* [dts] [PATCH] test_plans: add test plan of bbdev
@ 2018-02-07 2:58 Peng Yuan
2018-02-11 1:57 ` Liu, Yong
0 siblings, 1 reply; 2+ messages in thread
From: Peng Yuan @ 2018-02-07 2:58 UTC (permalink / raw)
To: dts; +Cc: Peng Yuan
Signed-off-by: Peng Yuan <yuan.peng@intel.com>
diff --git a/test_plans/bbdev_test_plan.rst b/test_plans/bbdev_test_plan.rst
new file mode 100644
index 0000000..93ba6a7
--- /dev/null
+++ b/test_plans/bbdev_test_plan.rst
@@ -0,0 +1,203 @@
+.. Copyright (c) <2018>, Intel Corporation
+ All rights reserved.
+
+ Redistribution and use in source and binary forms, with or without
+ modification, are permitted provided that the following conditions
+ are met:
+
+ - Redistributions of source code must retain the above copyright
+ notice, this list of conditions and the following disclaimer.
+
+ - Redistributions in binary form must reproduce the above copyright
+ notice, this list of conditions and the following disclaimer in
+ the documentation and/or other materials provided with the
+ distribution.
+
+ - Neither the name of Intel Corporation nor the names of its
+ contributors may be used to endorse or promote products derived
+ from this software without specific prior written permission.
+
+ THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
+ FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
+ COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
+ INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+ (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+ SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
+ STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
+ OF THE POSSIBILITY OF SUCH DAMAGE.
+
+=============================================================
+Wireless device for ICX-D (bbdev) for Turbo decoding/encoding
+=============================================================
+Description
+===========
+
+ The Wireless Baseband library provides a common programming framework that
+ abstracts HW accelerators based on FPGA and/or Fixed Function Accelerators
+ that assist with 3gpp Physical Layer processing. Furthermore, it decouples
+ the application from the compute-intensive wireless functions by abstracting
+ their optimized libraries to appear as virtual bbdev devices.
+ The functional scope of the BBDEV library are those functions in relation to
+ the 3gpp Layer 1 signal processing (channel coding, modulation, ...).
+ The framework currently only supports Turbo Code FEC function.
+ The Wireless Baseband library follows the same ideology of DPDK's Ethernet
+ Device and Crypto Device frameworks. Wireless Baseband provides a generic
+ acceleration abstraction framework which supports both physical (hardware)
+ and virtual (software) wireless acceleration functions.
+
+Prerequisites
+=============
+
+1. OS and Hardware
+ (a) An AVX2 supporting machine
+ (b) Windriver TS 2 or CentOS 7 operating systems
+ (Fedora 25 and Ubuntu 16.04 is ok.)
+ (c) Intel ICC compiler installed
+
+2. FlexRAN SDK Libraries
+ To build DPDK with the *turbo_sw* PMD the user is required to download
+ the export controlled ``FlexRAN SDK`` Libraries.
+ An account at Intel Resource Design Center needs to be registered from
+ https://www.intel.com/content/www/us/en/design/resource-design-center.html
+ Direct download link from Intel RDC ->
+ https://cdrd.intel.com/v1/dl/getContent/575367
+ After download is complete, the user needs to unpack and compile on their
+ system before building DPDK.
+ You can refer to the file dpdk/doc/guides/bbdevs/turbo_sw.rst.
+
+3. PMD setting
+ Current BBDEV framework is en-suited with two vdev PMD drivers:
+ null and turbo_sw.
+ 1) Null PMD is similar to cryptodev Null PMD, which is an empty driver to
+ measure the overhead added by the framework.
+ 2) Turbo_sw is a sw-only driver wrapper for FlexRAN SDK optimized Turbo
+ coding libraries.
+ It can be enabled by setting
+
+ ``CONFIG_RTE_LIBRTE_PMD_BBDEV_TURBO_SW=y``
+
+ It is located in the build configuration file ``common_base``.
+
+4. Test tool
+ A test suite for BBDEV is packaged with the framework to ease the
+ validation needs for various functions and use cases.
+ The tool to use for validation and testing is called: test-bbdev,
+ that is packaged with test vectors that are ready-to-use.
