From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga17.intel.com (mga17.intel.com [192.55.52.151]) by dpdk.org (Postfix) with ESMTP id CC6CF5F34 for ; Tue, 20 Mar 2018 04:00:58 +0100 (CET) X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga002.jf.intel.com ([10.7.209.21]) by fmsmga107.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 19 Mar 2018 20:00:57 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.48,333,1517904000"; d="scan'208";a="43538038" Received: from unknown (HELO dpdk-fedora20.icx.intel.com) ([10.240.176.135]) by orsmga002.jf.intel.com with ESMTP; 19 Mar 2018 20:00:56 -0700 From: Peng Yuan To: dts@dpdk.org Cc: Peng Yuan Date: Tue, 20 Mar 2018 11:02:12 +0800 Message-Id: <1521514932-88552-1-git-send-email-yuan.peng@intel.com> X-Mailer: git-send-email 1.9.3 Subject: [dts] [PATCH v2] test_plans: add test plan of bbdev X-BeenThere: dts@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: test suite reviews and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 20 Mar 2018 03:00:59 -0000 BBDEV allow DPDK as a 3gpp Layer 1 signal processing acceleration solution. Signed-off-by: Peng Yuan 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. + Now we only support virtual device, In the case of the hardware, of course + there will be change in results. This is the aim of having hardware + acceleration to get the latency and throughput improved. + +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(GCC compiler can be used too.) + +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