Added DPU platform test cases #14152

nissampa · 2024-08-16T18:01:20Z

Added test_show_platform_dpu.py file containing tests for the platform cli ... commands for dpu under tests/platform_tests/cli/
Added test_reload_dpu.py file containing tests for the reboot and reload ... commands for dpu under tests/platform_tests/
Added device_utils_dpu.py containing utility functions for dpu platform test cases under common/platform

Test Plan PR Link: #12701

Add new tests:

Under test_show_platform_dpu.py

test_midplane_ip    -    [1.5 Check midplane ip address between NPU and DPU]
test_shutdown_power_up_dpu     -     [1.6 Check DPU shutdown and power up individually]
test_reboot_cause     -    [1.11 Check reboot cause history]
test_pcie_link     -    [1.7 Check removal of pcie link between NPU and DPU]

Under test_reload_dpu.py:

test_dpu_ping_after_reboot    -    [1.12 Check the DPU state after OS reboot]
test_show_ping_int_after_reload     -    [Phase 2 Test plan document]

mssonicbld · 2024-08-16T18:02:38Z

The pre-commit check detected issues in the files touched by this pull request.
The pre-commit check is a mandatory check, please fix detected issues.

Detailed pre-commit check results:
trim trailing whitespace.................................................Failed - hook id: trailing-whitespace - exit code: 1 - files were modified by this hook Fixing tests/platform_tests/cli/test_show_platform_dpu.py Fixing tests/platform_tests/test_reload_dpu.py fix end of files.........................................................Passed check yaml...........................................(no files to check)Skipped check for added large files..............................................Passed check python ast.........................................................Passed flake8...................................................................Failed - hook id: flake8 - exit code: 1 tests/common/platform/device_utils_dpu.py:9:1: F403 'from tests.platform_tests.api.conftest import ' used; unable to detect undefined names tests/common/platform/device_utils_dpu.py:9:1: F401 'tests.platform_tests.api.conftest.' imported but unused tests/common/platform/device_utils_dpu.py:10:1: F403 'from tests.common.devices.sonic import ' used; unable to detect undefined names tests/common/platform/device_utils_dpu.py:10:1: F401 'tests.common.devices.sonic.' imported but unused ... [truncated extra lines, please run pre-commit locally to view full check results]

To run the pre-commit checks locally, you can follow below steps:

Ensure that default python is python3. In sonic-mgmt docker container, default python is python2. You can run
the check by activating the python3 virtual environment in sonic-mgmt docker container or outside of sonic-mgmt
docker container.
Ensure that the pre-commit package is installed:

sudo pip install pre-commit

Go to repository root folder
Install the pre-commit hooks:

pre-commit install

Use pre-commit to check staged file:

pre-commit

Alternatively, you can check committed files using:

pre-commit run --from-ref <commit_id> --to-ref <commit_id>

prgeor · 2024-08-26T18:03:22Z

@nissampa Please update the PR description with the HLD link

vvolam · 2024-08-26T22:15:39Z

tests/platform_tests/cli/test_show_platform_dpu.py

+                # Check will be added for ip address once dynamic implementation is in place
+                ip_address_list.append(module.get_midplane_ip(platform_api_conn, index))
+
+    check_dpu_ping_status(duthost, ip_address_list)


There is no failure check here, if ping fails to the DPU hosts. Also, it is good to check for each DPU and log accordingly to debug the failure easily.

added failure check to it.

vvolam · 2024-08-26T22:18:45Z

tests/platform_tests/cli/test_show_platform_dpu.py

+    # Checking the state of dpu health again after bringing up interface and waiting for db entry to get populated
+    dpu_db_after = wait_until(60, 60, 0, count_dpu_modules_in_system_health_cli, duthost)
+
+    pytest_assert(dpu_db_before == count_up == count_down == dpu_db_after, "Link Flap is Tested'{}'".format(duthost.hostname))


It is good to have the DPU information whose link didn't come up after the test. Also, can we have a debug log in successful case?

vvolam · 2024-08-26T22:20:29Z

tests/platform_tests/cli/test_show_platform_dpu.py

+        duthosts.shell("config chassis modules startup %s"%dpu)
+        time.sleep(2)
+
+    pytest_assert(wait_until(120, 60, 0, check_dpu_module_status, duthost, num_modules, "on"),


same here, it is good to log the DPUs that failed to come up. Something like to add a new utils function to get the status and log the information in case of failure.