+ Test-bbdev tool is located at this location:
+ app/test-bbdev/
+ The command-line options you can refer to:
+ dpdk/doc/guides/tools/testbbdev.rst
+
+Test case 1: bbdev null device
+==============================
+
+Executing bbdev null device with *bbdev_vector_null.data* helps in measuring
+the overhead introduced by the bbdev framework::
+
+ ./test-bbdev.py -e="--vdev=bbdev_null0"
+ -v ./test_vectors/bbdev_vector_null.data
+
+bbdev_null device does not have to be defined explicitly as it is created
+by default. so the command-line can be written as::
+
+ ./test-bbdev.py -v ./test_vectors/bbdev_vector_null.data
+
+the case only cover unittest, all the other cases are skipped.
+
+Test case 2: Turbo encoding validation
+======================================
+
+It runs **validation** test for Turbo encode vector file
+Number of operations to process on device is set to 64
+and operations timeout is set to 120s
+and enqueue/dequeue burst size is set to 8 and to 32.
+Moreover a bbdev (*turbo_sw*) device will be created::
+
+ ./test-bbdev.py -p ../../x86_64-native-linuxapp-icc/app/testbbdev
+ -e="--vdev=turbo_sw" -t 120 -c validation
+ -v ./test_vectors/bbdev_vector_te_default.data -n 64 -b 8 32
+
+you can check the result from the detailed printing, and compare
+the two results from different burst size.
+
+Test case 3: Turbo decoding validation
+======================================
+
+It runs **validation** test for Turbo decode vector file
+we use default options::
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c validation
+ -v ./test_vectors/bbdev_vector_td_default.data
+
+you can check the result from the detailed printing.
+
+Test case 4: Turbo encoding latency
+===================================
+
+It runs **latency** test for Turbo encode vector file::
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c latency
+ -v ./test_vectors/bbdev_vector_te_default.data -n 512 -b 64 512
+
+Test calculates three latency metrics:
+1) offload_latency_tc
+2) offload_latency_empty_q_tc
+3) operation_latency_tc
+You can compare the three latency from different burst size.
+
+Test case 5: Turbo decoding latency
+===================================
+
+It runs **latency** test for Turbo decode vector file::
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c latency
+ -v ./test_vectors/bbdev_vector_td_default.data -n 512 -b 64
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c latency
+ -v ./test_vectors/bbdev_vector_td_default.data -n 128 -b 64
+
+You can compare the three latency from different number of operations.
+
+Test case 6: Turbo encoding throughput
+======================================
+
+It runs **throughput** test for Turbo encode vector file::
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
+ -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 64
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
+ -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 8 -b 64
+
+You can compare the turbo encode throughput from different number of lcores.
+then different burst size::
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
+ -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 64
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
+ -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 32
+
+You can compare the turbo encode throughput from different burst size.
+then different number of operations::
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
+ -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 32
+
+ ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
+ -v ./test_vectors/bbdev_vector_te_default.data -n 32 -l 16 -b 32
+
+You can compare the turbo encode throughput from different number of
+operations.
+
+Test case 7: Turbo decoding throughput
+======================================
+
+It runs **throughput** test for Turbo decode vector file.
+you can run the three subcases of test case 6 which just needed to
+change the test vector file to "bbdev_vector_td_default.data", and
+compare the results.
--
2.5.0
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [dts] [PATCH] test_plans: add test plan of bbdev
2018-02-07 2:58 [dts] [PATCH] test_plans: add test plan of bbdev Peng Yuan
@ 2018-02-11 1:57 ` Liu, Yong
0 siblings, 0 replies; 2+ messages in thread
From: Liu, Yong @ 2018-02-11 1:57 UTC (permalink / raw)
To: Peng, Yuan, dts; +Cc: Peng, Yuan
Yuan,
Some comments are inline.
