moonbitlang / x / 939

Committed 21 Jul 2026 02:30AM UTC coverage: 84.504%. First build

Build # 939

Build Type

Pull #280

github

Committed by

web-flow

Commit Message

Merge 7553848b9 into 90bd61809

Pull Request Pull Request #280: Optimize ChaCha with target-aware SIMD

Coverage Stats

72 of 171 new or added lines in 3 files covered. (42.11%)

2634 of 3117 relevant lines covered (84.5%)

269.81 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

41.25

/crypto/simd.mbt

// Copyright 2025 International Digital Economy Academy
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

///|
#cfg(any(target="native", target="wasm"))
#inline
fn load_u32x16_be_into(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  words : FixedArray[UInt],
) -> Unit {
  for group in 0..<4 {
    let value = @v128.v128_load(bytes, byte_offset + group * 16)
    let value = @v128.i8x16_shuffle(
      value, value, 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12,
    )
    let word_offset = group * 4
    words.unsafe_set(word_offset, @v128.i32x4_extract_lane(value, 0))
    words.unsafe_set(word_offset + 1, @v128.i32x4_extract_lane(value, 1))
    words.unsafe_set(word_offset + 2, @v128.i32x4_extract_lane(value, 2))
    words.unsafe_set(word_offset + 3, @v128.i32x4_extract_lane(value, 3))
  }
}

///|
#cfg(not(any(target="native", target="wasm")))
#inline
fn load_u32x16_be_into(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  words : FixedArray[UInt],
) -> Unit {
  for word_offset in 0..<16 {
    let offset = byte_offset + word_offset * 4
    words.unsafe_set(
      word_offset,
      (bytes.unsafe_get(offset).to_uint() << 24) |
      (bytes.unsafe_get(offset + 1).to_uint() << 16) |
      (bytes.unsafe_get(offset + 2).to_uint() << 8) |
      bytes.unsafe_get(offset + 3).to_uint(),
    )
  }
}

///|
#cfg(any(target="native", target="wasm"))
#inline
fn load_i32x16_be_into(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  words : FixedArray[Int],
) -> Unit {
  for group in 0..<4 {
    let value = @v128.v128_load(bytes, byte_offset + group * 16)
    let value = @v128.i8x16_shuffle(
      value, value, 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12,
    )
    let word_offset = group * 4
    words.unsafe_set(
      word_offset,
      @v128.i32x4_extract_lane(value, 0).reinterpret_as_int(),
    )
    words.unsafe_set(
      word_offset + 1,
      @v128.i32x4_extract_lane(value, 1).reinterpret_as_int(),
    )
    words.unsafe_set(
      word_offset + 2,
      @v128.i32x4_extract_lane(value, 2).reinterpret_as_int(),
    )
    words.unsafe_set(
      word_offset + 3,
      @v128.i32x4_extract_lane(value, 3).reinterpret_as_int(),
    )
  }
}

///|
#cfg(not(any(target="native", target="wasm")))
#inline
fn load_i32x16_be_into(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  words : FixedArray[Int],
) -> Unit {
  for word_offset in 0..<16 {
    let offset = byte_offset + word_offset * 4
    words.unsafe_set(
      word_offset,
      (bytes.unsafe_get(offset).to_int() << 24) |
      (bytes.unsafe_get(offset + 1).to_int() << 16) |
      (bytes.unsafe_get(offset + 2).to_int() << 8) |
      bytes.unsafe_get(offset + 3).to_int(),
    )
  }
}

///|
#cfg(any(target="native", target="wasm"))
#inline
fn load_u32x16_le_into(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  words : FixedArray[UInt],
) -> Unit {
  for group in 0..<4 {
    let value = @v128.v128_load(bytes, byte_offset + group * 16)
    let word_offset = group * 4
    words.unsafe_set(word_offset, @v128.i32x4_extract_lane(value, 0))
    words.unsafe_set(word_offset + 1, @v128.i32x4_extract_lane(value, 1))
    words.unsafe_set(word_offset + 2, @v128.i32x4_extract_lane(value, 2))
    words.unsafe_set(word_offset + 3, @v128.i32x4_extract_lane(value, 3))
  }
}

///|
#cfg(not(any(target="native", target="wasm")))
#inline
fn load_u32x16_le_into(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  words : FixedArray[UInt],
) -> Unit {
  for word_offset in 0..<16 {
    let offset = byte_offset + word_offset * 4
    words.unsafe_set(
      word_offset,
      bytes.unsafe_get(offset).to_uint() |
      (bytes.unsafe_get(offset + 1).to_uint() << 8) |
      (bytes.unsafe_get(offset + 2).to_uint() << 16) |
      (bytes.unsafe_get(offset + 3).to_uint() << 24),
    )
  }
}

