blob: 45714ec4ede9e84f0efd89b36b9844103f5b1dc7 (
plain)
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
|
#!/bin/sh
set -e
echo "prim-benchmarks CPU-DPU alloc (dfatool edition)"
echo "Started at $(date)"
echo "Revision $(git describe --always)"
./make-size.sh 0
for i in $(seq 1 20); do
for k in BROADCAST; do
# BROADCAST sends the same data to all DPUs, so data size must not exceed the amount of MRAM available on a single DPU (i.e., 64 MB)
for l in 4194304 6291456; do
make -B NR_RANKS=$i NR_TASKLETS=1 BL=10 TRANSFER=$k NUMA=1
for numa_rank in 0 1; do
sudo limit_ranks_to_numa_node $numa_rank
for numa_in in 0 1; do
for numa_out in 0 1; do
for numa_cpu in 0 1; do
bin/host_code -a $numa_in -b $numa_out -c $numa_cpu -w 0 -e 100 -x 1 -N 0 -I $(size -A bin/dpu_code | awk '($1 == ".text") {print $2/8}') -i $l
done
done
done
done
done
done
# utilize 32MiB / 50% of per-DPU MRAM capacity -- otherwise DRAM capacity per NUMA node is insufficient
for numa_rank in 0 1; do
sudo limit_ranks_to_numa_node $numa_rank
for numa_in in 0 1; do
for numa_out in 0 1; do
for numa_cpu in 0 1; do
make -B NR_RANKS=$i NR_TASKLETS=1 BL=10 TRANSFER=PUSH NUMA=1
bin/host_code -a $numa_in -b $numa_out -c $numa_cpu -w 0 -e 100 -x 0 -N 0 -I $(size -A bin/dpu_code | awk '($1 == ".text") {print $2/8}') -i 1
bin/host_code -a $numa_in -b $numa_out -c $numa_cpu -w 0 -e 100 -x 0 -N 0 -I $(size -A bin/dpu_code | awk '($1 == ".text") {print $2/8}') -i 4194304
make -B NR_RANKS=$i NR_TASKLETS=1 BL=10 TRANSFER=SERIAL NUMA=1
bin/host_code -a $numa_in -b $numa_out -c $numa_cpu -w 0 -e 100 -x 0 -N 0 -I $(size -A bin/dpu_code | awk '($1 == ".text") {print $2/8}') -i 1
done
done
done
done
done
sudo limit_ranks_to_numa_node any
echo "Completed at $(date)"
|