added the loggin information in the util itself.

vvolam · 2024-08-26T22:21:13Z

tests/platform_tests/cli/test_show_platform_dpu.py

+        time.sleep(2)
+        output_startup_dpu = duthost.shell("config chassis module startup %s"%(dpus_name))["stdout_lines"]
+
+    pytest_assert(wait_until(120, 60, 0, check_dpu_reboot_cause, duthost, num_modules),


same here. please log the DPU status information for easy debugging.

vvolam · 2024-08-26T22:21:50Z

tests/platform_tests/test_reload_dpu.py

+
+def test_dpu_ping_after_reboot(duthosts, enum_rand_one_per_hwsku_hostname, localhost, platform_api_conn):
+    """
+    @summary: Verify output of `config chassis modules startup <DPU_Number>`


could you elaborate more on function summary?

Changed it.

congh-nvidia · 2024-08-30T04:08:49Z

tests/common/platform/device_utils_dpu.py

+from tests.common.devices.sonic import * #noqa
+from tests.common.helpers.assertions import pytest_assert
+
+@pytest.fixture(scope='function', autouse=True)


Why the fixture is autouse? It's better to not put an autouse fixture in a common util file. Or you can remove the autouse so this fixture can be used more flexibly.

And the scope is function, is it necessary to power on the DPUs before each test case.

Idea is to run all the cases with having all the dpus turned on.

r12f · 2024-08-30T04:05:32Z

tests/platform_tests/cli/test_show_platform_dpu.py

+    for index in range(num_modules):
+        dpu_name = module.get_name(platform_api_conn, index)
+        output_shutdown_dpu = duthost.shell("config chassis module shutdown %s"%(dpu_name))["stdout_lines"]
+        time.sleep(2)


will this sleep being enough for waiting for the shutdown? should we use other more stable signal?

Changed it to wait until DPUs turns off.

congh-nvidia · 2024-08-30T04:10:20Z

tests/common/platform/device_utils_dpu.py

+from tests.common.helpers.assertions import pytest_assert
+
+@pytest.fixture(scope='function', autouse=True)
+def dpu_poweron_fixture(duthosts, enum_rand_one_per_hwsku_hostname, request, platform_api_conn):


Usually we don't have the "fixture" in a fixture name.

Changed it.

congh-nvidia · 2024-08-30T04:30:33Z

tests/common/platform/device_utils_dpu.py

+        for i in range(len(output_ping)):
+            if "0% packet loss" in output_ping[i]:
+                ping_count += 1


You can take the full output from "stdout", no need to search in each line.

Changed it.

congh-nvidia · 2024-08-30T04:33:14Z

tests/common/platform/device_utils_dpu.py

+            if "0% packet loss" in output_ping[i]:
+                ping_count += 1
+
+    return (ping_count == len(ip_address_list))


Looks like the function can return with False when any of the ping fails, no need to try the remaining ones.

And the outer parentheses are not necessary.

removed outer parentheses

congh-nvidia · 2024-08-30T04:39:08Z

tests/common/platform/device_utils_dpu.py

+        parse_output = output_dpu_status[index]
+        if parse_output['oper-status'] == 'Offline':
+            dpu_off_count += 1
+        else:


Here should be a condition for the on status in case there is unexpected status output.

The cli output has only three options: offline, online or partial online.

congh-nvidia · 2024-08-30T04:42:07Z

tests/common/platform/device_utils_dpu.py

+    elif power_status == "off":
+        dpu_status_count = dpu_off_count
+
+    return (dpu_status_count == num_modules)


The parentheses are not necessary.

Changed it.

congh-nvidia · 2024-08-30T04:45:19Z

tests/common/platform/device_utils_dpu.py

+    return (ping_count == len(ip_address_list))
+
+
+def check_dpu_module_status(duthost, num_modules, power_status):


It's better to make this function also able to check specific DPU/DUPs.

Changed it.

congh-nvidia · 2024-08-30T04:45:41Z

tests/common/platform/device_utils_dpu.py

+    return (dpu_status_count == num_modules)
+
+
+def check_dpu_reboot_cause(duthost, num_modules):


It's better to make this function also able to check specific DPU/DUPs.