Thanks,
Marvin
> -----Original Message-----
> From: dts [mailto:dts-bounces@dpdk.org] On Behalf Of Peng Yuan
> Sent: Wednesday, February 07, 2018 10:59 AM
> To: dts@dpdk.org
> Cc: Peng, Yuan <yuan.peng@intel.com>
> Subject: [dts] [PATCH] test_plans: add test plan of bbdev
>
> Signed-off-by: Peng Yuan <yuan.peng@intel.com>
>
> diff --git a/test_plans/bbdev_test_plan.rst
> b/test_plans/bbdev_test_plan.rst
> new file mode 100644
> index 0000000..93ba6a7
> --- /dev/null
> +++ b/test_plans/bbdev_test_plan.rst
> @@ -0,0 +1,203 @@
> +.. Copyright (c) <2018>, Intel Corporation
> + All rights reserved.
> +
> + Redistribution and use in source and binary forms, with or without
> + modification, are permitted provided that the following conditions
> + are met:
> +
> + - Redistributions of source code must retain the above copyright
> + notice, this list of conditions and the following disclaimer.
> +
> + - Redistributions in binary form must reproduce the above copyright
> + notice, this list of conditions and the following disclaimer in
> + the documentation and/or other materials provided with the
> + distribution.
> +
> + - Neither the name of Intel Corporation nor the names of its
> + contributors may be used to endorse or promote products derived
> + from this software without specific prior written permission.
> +
> + THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
> + "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
> + LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
> + FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
> + COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
> + INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
> + (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
> + SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
> + HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
> + STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
> + ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED
> + OF THE POSSIBILITY OF SUCH DAMAGE.
> +
> +=============================================================
> +Wireless device for ICX-D (bbdev) for Turbo decoding/encoding
> +=============================================================
> +Description
> +===========
> +
> + The Wireless Baseband library provides a common programming framework
> that
> + abstracts HW accelerators based on FPGA and/or Fixed Function
> Accelerators
> + that assist with 3gpp Physical Layer processing. Furthermore, it
> decouples
> + the application from the compute-intensive wireless functions by
> abstracting
> + their optimized libraries to appear as virtual bbdev devices.
> + The functional scope of the BBDEV library are those functions in
> relation to
> + the 3gpp Layer 1 signal processing (channel coding, modulation, ...).
> + The framework currently only supports Turbo Code FEC function.
> + The Wireless Baseband library follows the same ideology of DPDK's
> Ethernet
> + Device and Crypto Device frameworks. Wireless Baseband provides a
> generic
> + acceleration abstraction framework which supports both physical
> (hardware)
> + and virtual (software) wireless acceleration functions.
> +
Please remove the spaces in the frontend of each line.
> +Prerequisites
> +=============
> +
> +1. OS and Hardware
> + (a) An AVX2 supporting machine
> + (b) Windriver TS 2 or CentOS 7 operating systems
> + (Fedora 25 and Ubuntu 16.04 is ok.)
> + (c) Intel ICC compiler installed
> +
Is this suite only work with ICC or also can work with GCC?
> +2. FlexRAN SDK Libraries
> + To build DPDK with the *turbo_sw* PMD the user is required to download
> + the export controlled ``FlexRAN SDK`` Libraries.
> + An account at Intel Resource Design Center needs to be registered from
> + https://www.intel.com/content/www/us/en/design/resource-design-
> center.html
> + Direct download link from Intel RDC ->
> + https://cdrd.intel.com/v1/dl/getContent/575367
> + After download is complete, the user needs to unpack and compile on
> their
> + system before building DPDK.
> + You can refer to the file dpdk/doc/guides/bbdevs/turbo_sw.rst.
> +
> +3. PMD setting
> + Current BBDEV framework is en-suited with two vdev PMD drivers:
> + null and turbo_sw.
> + 1) Null PMD is similar to cryptodev Null PMD, which is an empty driver
> to
> + measure the overhead added by the framework.
> + 2) Turbo_sw is a sw-only driver wrapper for FlexRAN SDK optimized
> Turbo
> + coding libraries.
> + It can be enabled by setting
> +
> + ``CONFIG_RTE_LIBRTE_PMD_BBDEV_TURBO_SW=y``
> +
> + It is located in the build configuration file ``common_base``.
> +
> +4. Test tool
> + A test suite for BBDEV is packaged with the framework to ease the
> + validation needs for various functions and use cases.
