Bandwidth Test Plugins Design, Switches and Parameters

This section describes plugin specific switches, however, it will not focus on the common switches although these switches will be mentioned here for the completeness of the command examples. To see the common plugin switches and parameters, refer to hl_qual Common Plugin Switches and Parameters.

Note

Before running the plugin tests described in this document, make sure to set export  __python_cmd=python3.8 environment variable.

For PCI Parallel mode testing, python scripts are used rely on the numpy library. The qual package does not install the numpy package, leaving it to the user to install their preferred version. The tested version of the numpy library for Habana is V1.24.3, but newer versions should also be compatible.

Please note that running PCI parallel testing using the memory bandwidth plugin or PCI bandwidth plugins can only be performed on bare metal machines.

Memory Bandwidth Plugin Design Consideration and Responsibilities

Note

The memory bandwidth test plugin is applicable for both first-gen Gaudi and Gaudi2.

The memory bandwidth test plugin is an hl_thunk based DMA bandwidth measurement test. The test includes PCI bandwidth and a variation of memory transfer tests such as HBM and SRAM. The tests are built on top of the hlthunk API - a lower level API wrapping the Habana driver.

Memory Bandwidth Sub-testing Modes

PCI Tests Host-to-Device Communication

HOST ==> HBM

HBM ==> HOST

HBM <==> HOST - bidirectional test

HOST ==> SRAM

SRAM ==> HOST

SRAM <==> HOST - bidirectional test

Device Memory to Memory Tests

SRAM ==> HBM

Using single and multiple DMA channels

HBM ==> SRAM

Using single and multiple DMA channels

HBM ==> HBM

Using single and multiple DMA channels

Serial and Parallel Running Mode Considerations

Since the PCI link is common to all devices, running PCI HOST/DEViCE test variants on multiple devices in parallel mode may result in reduced bandwidth results. To get the most accurate bandwidth result, run the PCI test in serial mode. If you are using Bare Metal setup, you can run the test either in serial or parallel mode.

Memory to memory tests can be run in serial and parallel mode. However, running the tests in different modes is crucial since the memory test takes longer to run, and running the test on multiple devices can decrease the overall execution time.

Split the test into two:

  • Running memory test in parallel mode example:

./hl_qual -c all -rmod parallel -mb -memOnly -gaudi2

  • Running PCI test in serial mode example:

./hl_qual -c all -rmod serial -mb -b -pciOnly -sramOnly -gaudi2

Memory Bandwidth Test - Pass/fail Criteria

The pass/fail criteria depends on the sub-test type executed. Memory to memory tests pass/fail criteria are pre-calibrated.

The PCI sub-test is similar to the SynapseAI PCI load test pass/fail criteria. The following lists some assumptions made in the test plugin code:

  • The full PCI path between HOST CPU and device with predefined Gen-3 setup.

  • The full PCI path between HOST CPU and device is composed out of 16 lanes.

When conducting a test on a different host setup, you can change these assumptions by using the applicable test switches. This will cause the test plugin to change the pass/fail criteria accordingly.

The below pass criteria numbers are precalculated for a setup with 16 lanes and Gen-3 compatibility:

  • Unidirectional download from host to device with an expected bandwidth of 11.9GB/s, assuming CPU with Gen-3 PCI link.

  • Unidirectional upload from device to host with an expected bandwidth of 12.9GB/s, assuming CPU with Gen-3 PCI link.

  • Bidirectional test which calculates the bandwidth on a simultaneous upload and download with an expected bandwidth of 19.5GB/s.

The calculated pass/fail criteria threshold is theoretical, hence the PCI test has a 10% allowable degradation. Below that, the plugin will fail the test run.