Changed it.

congh-nvidia · 2024-08-30T04:50:28Z

tests/common/platform/device_utils_dpu.py

+        if 'DPU' in parse_output['device'] and parse_output['cause'] == 'Unknown':
+            len_show_cmd += 1
+
+    return (num_modules == len_show_cmd)


The parentheses are not necessary.

Changed it.

congh-nvidia · 2024-08-30T06:05:43Z

tests/platform_tests/cli/test_show_platform_dpu.py

+        interface = parse_output['interface']
+
+        if interface != "eth0" and 'eth' in interface:
+            output_intf_down = duthost.shell("ifconfig %s down"%(interface))["stdout_lines"]


looks like output_intf_down is not used

Changed it.

congh-nvidia · 2024-08-30T08:45:09Z

tests/platform_tests/cli/test_show_platform_dpu.py

+
+    # Deleting the DB table entry after bringing down the interface
+    duthost.shell(
+            "for key in `redis-cli -p 6380 -h 127.0.0.1 -n 13 --scan --pattern \"DPU_STATE|DPU*\" `; "


Why the port is specified here? Is it static?

Yes it is static port.

congh-nvidia · 2024-08-30T09:03:24Z

tests/platform_tests/cli/test_show_platform_dpu.py

+            "do redis-cli -p 6380 -h 127.0.0.1 -n 13 del $key ; done")
+
+    # Checking the state of dpu health again after bringing down interface and deleting db entry
+    dpu_db = wait_until(60, 60, 0, count_dpu_modules_in_system_health_cli, duthost)


Looks like here the expectation is the failure of the wait_until, which actually equals to a 60s sleep, isn't it? And if in the first check the return value of count_dpu_modules_in_system_health_cli is not 0, the wait_until will pass and the following pytest_assert will fail?

Looks like here the expectation is the failure of the wait_until, which actually equals to a 60s sleep, isn't it? Yes

And if in the first check the return value of count_dpu_modules_in_system_health_cli is not 0, the wait_until will pass and the following pytest_assert will fail? Yes

congh-nvidia · 2024-08-30T09:10:20Z

tests/platform_tests/cli/test_show_platform_dpu.py

+            if status == "down/down":
+                count_down += 1
+            output_intf_up = duthost.shell("ifconfig %s up"%(interface))["stdout_lines"]
+            time.sleep(1)


Is 1 second enough for the port to up? Why not combine this sleep with the wait_until check?

We dont need to wait actually, the interface face would be up within a second.

congh-nvidia · 2024-08-30T09:11:12Z

tests/platform_tests/cli/test_show_platform_dpu.py

+                count_up += 1
+
+    # Checking the state of dpu health again after bringing up interface and waiting for db entry to get populated
+    dpu_db_after = wait_until(60, 60, 0, count_dpu_modules_in_system_health_cli, duthost)


If the upper delay is not enough, here could be false negative.

I can increase the delay, but db reflects the changes within a minute, that is why kept at 60 secs.

congh-nvidia · 2024-08-30T09:17:40Z

tests/platform_tests/cli/test_show_platform_dpu.py

+
+    logging.info("Verifying output of '{}' on '{}'...".format(CMD_PCIE_INFO, duthost.hostname))
+    output_pcie_info = duthost.command(CMD_PCIE_INFO)["stdout_lines"]
+    pytest_assert("PASSED" == output_pcie_info[-1], "PCIe Link is good'{}'".format(duthost.hostname))


Isn't here should be a "in" instead of "=="? The last line of the output is like this:
"PCIe Device Checking All Test ----------->>> PASSED"

Changed it.

congh-nvidia · 2024-08-30T09:20:13Z

tests/platform_tests/cli/test_show_platform_dpu.py

+    output_pcie_info = duthost.command(CMD_PCIE_INFO)["stdout_lines"]
+    pytest_assert("PASSED" == output_pcie_info[-1], "PCIe Link is good'{}'".format(duthost.hostname))
+
+    num_modules = int(chassis.get_num_modules(platform_api_conn))


Since this is used every, why not make it a fixture?

made it into a fixture.

congh-nvidia · 2024-08-30T09:24:24Z

tests/platform_tests/cli/test_show_platform_dpu.py

+
+    pytest_assert(wait_until(120, 60, 0, check_dpu_module_status, duthost, num_modules, "off"),
+                          "Not all DPUs operationally down")
+


Should here be a check for the pcie after the DPUs are shutdown?

