weld 0.4.0

Weld is a language and runtime for improving the performance of data-intensive applications.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167

//! Tests related to Weld's SIMD vectorizer transformation.

use weld;
use weld::WeldValue;

mod common;
use crate::common::*;

#[test]
fn simple_for_vectorizable_loop() {
    #[allow(dead_code)]
    struct Args {
        x: WeldVec<i32>,
    }

    let code = "|x:vec[i32]| result(for(simditer(x), merger[i32,+], |b,i,e:simd[i32]| let a = broadcast(1); let a2 = a +
                    broadcast(1); merge(b, e+a2)))";
    let ref conf = default_conf();

    let size = 1000;
    let input_vec = vec![1 as i32; size as usize];
    let ref input_data = WeldVec::from(&input_vec);

    let ret_value = compile_and_run(code, conf, input_data);
    let data = ret_value.data() as *const i32;
    let result = unsafe { (*data).clone() };
    let output = size * 3;
    assert_eq!(result, output);
}

#[test]
fn fringed_for_vectorizable_loop() {
    #[allow(dead_code)]
    struct Args {
        x: WeldVec<i32>,
    }

    let code = "|x:vec[i32]|
	let b1 = for(
		simditer(x),
		merger[i32,+],
		|b,i,e:simd[i32]| let a = broadcast(1); let a2 = a + broadcast(1); merge(b, e+a2));
	result(for(fringeiter(x),
		b1,
		|b,i,e| let a = 1; let a2 = a + 1; merge(b, e+a2)
	))";

    let ref conf = default_conf();

    let size = 1002;
    let input_vec = vec![1 as i32; size as usize];
    let ref input_data = Args {
        x: WeldVec::from(&input_vec),
    };

    let ret_value = compile_and_run(code, conf, input_data);
    let data = ret_value.data() as *const i32;
    let result = unsafe { (*data).clone() };
    let output = size * 3;
    assert_eq!(result, output);
}

#[test]
fn fringed_for_vectorizable_loop_with_par() {
    #[allow(dead_code)]
    struct Args {
        x: WeldVec<i32>,
    }

    let code = "|x:vec[i32]|
	let b1 = for(
		simditer(x),
		merger[i32,+],
		|b,i,e:simd[i32]| let a = broadcast(1); let a2 = a + broadcast(1); merge(b, e+a2));
	result(for(fringeiter(x),
		b1,
		|b,i,e| let a = 1; let a2 = a + 1; merge(b, e+a2)
	))";

    let ref conf = many_threads_conf();

    // Large size to invoke parallel runtime + some fringing.
    let size = 10 * 1000 * 1000 + 123;
    let input_vec = vec![1 as i32; size as usize];
    let ref input_data = Args {
        x: WeldVec::from(&input_vec),
    };

    let ret_value = compile_and_run(code, conf, input_data);
    let data = ret_value.data() as *const i32;
    let result = unsafe { (*data).clone() };
    let output = size * 3;
    assert_eq!(result, output);
}

#[test]
fn for_predicated_vectorizable_loop() {
    #[allow(dead_code)]
    struct Args {
        x: WeldVec<i32>,
    }

    let code = "|x:vec[i32]|
	result(for(
			simditer(x:vec[i32]),
			merger[i32,+],
			|b:merger[i32,+],i:i64,e:simd[i32]|
			  merge(b:merger[i32,+],select(
				(e:simd[i32]>broadcast(0)),
				e:simd[i32],
				broadcast(0):simd[i32]
			  ))
	))
	";
    let ref conf = default_conf();

    let size = 1000;
    let input_vec = vec![1 as i32; size as usize];
    let ref input_data = Args {
        x: WeldVec::from(&input_vec),
    };

    let ret_value = compile_and_run(code, conf, input_data);
    let data = ret_value.data() as *const i32;
    let result = unsafe { (*data).clone() };
    let output = size;
    assert_eq!(result, output);
}

fn check_result_and_free(ret_value: WeldValue, expected: i32) {
    let data = ret_value.data() as *const i32;
    let result = unsafe { (*data).clone() };
    assert_eq!(result, expected);
}

#[test]
fn predicate_if_iff_annotated() {
    #[allow(dead_code)]
    struct Args {
        v: WeldVec<i32>,
    }

    let input_vec = vec![-1, 2, 3, 4, 5];
    let ref input_data = Args {
        v: WeldVec::from(&input_vec),
    };

    let expected = 14;

    /* annotation true */
    let code = "|v:vec[i32]| result(for(v, merger[i32,+], |b,i,e| @(predicate:true)if(e>0, merge(b,e), b)))";
    let ref conf = default_conf();
    let ret_value = compile_and_run(code, conf, input_data);
    check_result_and_free(ret_value, expected);

    /* annotation false */
    let code = "|v:vec[i32]| result(for(v, merger[i32,+], |b,i,e| @(predicate:false)if(e>0, merge(b,e), b)))";
    let ref conf = default_conf();
    let ret_value = compile_and_run(code, conf, input_data);
    check_result_and_free(ret_value, expected);

    /* annotation missing */
    let code = "|v:vec[i32]| result(for(v, merger[i32,+], |b,i,e| if(e>0, merge(b,e), b)))";
    let ref conf = default_conf();
    let ret_value = compile_and_run(code, conf, input_data);
    check_result_and_free(ret_value, expected);
}