Memory Bandwidth Test Plugin Switches and Parameters

hl_qual -gaudi|-gaudi2  -c <pci bus id> -rmod <serial | parallel>  [-dis_mon] [-mon_cfg <monitor INI path>]
         [-n] [-b]  [-memOnly | -pciOnly | -sramOnly | -pciall] -mb [-gen gen_specifier] [-lanes lane_specifier]

Switches and Parameters

Description

-mb

Memory bandwidth test selector.

-n

Cancels checking pass/fail criteria for bandwidth speed.

-b

Activates bidirectional bandwidth tests.

-memOnly

Only activates device memory tests:

  • SRAM ==> HBM

  • HBM ==> SRAM

  • HBM ==> HBM

-pciOnly

Only activates PCI tests:

  • HOST ==> HBM

  • HBM ==> HOST

  • HBM <==> HOST [only active with -b flag]

-sramOnly

Only activates PCI tests:

  • HOST ==> SRAM

  • SRAM ==> HOST

  • SRAM <==> HOST [only active with -b flag]

-pciall

Activates all PCI tests listed in both -pciOnly and -sramOnly.

-gen gen_specifier

Sets up the applicable PCI generation. It can be either gen3 or gen4.

-lanes lane_specifier

Sets up the applicable amount of PCI link lanes. It can be either 8 or 16.

The below command line executes all of the memory bandwidth tests listed above in serial mode, and sets a pass/fail criteria for transfer speed.

./hl_qual -gaudi -c all -rmod serial -dis_mon -b

Note

  • -pciOnly, -memOnly and -sramOnly are optional. When these switches are not specified, a full test run including all available tests is executed. When a full mode run is executed, the test must run in serial mode.

  • -sramOnly and -pciOnly must be run using serial mode, -rmod serial.

  • -memOnly can be run in serial or parallel mode, but any combination of -memOnly that also includes -sramOnly or -pciOnly must be run in serial mode.

  • -pciall executes all tests controlled by -pciOnly and -sramOnly switches.

  • Valid examples:

    • ./hl_qual -c all -rmod parallel -mb -memOnly -gaudi2

    • ./hl_qual -c all -rmod serial -mb -gaudi2

    • ./hl_qual -c all -rmod serial -mb -b -pciall -gaudi2

  • Non-valid examples:

    • ./hl_qual -c all -rmod parallel -mb -gaudi2

    • ./hl_qual -c all -rmod parallel -mb -b -pciOnly  -gaudi2

    • ./hl_qual -c all -rmod parallel -mb -b -pciOnly -sramOnly -gaudi2

PCI Bandwidth Test Design Considerations and Requirements

Note

The PCI bandwidth test plugin is applicable for both first-gen Gaudi and Gaudi2.

The test runs are partitioned into three sub-tests: upload, download and bidirectional.

The PCI bandwidth test plugin measures the PCI bandwidth when moving data from the host to/from the device HBM memory. The hl_qual can run this test using the two running modes - serial and parallel.

Serial Mode

In this testing mode, each device should achieve maximal bandwidth.

Running the test in serial mode imposes certain restrictions on the PCI test time duration. For example, if you run the test for 20 seconds on an 8 first-gen Gaudi machine, the actual test duration will be Total_test duration = 20 * 3 * 8 = 240 seconds.

To average the bandwidth calculation, 10-20 seconds is sufficient (about 200 GB for upload/download). Running the test beyond 1 minute increases the test duration.

Parallel Mode

Use Linux PCIe information for this testing mode. In this scenario, the test compares the card’s performance against the switch/bridge bottle-neck in cards connectivity paths.

The example below illustrates how pass/fail criteria is determined in this mode:

For each path <P> to Habana devices <D1, D2, …> in the PCIe tree:

  1. Extract the bottle-neck <B> switch/bridge/device in P.

  2. If ( BandWidth(B) <= Sum(BandWidth(D1) + BandWidth(D2) + …) ) == True, then PASS else FAIL.

For example:

../../_images/PCIe_tree_example.PNG

Figure 22 Example: PCIe tree

  • Path A : Pass if BandWidth(Second Bridge) <= Sum(BandWidth(D1) + BandWidth(D2)).

