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
|
import pytest
from efa.efa_common import efa_run_client_server_test
SHM_DEFAULT_MAX_INJECT_SIZE = 4096
SHM_DEFAULT_RX_SIZE = 1024
# This test skips efa-direct because it does not have unexpected message
@pytest.mark.functional
@pytest.mark.parametrize("msg_size", [1, 512, 9000, 1048576]) # cover various switch points of shm/efa protocols
@pytest.mark.parametrize("msg_count", [1, 1024, 2048]) # below and above shm's default rx size
def test_unexpected_msg(cmdline_args, msg_size, msg_count, memory_type, completion_semantic):
from common import ClientServerTest
if cmdline_args.server_id == cmdline_args.client_id:
if (msg_size > SHM_DEFAULT_MAX_INJECT_SIZE or memory_type != "host_to_host" or completion_semantic == "delivery_complete") and msg_count > SHM_DEFAULT_RX_SIZE:
pytest.skip("SHM's CMA/IPC protocol currently cannot handle > rx size number of unexpected messages")
# This fabtests will allocate msg_size * 2 * msg_count memory for send/recv
allocated_memory = msg_size * 2 * msg_count
# The limit size (4 GB) of neuron_tensor_alloc
neuron_maximal_buffer_size = 2**32
if "neuron" in memory_type and allocated_memory >= neuron_maximal_buffer_size:
pytest.skip("Cannot hit neuron allocation limit")
# The EFA limit for single MR that enables remote write is 1M pages aka 4GB for regular pages
maximal_mr_size = 2**32
if allocated_memory >= maximal_mr_size:
pytest.skip("Cannot hit EFA MR limit")
efa_run_client_server_test(cmdline_args, f"fi_unexpected_msg -e rdm -M {msg_count}", iteration_type="short",
completion_semantic=completion_semantic, memory_type=memory_type,
message_size=msg_size, completion_type="queue", timeout=1800, fabric="efa")
|