This check is in the testplan:

1.7 Check removal of pcie link between NPU and DPU

Steps

Use show platform pcieinfo -c to run the pcie info test to check everything is passing

Use command config chassis modules shutdown DPU<DPU_NUM> to bring down the dpu (This will bring down the pcie link between npu and dpu)

Use show platform pcieinfo -c to run the pcie info test to check pcie link has been removed

Use command config chassis modules startup DPU<DPU_NUM> to bring up the dpu (This will rescan pcie links)

Use show platform pcieinfo -c to run the pcie info test to check everything is passing

This test is to check the PCie hot plug functinality since there is no OIR possible

Added the check according to test plan.

congh-nvidia · 2024-08-30T09:30:27Z

tests/platform_tests/test_reload_dpu.py

+    reboot(duthost, localhost, reboot_type=REBOOT_TYPE_COLD, wait_for_ssh=False)
+    wait_for_startup(duthost, localhost, 10, 300)
+    pytest_assert(wait_until(300, 5, 0, check_interface_status_of_up_ports, duthost),
+                          "Not all ports that are admin up on are operationally up")
+    logging.info("Interfaces are up")


Why not directly use a safe_reboot?

sonic-mgmt/tests/common/reboot.py

Line 225 in 7b5156e

safe_reboot=False, check_intf_up_ports=False):

congh-nvidia · 2024-08-30T09:35:40Z

tests/platform_tests/test_reload_dpu.py

+
+    for index in range(len(output_dpu_status)):
+        parse_output = output_dpu_status[index]
+        if parse_output['oper-status'] != 'Offline':


Why some of the DPUs are expected to be in “Offline” status?

Changed it. It was added before creating the power on fixture, hence the confusion.

congh-nvidia · 2024-08-30T09:36:18Z

tests/platform_tests/test_reload_dpu.py

+    for index in range(num_modules):
+        dpu = module.get_name(platform_api_conn, index)
+        ip_address_list.append(module.get_midplane_ip(platform_api_conn, index))
+        duthosts.shell("config chassis modules startup %s"%dpu)
+        time.sleep(2)


Why need to start the DPUs before reload? You already have dpu_poweron_fixture imported.

Changed it.

congh-nvidia · 2024-08-30T09:42:15Z

tests/platform_tests/test_reload_dpu.py

+    logging.info("Reload configuration")
+    duthost.shell("sudo config reload -y &>/dev/null", executable="/bin/bash")
+
+    logging.info("Wait until all critical services are fully started")
+    wait_critical_processes(duthost)


Why not directly use the safe config reload?

sonic-mgmt/tests/common/config_reload.py

Line 110 in 7b5156e

safe_reload=False, wait_before_force_reload=0, wait_for_bgp=False,

Even in the above case safe reload is set to false

vvolam

Please fix these comments. I am in process of reviewing other 2 files also.

vvolam · 2024-10-07T17:27:42Z

tests/smartswitch/common/device_utils_dpu.py

+
+@pytest.fixture(scope='function')
+def num_dpu_modules(platform_api_conn):
+


This new line is not neccessary.

vvolam · 2024-10-07T17:29:04Z

tests/smartswitch/common/device_utils_dpu.py

+            parse_version(duthost.os_version) <= parse_version("202405"):
+        pytest.skip("Test is not supported for this testbed and os version")
+
+    darkmode = is_dark_mode_enabled(duthost, platform_api_conn)


As commented earlier, either rename function name or move these dpu check for darkmode to a new function for providing more clarity?

I changed the function name.

vvolam · 2024-10-07T17:29:31Z

tests/smartswitch/common/device_utils_dpu.py

+
+
+def is_dark_mode_enabled(duthost, platform_api_conn):
+


This new line is not neccessary.