///|
#cfg(any(target="native", target="wasm"))
#inline
fn store_u32x4_le(
  words : FixedArray[UInt],
  word_offset : Int,
  bytes : FixedArray[Byte],
  byte_offset : Int,
) -> Unit {
  @v128.v128_store(
    bytes,
    byte_offset,
    @v128.i32x4_const(
      words.unsafe_get(word_offset),
      words.unsafe_get(word_offset + 1),
      words.unsafe_get(word_offset + 2),
      words.unsafe_get(word_offset + 3),
    ),
  )
}

///|
#cfg(not(any(target="native", target="wasm")))
#inline
fn store_u32x4_le(
  words : FixedArray[UInt],
  word_offset : Int,
  bytes : FixedArray[Byte],
  byte_offset : Int,
) -> Unit {
  for lane in 0..<4 {
    let word = words.unsafe_get(word_offset + lane)
    let offset = byte_offset + lane * 4
    bytes.unsafe_set(offset, word.to_byte())
    bytes.unsafe_set(offset + 1, (word >> 8).to_byte())
    bytes.unsafe_set(offset + 2, (word >> 16).to_byte())
    bytes.unsafe_set(offset + 3, (word >> 24).to_byte())
  }
}

///|
#cfg(any(target="native", target="wasm"))
fn xor_bytes_with_splat_in_place(
  bytes : FixedArray[Byte],
  length : Int,
  mask : Byte,
) -> Unit {
  let simd_end = length / 16 * 16
  let vector_mask = @v128.i8x16_splat(mask)
  for offset = 0; offset < simd_end; offset = offset + 16 {
    @v128.v128_load(bytes, offset)
    |> @v128.v128_xor(vector_mask)
    |> value => { @v128.v128_store(bytes, offset, value) }
  }
  for offset in simd_end..<length {
    bytes.unsafe_set(offset, bytes.unsafe_get(offset) ^ mask)
  }
}

///|
#cfg(any(target="native", target="wasm"))
#inline
fn xor_byte_ranges_in_place(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  mask : FixedArray[Byte],
  mask_offset : Int,
  length : Int,
) -> Unit {
  let simd_end = length / 16 * 16
  for offset = 0; offset < simd_end; offset = offset + 16 {
    @v128.v128_store(
      bytes,
      byte_offset + offset,
      @v128.v128_xor(
        @v128.v128_load(bytes, byte_offset + offset),
        @v128.v128_load(mask, mask_offset + offset),
      ),
    )
  }
  for offset = simd_end; offset < length; offset = offset + 1 {
    bytes.unsafe_set(
      byte_offset + offset,
      bytes.unsafe_get(byte_offset + offset) ^
      mask.unsafe_get(mask_offset + offset),
    )
  }
}

///|
#cfg(not(any(target="native", target="wasm", target="js")))
#inline
fn xor_byte_ranges_in_place(
  bytes : FixedArray[Byte],
  byte_offset : Int,
  mask : FixedArray[Byte],
  mask_offset : Int,
  length : Int,
) -> Unit {
  for offset in 0..<length {
    bytes.unsafe_set(
      byte_offset + offset,
      bytes.unsafe_get(byte_offset + offset) ^
      mask.unsafe_get(mask_offset + offset),
    )
  }
}

///|
#cfg(not(any(target="native", target="wasm")))
fn xor_bytes_with_splat_in_place(
  bytes : FixedArray[Byte],
  length : Int,
  mask : Byte,
) -> Unit {
  for offset in 0..<length {
    bytes.unsafe_set(offset, bytes.unsafe_get(offset) ^ mask)
  }
}