> + The tool to use for validation and testing is called: test-bbdev,
> + that is packaged with test vectors that are ready-to-use.
> + Test-bbdev tool is located at this location:
> + app/test-bbdev/
> + The command-line options you can refer to:
> + dpdk/doc/guides/tools/testbbdev.rst
> +
> +Test case 1: bbdev null device
> +==============================
> +
> +Executing bbdev null device with *bbdev_vector_null.data* helps in
> measuring
> +the overhead introduced by the bbdev framework::
> +
> + ./test-bbdev.py -e="--vdev=bbdev_null0"
> + -v ./test_vectors/bbdev_vector_null.data
> +
> +bbdev_null device does not have to be defined explicitly as it is created
> +by default. so the command-line can be written as::
> +
> + ./test-bbdev.py -v ./test_vectors/bbdev_vector_null.data
> +
> +the case only cover unittest, all the other cases are skipped.
Is there any expectation of the overhead on certain platform, or just record this number?
> +
> +Test case 2: Turbo encoding validation
> +======================================
> +
> +It runs **validation** test for Turbo encode vector file
> +Number of operations to process on device is set to 64
> +and operations timeout is set to 120s
> +and enqueue/dequeue burst size is set to 8 and to 32.
> +Moreover a bbdev (*turbo_sw*) device will be created::
> +
> + ./test-bbdev.py -p ../../x86_64-native-linuxapp-icc/app/testbbdev
> + -e="--vdev=turbo_sw" -t 120 -c validation
> + -v ./test_vectors/bbdev_vector_te_default.data -n 64 -b 8 32
> +
> +you can check the result from the detailed printing, and compare
> +the two results from different burst size.
> +
> +Test case 3: Turbo decoding validation
> +======================================
> +
> +It runs **validation** test for Turbo decode vector file
> +we use default options::
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c validation
> + -v ./test_vectors/bbdev_vector_td_default.data
> +
> +you can check the result from the detailed printing.
> +
> +Test case 4: Turbo encoding latency
> +===================================
> +
> +It runs **latency** test for Turbo encode vector file::
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c latency
> + -v ./test_vectors/bbdev_vector_te_default.data -n 512 -b 64 512
> +
> +Test calculates three latency metrics:
> +1) offload_latency_tc
> +2) offload_latency_empty_q_tc
> +3) operation_latency_tc
> +You can compare the three latency from different burst size.
> +
> +Test case 5: Turbo decoding latency
> +===================================
> +
> +It runs **latency** test for Turbo decode vector file::
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c latency
> + -v ./test_vectors/bbdev_vector_td_default.data -n 512 -b 64
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c latency
> + -v ./test_vectors/bbdev_vector_td_default.data -n 128 -b 64
> +
> +You can compare the three latency from different number of operations.
Not sure why decoding not need range the descriptor number.
> +
> +Test case 6: Turbo encoding throughput
> +======================================
> +
> +It runs **throughput** test for Turbo encode vector file::
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
> + -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 64
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
> + -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 8 -b 64
> +
> +You can compare the turbo encode throughput from different number of
> lcores.
> +then different burst size::
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
> + -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 64
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
> + -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 32
> +
> +You can compare the turbo encode throughput from different burst size.
> +then different number of operations::
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
> + -v ./test_vectors/bbdev_vector_te_default.data -n 64 -l 16 -b 32
> +
> + ./test-bbdev.py -e="--vdev=turbo_sw" -c throughput
> + -v ./test_vectors/bbdev_vector_te_default.data -n 32 -l 16 -b 32
> +
> +You can compare the turbo encode throughput from different number of
> +operations.
> +
> +Test case 7: Turbo decoding throughput
> +======================================
> +
> +It runs **throughput** test for Turbo decode vector file.
> +you can run the three subcases of test case 6 which just needed to
> +change the test vector file to "bbdev_vector_td_default.data", and
> +compare the results.
> --
> 2.5.0
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2018-02-11 1:57 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-02-07 2:58 [dts] [PATCH] test_plans: add test plan of bbdev Peng Yuan
2018-02-11 1:57 ` Liu, Yong
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).