Removed it.

vvolam · 2024-10-07T17:33:11Z

tests/smartswitch/common/device_utils_dpu.py

+        output_config_db = duthost.command(
+                           'redis-cli -p 6379 -h 127.0.0.1 \
+                            -n 4 hgetall "CHASSIS_MODULE|{}"'.format(dpu))
+        if 'down' in output_config_db['stdout']:


It is good to add a check, if the output_config_db is empty for some failure or a reason. Also add a warning log if the output is empty.

vvolam · 2024-10-07T17:34:08Z

tests/smartswitch/common/device_utils_dpu.py

+    Returns the number of DPU modules
+    """
+
+    num_modules = int(chassis.get_num_modules(platform_api_conn))


Add a debug log to log num_modules for easy troubleshooting.

vvolam · 2024-10-07T17:35:08Z

tests/smartswitch/common/device_utils_dpu.py

+                            -n 4 hgetall "CHASSIS_MODULE|{}"'.format(dpu))
+        if 'down' in output_config_db['stdout']:
+            count_admin_down += 1
+


Add debug log here to print dark mode

vvolam · 2024-10-07T17:36:37Z

tests/smartswitch/common/device_utils_dpu.py

+    if count_admin_down == num_modules:
+        return True
+
+    return False


What will happen in case of only few DPUs are in power-on state. Does the tests proceed only on enabled DPUs?

Yes, the test proceeds with enabled dpus.

vvolam · 2024-10-07T18:05:48Z

tests/smartswitch/common/device_utils_dpu.py

+        duthost.shell("config chassis modules startup %s" % (dpu))
+
+    pytest_assert(wait_until(180, 60, 0, check_dpu_ping_status,
+                  duthost, ip_address_list), "Not all DPUs operationally up")


Minor nit: Use "Not all DPUs are operationally up"

Changed it.

vvolam · 2024-10-07T18:06:34Z

tests/smartswitch/common/device_utils_dpu.py

+
+    ping_count = 0
+    for ip_address in ip_address_list:
+        output_ping = duthost.command("ping -c 3 %s" % (ip_address))


Add debug logs for all ping outputs for easy troubleshooting.

vvolam

Remaining PR LGTM. Thanks!

vvolam · 2024-10-07T18:30:49Z

tests/smartswitch/platform_tests/test_reload_dpu.py

+    ip_address_list = []
+    num_modules = num_dpu_modules(platform_api_conn)
+
+    reboot(duthost, localhost, reboot_type=REBOOT_TYPE_COLD,


Are you rebooting entire switch here? If yes, could you add a debug/info log?

Yes, I added the logs as well.

vvolam · 2024-10-08T19:48:50Z

tests/smartswitch/common/device_utils_dpu.py

+    """
+
+    num_modules = int(chassis.get_num_modules(platform_api_conn))
+    logging.info("Num of moduels: '{}'".format(num_modules))


nit: spelling mistake

changed it.

vvolam · 2024-10-08T19:50:15Z

tests/smartswitch/common/device_utils_dpu.py

+                           'redis-cli -p 6379 -h 127.0.0.1 \
+                            -n 4 hgetall "CHASSIS_MODULE|{}"'.format(dpu))
+        if output_config_db['stdout'] is None:
+            logging.warn("redis cli output for chassis module state is empty")


If output_config_db is empty, you need to return False here right?

yes, returning false here. Changed it.

vvolam

Couple of minor suggestions, else LGTM. Thank you!

vvolam

Thank you addressing the comments!

vvolam · 2024-10-09T16:01:49Z

/azpw run

nissampa · 2024-10-09T16:28:13Z

/azpw run

mssonicbld · 2024-10-09T16:28:15Z

/AzurePipelines run

azure-pipelines · 2024-10-09T16:28:27Z

Azure Pipelines successfully started running 1 pipeline(s).

nissampa · 2024-10-10T16:55:58Z

@vvolam and @prgeor,

All PR checker passed and comments are addressed.
Can we close on this ?

vvolam · 2024-10-10T20:03:02Z

@vvolam and @prgeor,

All PR checker passed and comments are addressed. Can we close on this ?