1	// Copyright 2025 International Digital Economy Academy
2	//
3	// Licensed under the Apache License, Version 2.0 (the "License");
4	// you may not use this file except in compliance with the License.
5	// You may obtain a copy of the License at
6	//
7	// http://www.apache.org/licenses/LICENSE-2.0
8	//
9	// Unless required by applicable law or agreed to in writing, software
10	// distributed under the License is distributed on an "AS IS" BASIS,
11	// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12	// See the License for the specific language governing permissions and
13	// limitations under the License.
14
15	///\|
16	#cfg(any(target="native", target="wasm"))
17	#inline
18	fn load_u32x16_be_into(
19	bytes : FixedArray[Byte],
20	byte_offset : Int,
21	words : FixedArray[UInt],
22	) -> Unit {
23	for group in 0..<4 {	×
24	let value = @v128.v128_load(bytes, byte_offset + group * 16)	×
25	let value = @v128.i8x16_shuffle(	×
26	value, value, 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12,
27	)
28	let word_offset = group * 4
29	words.unsafe_set(word_offset, @v128.i32x4_extract_lane(value, 0))	×
30	words.unsafe_set(word_offset + 1, @v128.i32x4_extract_lane(value, 1))	×
31	words.unsafe_set(word_offset + 2, @v128.i32x4_extract_lane(value, 2))	×
32	words.unsafe_set(word_offset + 3, @v128.i32x4_extract_lane(value, 3))	×
33	}
34	}
35
36	///\|
37	#cfg(not(any(target="native", target="wasm")))
38	#inline
39	fn load_u32x16_be_into(
40	bytes : FixedArray[Byte],
41	byte_offset : Int,
42	words : FixedArray[UInt],
43	) -> Unit {
44	for word_offset in 0..<16 {	122✔
45	let offset = byte_offset + word_offset * 4	1,952✔
46	words.unsafe_set(	1,952✔
47	word_offset,
48	(bytes.unsafe_get(offset).to_uint() << 24) \|	1,952✔
49	(bytes.unsafe_get(offset + 1).to_uint() << 16) \|	1,952✔
50	(bytes.unsafe_get(offset + 2).to_uint() << 8) \|	1,952✔
51	bytes.unsafe_get(offset + 3).to_uint(),	1,952✔
52	)
53	}
54	}
55
56	///\|
57	#cfg(any(target="native", target="wasm"))
58	#inline
59	fn load_i32x16_be_into(
60	bytes : FixedArray[Byte],
61	byte_offset : Int,
62	words : FixedArray[Int],
63	) -> Unit {
64	for group in 0..<4 {	×
65	let value = @v128.v128_load(bytes, byte_offset + group * 16)	×
66	let value = @v128.i8x16_shuffle(	×
67	value, value, 3, 2, 1, 0, 7, 6, 5, 4, 11, 10, 9, 8, 15, 14, 13, 12,
68	)
69	let word_offset = group * 4
70	words.unsafe_set(	×
71	word_offset,
72	@v128.i32x4_extract_lane(value, 0).reinterpret_as_int(),	×
73	)
74	words.unsafe_set(	×
75	word_offset + 1,
76	@v128.i32x4_extract_lane(value, 1).reinterpret_as_int(),	×
77	)
78	words.unsafe_set(	×
79	word_offset + 2,
80	@v128.i32x4_extract_lane(value, 2).reinterpret_as_int(),	×
81	)
82	words.unsafe_set(	×
83	word_offset + 3,
84	@v128.i32x4_extract_lane(value, 3).reinterpret_as_int(),	×
85	)
86	}
87	}
88
89	///\|
90	#cfg(not(any(target="native", target="wasm")))
91	#inline
92	fn load_i32x16_be_into(
93	bytes : FixedArray[Byte],
94	byte_offset : Int,
95	words : FixedArray[Int],
96	) -> Unit {
97	for word_offset in 0..<16 {	534✔
98	let offset = byte_offset + word_offset * 4	8,544✔
99	words.unsafe_set(	8,544✔
100	word_offset,
101	(bytes.unsafe_get(offset).to_int() << 24) \|	8,544✔
102	(bytes.unsafe_get(offset + 1).to_int() << 16) \|	8,544✔
103	(bytes.unsafe_get(offset + 2).to_int() << 8) \|	8,544✔
104	bytes.unsafe_get(offset + 3).to_int(),	8,544✔
105	)
106	}
107	}
108
109	///\|
110	#cfg(any(target="native", target="wasm"))
111	#inline
112	fn load_u32x16_le_into(
113	bytes : FixedArray[Byte],
114	byte_offset : Int,
115	words : FixedArray[UInt],
116	) -> Unit {
117	for group in 0..<4 {	×
118	let value = @v128.v128_load(bytes, byte_offset + group * 16)	×
119	let word_offset = group * 4
120	words.unsafe_set(word_offset, @v128.i32x4_extract_lane(value, 0))	×
121	words.unsafe_set(word_offset + 1, @v128.i32x4_extract_lane(value, 1))	×
122	words.unsafe_set(word_offset + 2, @v128.i32x4_extract_lane(value, 2))	×
