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
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
//! Persistent collectives example (MPI 4.0+).
//!
//! Demonstrates the use of persistent collectives for iterative algorithms
//! where the same communication pattern is repeated many times.
//!
//! Run with: mpiexec -n 4 cargo run --example persistent_bcast
//!
//! Note: This example requires MPICH 4.0+ or OpenMPI 5.0+ with MPI 4.0 support.
use ferrompi::{Mpi, ReduceOp, Result};
fn main() -> Result<()> {
let mpi = Mpi::init()?;
let world = mpi.world();
let rank = world.rank();
let size = world.size();
if rank == 0 {
println!("Testing persistent collectives (MPI 4.0+)");
println!("MPI Version: {}", Mpi::version()?);
println!("Running on {} processes\n", size);
}
world.barrier()?;
// ============================================================
// Test 1: Persistent Broadcast
// ============================================================
if rank == 0 {
println!("Test 1: Persistent Broadcast");
println!("----------------------------");
}
// Buffer that will be reused for all broadcasts
let mut bcast_buffer = vec![0.0f64; 1000];
// Try to initialize persistent broadcast
match world.bcast_init(&mut bcast_buffer, 0) {
Ok(mut persistent_bcast) => {
let num_iterations = 100;
let start_time = Mpi::wtime();
for iter in 0..num_iterations {
// Root updates the buffer
if rank == 0 {
for (i, x) in bcast_buffer.iter_mut().enumerate() {
*x = (iter * 1000 + i) as f64;
}
}
// Start the persistent operation
persistent_bcast.start()?;
// Optionally do other work here while communication proceeds...
// Wait for completion
persistent_bcast.wait()?;
// Verify on non-root processes
if rank != 0 {
for (i, &x) in bcast_buffer.iter().enumerate() {
let expected = (iter * 1000 + i) as f64;
debug_assert!(
(x - expected).abs() < 1e-10,
"Mismatch at iter {}, index {}: expected {}, got {}",
iter,
i,
expected,
x
);
}
}
}
let elapsed = Mpi::wtime() - start_time;
let throughput = num_iterations as f64 / elapsed;
world.barrier()?;
if rank == 0 {
println!(
" ✓ {} iterations completed in {:.4}s",
num_iterations, elapsed
);
println!(" ✓ Throughput: {:.1} broadcasts/second", throughput);
}
}
Err(e) => {
if rank == 0 {
println!(" ⚠ Persistent broadcast not available: {}", e);
println!(" (This requires MPI 4.0+)");
}
}
}
world.barrier()?;
// ============================================================
// Test 2: Persistent All-Reduce
// ============================================================
if rank == 0 {
println!("\nTest 2: Persistent All-Reduce");
println!("-----------------------------");
}
let mut allreduce_send = vec![0.0f64; 500];
let mut allreduce_recv = vec![0.0f64; 500];
match world.allreduce_init(&allreduce_send, &mut allreduce_recv, ReduceOp::Sum) {
Ok(mut persistent_allreduce) => {
let num_iterations = 100;
let start_time = Mpi::wtime();
for iter in 0..num_iterations {
// Each rank contributes its rank value
for x in allreduce_send.iter_mut() {
*x = rank as f64 + iter as f64;
}
persistent_allreduce.start()?;
persistent_allreduce.wait()?;
// Verify: sum should be (0 + 1 + ... + (size-1)) + iter*size
let expected: f64 = (0..size).map(|r| r as f64 + iter as f64).sum();
for &x in &allreduce_recv {
debug_assert!(
(x - expected).abs() < 1e-10,
"Allreduce mismatch at iter {}",
iter
);
}
}
let elapsed = Mpi::wtime() - start_time;
let throughput = num_iterations as f64 / elapsed;
world.barrier()?;
if rank == 0 {
println!(
" ✓ {} iterations completed in {:.4}s",
num_iterations, elapsed
);
println!(" ✓ Throughput: {:.1} all-reduces/second", throughput);
}
}
Err(e) => {
if rank == 0 {
println!(" ⚠ Persistent all-reduce not available: {}", e);
}
}
}
world.barrier()?;
// ============================================================
// Test 3: Comparison with Non-Persistent
// ============================================================
if rank == 0 {
println!("\nTest 3: Performance Comparison");
println!("------------------------------");
}
let compare_iterations = 1000;
let buffer_size = 100;
// Non-persistent broadcast timing
{
let mut buffer = vec![0.0f64; buffer_size];
world.barrier()?;
let start = Mpi::wtime();
for iter in 0..compare_iterations {
if rank == 0 {
for (i, x) in buffer.iter_mut().enumerate() {
*x = (iter + i) as f64;
}
}
world.broadcast(&mut buffer, 0)?;
}
let non_persistent_time = Mpi::wtime() - start;
if rank == 0 {
println!(
" Non-persistent: {} broadcasts in {:.4}s ({:.1}/s)",
compare_iterations,
non_persistent_time,
compare_iterations as f64 / non_persistent_time
);
}
}
// Persistent broadcast timing
{
let mut buffer = vec![0.0f64; buffer_size];
if let Ok(mut persistent) = world.bcast_init(&mut buffer, 0) {
world.barrier()?;
let start = Mpi::wtime();
for iter in 0..compare_iterations {
if rank == 0 {
for (i, x) in buffer.iter_mut().enumerate() {
*x = (iter + i) as f64;
}
}
persistent.start()?;
persistent.wait()?;
}
let persistent_time = Mpi::wtime() - start;
if rank == 0 {
println!(
" Persistent: {} broadcasts in {:.4}s ({:.1}/s)",
compare_iterations,
persistent_time,
compare_iterations as f64 / persistent_time
);
}
}
}
world.barrier()?;
if rank == 0 {
println!("\n========================================");
println!("Persistent collectives tests complete!");
println!("========================================");
}
Ok(())
}