@nissampa I have approved the PR as it looks good to me.

nissampa · 2024-10-17T18:44:25Z

/AzurePipelines run

azure-pipelines · 2024-10-17T18:44:31Z

Commenter does not have sufficient privileges for PR 14152 in repo sonic-net/sonic-mgmt

nissampa · 2024-10-17T18:45:00Z

/azpw run

mssonicbld · 2024-10-17T18:45:03Z

/AzurePipelines run

azure-pipelines · 2024-10-17T18:45:15Z

Azure Pipelines successfully started running 1 pipeline(s).

bpar9

lgtm

* Added DPU platform test cases * modified imports for flake8 test * Changed dirrectory structure for smartswitch test scripts * adderessed few comments * moved tests scripts to platform_tests dir under smartswitch * addressed few set of comments in PR * added logging info * changed ping check * addressed set of comments on PR * flake8 errors resolved for smartswitch/common directory * flake8 error resolved for platform_tests/test_reload_dpu.py * flake8 error resolved for platform_tests/test_show_platform_dpu.py * flake8 error resolved for platform_tests/test_show_platform_dpu.py * flake8 error resolved for platform_tests/test_show_platform_dpu.py * added test skip fixture for smartswitch and non-dark mode * utils flak8 fix * added platform file as variable * missed the \ in added code for platform file * added smartswtich and darkmode detection and lighting up dpus * rectified the syntax error * rectified the syntax error * rectified the syntax error * rectified the syntax error * rectified the syntax error * rectified the syntax error * rectified the syntax error * rectified the syntax error * rectified the syntax error * resolved flake8 errors * added comments for util functions * Made clear with utility function descriptions * retified the error * num of dpu modules fixtures * resolved flake8 errors * resolved flake8 errors * changed link flap case to accomodate bridge interface * removed link flap case * resolved flake8 error * eof fixed * resolving pre-commit check * Addressed set of comments on PR * minor mistakes * added extra line to PR checker * minor change for PR checker * removed functions parameter which are not fixtures

Added DPU platform test cases

e0ebe52

modified imports for flake8 test

8dc74e7

nissampa marked this pull request as ready for review August 20, 2024 16:53

nissampa requested a review from prgeor as a code owner August 20, 2024 16:53

vvolam reviewed Aug 26, 2024

View reviewed changes

congh-nvidia reviewed Aug 30, 2024

View reviewed changes

r12f reviewed Aug 30, 2024

View reviewed changes

congh-nvidia reviewed Aug 30, 2024

View reviewed changes

Changed dirrectory structure for smartswitch test scripts

3f8d0d8

vvolam reviewed Oct 7, 2024

View reviewed changes

Addressed set of comments on PR

bf9c574

vvolam reviewed Oct 8, 2024

View reviewed changes

minor mistakes

ecc60e8

nissampa requested a review from vvolam October 8, 2024 21:07

vvolam approved these changes Oct 8, 2024

View reviewed changes

vvolam requested a review from prgeor October 8, 2024 21:29

added extra line to PR checker

04d829c

nissampa and others added 2 commits October 9, 2024 13:57

Merge branch 'sonic-net:master' into dpu_platform_test_cases

2ab8055

minor change for PR checker

e1c8039

nissampa and others added 2 commits October 17, 2024 10:11

removed functions parameter which are not fixtures

562a595

Merge branch 'sonic-net:master' into dpu_platform_test_cases

2ad407b

bpar9 approved these changes Oct 17, 2024

View reviewed changes

bpar9 merged commit 3ceada7 into sonic-net:master Oct 17, 2024
15 checks passed

		return (ping_count == len(ip_address_list))


		def check_dpu_module_status(duthost, num_modules, power_status):

		return (dpu_status_count == num_modules)


		def check_dpu_reboot_cause(duthost, num_modules):


		pytest_assert(wait_until(120, 60, 0, check_dpu_module_status, duthost, num_modules, "off"),
		"Not all DPUs operationally down")


		@pytest.fixture(scope='function')
		def num_dpu_modules(platform_api_conn):

Added DPU platform test cases #14152

Added DPU platform test cases #14152

Conversation

nissampa commented Aug 16, 2024 • edited Loading

mssonicbld commented Aug 16, 2024

prgeor commented Aug 26, 2024

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

congh-nvidia Aug 30, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nissampa Sep 4, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

1.7 Check removal of pcie link between NPU and DPU

Steps

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

congh-nvidia Aug 30, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

vvolam left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nissampa Oct 7, 2024 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

nissampa commented Aug 16, 2024 •

edited

Loading

congh-nvidia Aug 30, 2024 •

edited

Loading

nissampa Sep 4, 2024 •

edited

Loading

congh-nvidia Aug 30, 2024 •

edited

Loading

nissampa Oct 7, 2024 •

edited

Loading