123	words.unsafe_set(word_offset + 3, @v128.i32x4_extract_lane(value, 3))	×
124	}
125	}
126
127	///\|
128	#cfg(not(any(target="native", target="wasm")))
129	#inline
130	fn load_u32x16_le_into(
131	bytes : FixedArray[Byte],
132	byte_offset : Int,
133	words : FixedArray[UInt],
134	) -> Unit {
135	for word_offset in 0..<16 {	53✔
136	let offset = byte_offset + word_offset * 4	848✔
137	words.unsafe_set(	848✔
138	word_offset,
139	bytes.unsafe_get(offset).to_uint() \|	848✔
140	(bytes.unsafe_get(offset + 1).to_uint() << 8) \|	848✔
141	(bytes.unsafe_get(offset + 2).to_uint() << 16) \|	848✔
142	(bytes.unsafe_get(offset + 3).to_uint() << 24),	848✔
143	)
144	}
145	}
146
147	///\|
148	#cfg(any(target="native", target="wasm"))
149	#inline
150	fn store_u32x4_le(
151	words : FixedArray[UInt],
152	word_offset : Int,
153	bytes : FixedArray[Byte],
154	byte_offset : Int,
155	) -> Unit {
156	@v128.v128_store(	×
157	bytes,
158	byte_offset,
159	@v128.i32x4_const(	×
160	words.unsafe_get(word_offset),	×
161	words.unsafe_get(word_offset + 1),	×
162	words.unsafe_get(word_offset + 2),	×
163	words.unsafe_get(word_offset + 3),	×
164	),
165	)
166	}
167
168	///\|
169	#cfg(not(any(target="native", target="wasm")))
170	#inline
171	fn store_u32x4_le(
172	words : FixedArray[UInt],
173	word_offset : Int,
174	bytes : FixedArray[Byte],
175	byte_offset : Int,
176	) -> Unit {
177	for lane in 0..<4 {	16✔
178	let word = words.unsafe_get(word_offset + lane)	64✔
179	let offset = byte_offset + lane * 4
180	bytes.unsafe_set(offset, word.to_byte())	64✔
181	bytes.unsafe_set(offset + 1, (word >> 8).to_byte())	64✔
182	bytes.unsafe_set(offset + 2, (word >> 16).to_byte())	64✔
183	bytes.unsafe_set(offset + 3, (word >> 24).to_byte())	64✔
184	}
185	}
186
187	///\|
188	#cfg(any(target="native", target="wasm"))
189	fn xor_bytes_with_splat_in_place(
190	bytes : FixedArray[Byte],
191	length : Int,
192	mask : Byte,
193	) -> Unit {
194	let simd_end = length / 16 * 16	×
195	let vector_mask = @v128.i8x16_splat(mask)	×
196	for offset = 0; offset < simd_end; offset = offset + 16 {	×
197	@v128.v128_load(bytes, offset)	×
198	\|> @v128.v128_xor(vector_mask)	×
199	\|> value => { @v128.v128_store(bytes, offset, value) }	×
200	}
201	for offset in simd_end..<length {
202	bytes.unsafe_set(offset, bytes.unsafe_get(offset) ^ mask)	×
203	}
204	}
205
206	///\|
207	#cfg(any(target="native", target="wasm"))
208	#inline
209	fn xor_byte_ranges_in_place(
210	bytes : FixedArray[Byte],
211	byte_offset : Int,
212	mask : FixedArray[Byte],
213	mask_offset : Int,
214	length : Int,
215	) -> Unit {
NEW 216	let simd_end = length / 16 * 16	×
NEW 217	for offset = 0; offset < simd_end; offset = offset + 16 {	×
NEW 218	@v128.v128_store(	×
219	bytes,
220	byte_offset + offset,
NEW 221	@v128.v128_xor(	×
NEW 222	@v128.v128_load(bytes, byte_offset + offset),	×
NEW 223	@v128.v128_load(mask, mask_offset + offset),	×
224	),
225	)
226	}
NEW 227	for offset = simd_end; offset < length; offset = offset + 1 {	×
NEW 228	bytes.unsafe_set(	×
229	byte_offset + offset,
NEW 230	bytes.unsafe_get(byte_offset + offset) ^	×
NEW 231	mask.unsafe_get(mask_offset + offset),	×
232	)
233	}
234	}
235
236	///\|
237	#cfg(not(any(target="native", target="wasm", target="js")))
238	#inline
239	fn xor_byte_ranges_in_place(
240	bytes : FixedArray[Byte],
241	byte_offset : Int,
242	mask : FixedArray[Byte],
243	mask_offset : Int,
244	length : Int,
245	) -> Unit {
246	for offset in 0..<length {	19✔
247	bytes.unsafe_set(	1,216✔
248	byte_offset + offset,
249	bytes.unsafe_get(byte_offset + offset) ^	1,216✔
250	mask.unsafe_get(mask_offset + offset),	1,216✔
251	)
252	}
253	}
254
255	///\|
256	#cfg(not(any(target="native", target="wasm")))
257	fn xor_bytes_with_splat_in_place(
258	bytes : FixedArray[Byte],
259	length : Int,
260	mask : Byte,
261	) -> Unit {
262	for offset in 0..<length {	10✔
263	bytes.unsafe_set(offset, bytes.unsafe_get(offset) ^ mask)	640✔
264	}
265	}

moonbitlang / x / 939

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous