v11 = u32_wrapping_add(v11, v12);
v6 = rotr12(v6 ^ v11);
v1 = u32_wrapping_add(u32_wrapping_add(v1, v6), *m_span.at(*sigma[11]));
v12 = rotr8(v12 ^ v1);
v11 = u32_wrapping_add(v11, v12);
v6 = rotr7(v6 ^ v11);

// 6 - 2,7,8,13
v2 = u32_wrapping_add(u32_wrapping_add(v2, v7), *m_span.at(*sigma[12]));
v13 = rotr16(v13 ^ v2);
v8 = u32_wrapping_add(v8, v13);
v7 = rotr12(v7 ^ v8);
v2 = u32_wrapping_add(u32_wrapping_add(v2, v7), *m_span.at(*sigma[13]));
v13 = rotr8(v13 ^ v2);
v8 = u32_wrapping_add(v8, v13);
v7 = rotr7(v7 ^ v8);

// 7 - 3,4,9,14
v3 = u32_wrapping_add(u32_wrapping_add(v3, v4), *m_span.at(*sigma[14]));
v14 = rotr16(v14 ^ v3);
v9 = u32_wrapping_add(v9, v14);
v4 = rotr12(v4 ^ v9);
v3 = u32_wrapping_add(u32_wrapping_add(v3, v4), *m_span.at(*sigma[15]));
v14 = rotr8(v14 ^ v3);
v9 = u32_wrapping_add(v9, v14);
v4 = rotr7(v4 ^ v9);

r += 1;
};
Expand Down
Loading
Loading