  • Path B : Pass if BandWidth(first Bridge) <= Sum(BandWidth(D1)).

  • Path C : Pass if BandWidth(D1) <= Sum(BandWidth(D1)).

  • Path D : Pass if BandWidth(Second Switch) <= Sum(BandWidth(D1) + BandWidth(D2) + BandWidth(D3) + BandWidth(D4)).

Note

The keyword BandWidth refers to the measured bandwidth on the switch/device.

PCI Bandwidth Test - Pass/fail Criteria

The PCI test plugin runs multiple tests to check the download and upload bandwidth with the following pass/fail criteria:

  • Unidirectional download from host to device with an expected bandwidth of 20.9GB/s assuming CPU with Gen-4 PCI link.

  • Unidirectional upload from device to host with an expected bandwidth of 23.1GB/s assuming CPU with Gen-4 PCI link.

  • Bidirectional test which calculates the bandwidth on a simultaneous upload and download with an expected bandwidth of 34.6GB/s assuming CPU with Gen-4 PCI link.

  • Unidirectional download from host to device with an expected bandwidth of 10.4GB/s assuming CPU with Gen-3 PCI link.

  • Unidirectional upload from device to host with an expected bandwidth of 11.5GB/s assuming CPU with Gen-3 PCI link.

  • Bidirectional test which calculates the bandwidth on a simultaneous upload and download with an expected bandwidth of 17.3GB/s assuming CPU with Gen-3 PCI link.

The calculated pass/fail criteria threshold is theoretical, hence the PCI test has a 10% allowable degradation.

PCI Bandwidth Test Plugin Switches and Parameters

hl_qual -gaudi -gaudi2 -c <pci bus id> [-t <time in seconds>]  -rmod <serial | parallel>  [-dis_mon] [-mon_cfg <monitor INI path>]
      -p [-b] [-n] [-size <size in bytes> ] [-gen gen_specifier] [-lanes lane_specifier]

Switches and Parameters

Description

-p

PCI test plugin selector.

-b

Enables bidirectional PCI test, simultaneous upload and download test. By default, if this switch is not specified, the test plugin performs only two upload and download bandwidth tests. When this switch is omitted, the default behavior is to skip the bidirectional test and perform an upload and download bandwidth check.

-n

Disables bandwidth checks. This option is useful when bandwidth calculation in parallel mode is required where all devices are simultaneously being tested. When this switch is omitted, the PCI bandwidth plugin conducts a bandwidth validation check.

-size <buffer size in bytes>

Upload/Download buffer size specification. The minimal buffer size must be 536870912. If this switch is omitted, the default upload/download buffer size is 4GB.

./hl_qual -gaudi -c all -rmod serial -t 20 -p
-b -size 102400000

-gen <gen modifier>

Specifies the expected PCI device generation. There are two applicable modifiers:

  • gen3 - The test is running on a Gen-3 PCI system (Host + Habana device).

  • gen4 - The test is running on a Gen-4 PCI system (Host + Habana device).

./hl_qual -gaudi -c all -rmod serial -t 20 -p -b -gen gen3
./hl_qual -gaudi -c all -rmod serial -t 20 -p -b -gen gen4
./hl_qual -gaudi2 -c all -rmod serial -t 20 -p -b -gen gen4

If this switch is omitted, it is assumed that the system under test (HOST + Habana device) is a Gen-3 PCI data path.

-t

PCI test duration in seconds. If this switch is omitted, the default value is 40 seconds.

Note

  • Since the PCI bandwidth test plugin conducts up to 3 sub-tests (upload, download and bidirectional), the duration given in the command line should be multiplied by 3.

  • For systems with malfunctioning PCI link (such as low bandwidth), the test duration will be different from what was specified by the -t option, as the PCI plugin calculates the number of test iterations based on the expected bandwidth.