Re: [PATCH v2] app/eventdev: add crypto producer mode

[Aaron Conole <aconole@redhat.com>]

Shijith Thotton <sthotton@marvell.com> writes:

>>> >>
>>> >> In crypto producer mode, producer core enqueues cryptodev with
>>> >> software generated crypto ops and worker core dequeues crypto
>>> >> completion events from the eventdev. Event crypto metadata used for
>>> >> above processing is pre- populated in each crypto session.
>>> >>
>>> >> Parameter --prod_type_cryptodev can be used to enable crypto producer
>>> >mode.
>>> >> Parameter --crypto_adptr_mode can be set to select the crypto adapter
>>> >> mode,
>>> >0
>>> >> for OP_NEW and 1 for OP_FORWARD.
>>> >>
>>> >> This mode can be used to measure the performance of crypto adapter.
>>> >>
>>> >> Example:
>>> >>   ./dpdk-test-eventdev -l 0-2 -w <EVENTDEV> -w <CRYPTODEV> -- \
>>> >>   --prod_type_cryptodev --crypto_adptr_mode 1 --test=perf_atq \
>>> >>   --stlist=a --wlcores 1 --plcores 2
>>> >
>>> >This patch has some perf failure as shown below. Could you please look into
>>> this?
>>> >105300 --> performance testing fail
>>> >
>>> >Test environment and result as below:
>>> >
>>> >Ubuntu 20.04
>>> >Kernel: 4.15.0-generic
>>> >Compiler: gcc 7.4
>>> >NIC: Intel Corporation Ethernet Converged Network Adapter 82599ES 10000
>>> >Mbps
>>> >Target: x86_64-native-linuxapp-gcc
>>> >Fail/Total: 0/4
>>> >
>>> >Detail performance results:
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| frame_size | txd/rxd | num_cpus | num_threads |  throughput
>>> >| difference from
>>> >|
>>> >|            |         |          |             |           expected           |
>>>
>>>+============+=========+==========+=============+============
>>=
>>> >=================+
>>> >| 64         | 512     | 1        | 1           | 0.3%                         |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 2048    | 1        | 1           | -0.2%                        |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 512     | 1        | 2           | 0.0%                         |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 2048    | 1        | 2           | 0.3%                         |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >
>>> >Ubuntu 20.04
>>> >Kernel: 4.15.0-generic
>>> >Compiler: gcc 7.4
>>> >NIC: Intel Corporation Ethernet Converged Network Adapter XL710-QDA2
>>> >40000 Mbps
>>> >Target: x86_64-native-linuxapp-gcc
>>> >Fail/Total: 1/4
>>> >
>>> >Detail performance results:
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| frame_size | txd/rxd | num_cpus | num_threads |  throughput
>>> >| difference from
>>> >|
>>> >|            |         |          |             |           expected           |
>>>
>>>+============+=========+==========+=============+============
>>=
>>> >=================+
>>> >| 64         | 512     | 1        | 1           | 0.2%                         |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 2048    | 1        | 1           | -0.7%                        |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 512     | 1        | 2           | -1.5%                        |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 2048    | 1        | 2           | -5.3%                        |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >
>>> >Ubuntu 20.04 ARM
>>> >Kernel: 4.15.0-132-generic
>>> >Compiler: gcc 7.5
>>> >NIC: Arm Intel Corporation Ethernet Converged Network Adapter
>>> >XL710-QDA2
>>> >40000 Mbps
>>> >Target: x86_64-native-linuxapp-gcc
>>> >Fail/Total: 0/2
>>> >
>>> >Detail performance results:
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| frame_size | txd/rxd | num_cpus | num_threads |  throughput
>>> >| difference from
>>> >|
>>> >|            |         |          |             |           expected           |
>>>
>>>+============+=========+==========+=============+============
>>=
>>> >=================+
>>> >| 64         | 512     | 1        | 1           | 0.1%                         |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >| 64         | 2048    | 1        | 1           | -0.5%                        |
>>> >+------------+---------+----------+-------------+------------------------------+
>>> >
>>> >To view detailed results, visit:
>>> >https://urldefense.proofpoint.com/v2/url?u=https-
>>>
>>>3A__lab.dpdk.org_results_dashboard_patchsets_20534_&d=DwIFAg&c=nKjWe
>>> c2
>>> >b6R0mOyPaz7xtfQ&r=G9w4KsPaQLACBfGCL35PtiRH996yqJDxAZwrWegU2qQ&
>>> m=
>>>
>>>dPDgmKiLC_KqhgjwZrodyFlBdlFPgckVr9IW0c7t2cIHpkw6u9pGqgElFqqwuU1v&s
>>> =E
>>> >ajYUFziZY27q6LB4p5sOGuFlty3VlOjbMsLXaQ5MeI&e=
>>> >
>>>
>>> Performance regression is reported by nic_single_core_perf, but is using
>>> testpmd to measure performance. As this patch does not touch testpmd, the
>>> issue could be with DPDK CI. Is it possible to re-trigger CI ?
>>I have don't have option to trigger CI from my side.
>>So, I have added dpdklab and Aaron to the loop.
>>Please check with them.
>>
>  
> Thanks Abhinandan.
>
> Aaron,
> From DTS, I only see testpmd being used for nic_single_core_perf test.
> This patch is not touching testpmd, but still got failure for the perf test.
> If possible, Please re-retrigger CI for this patch.

Seems this was superseded - I am guessing no new test needs to be run
for this:

https://patches.dpdk.org/project/dpdk/patch/736dc424a780b9547033f76f8911342ff06d3f87.1641292088.git.sthotton@marvell.com/

Is that correct?