QorIQ Optimization Suite (QOS) Packet Analysis Tool

Transcription

1 QorIQ Optimization Suite (QOS) Packet Analysis Tool FTF-SDS-F0004 Petru Lauric Dragos Badea A P R TM External Use

2 Introduction Performance analysis and debug tool designed specifically for the Freescale QorIQ network processors Networking-oriented system-level analysis Packet-focused performance analysis Packet latency measurement SoC data plane configuration debug The speaker is a software developer and a co-author of this tool External Use 1

3 Objectives After completing this session you will be able to: Use the Packet Analysis Tool to collect data from your QorIQ based system Visualize the analysis data collected by the tool Interpret the performance and the behavior of your system in a network application context Measure the packet processing latencies Analyze packet loss issues Debug the data path configuration Integrate the tool into your environment Correlate the tool s analysis data with open-source and third-party tools data External Use 2

4 Agenda Packet Analysis Overview DPAA Trace Packet Latency Analysis Packet Loss Analysis Data Path Configuration Debug User Environment Integration External Use 3

5 Packet Analysis Tool Overview Customer Benefits QorIQ system-level analysis and debug complexity abstraction and ease-of-use Enables key use cases: Packet-Oriented System Level Performance Analysis SoC Data Plane Configuration Debug Packet Processing Latency Analysis Packet Processing Critical Resource Monitoring Target areas: SoC debug/analysis feature enablement Linux Systems Analysis data interpretation Visualization Users Network application developers QA Customer support teams External Use 4

6 Packet Analysis Tool Customer Benefits Provides Packet-Oriented, System-Level Analysis Functionality Key Benefits: Abstracts the complexity of the DPAA debug resources Automatically identifies the frame processing paths (data flows) of the system Provides visibility into networking tasks offloaded to SoC Non-intrusive or low-intrusive data collection Remotely analyze systems deployed in the field, without interruption External Use 5

7 Packet Analysis Tool What Is the Packet-Oriented Analysis? Traditional analysis tools: Software-centric: software debuggers, software profilers, program trace, etc. Hardware-centric: provide low-level hardware data (e.g. status info) Difficult or impossible to associate this data with the network packets processed by the device under test (DUT) Networking-specific analysis tools Traffic-centric: packet log analysis (Wireshark), network performance measurement (iperf), etc. Difficult or impossible to associate this data with the internals of the DUT New paradigm Packet-centric analysis of the internals of the DUT Allows integrating with other analysis tools and technologies (e.g. software-centric or traffic-centric) External Use 6

8 Packet Analysis Tool - Key Use Cases Packet Tracing Packet Loss Analysis Latency Analysis Packet Sequence Analysis Data Flow Level Analysis Remote System Analysis Load Balance Analysis Shows which parts of the system process the frames. For example, use this to verify that the frame flow is what you expect Understand why the frames become lost in the system. For example, use this to check how the FM PCD changes affect where frames are sent Precisely measure the time spent processing frames at various points in the system See how an entire sequence of frames was processed. For example, use this to measure the performance of the SEC Use QM profile data to measure the performance of the system, at data-flow level Analyze and debug remote QorIQ devices, such as systems deployed in the field Measure how the load is distributed across the system External Use 7

9 Packet Analysis Tool Product Summary Supported Designs: P4080, P3041, P5020, P5040, P2040 T2080 T4240 B4860rev2 Target device OS agnostic supports both Linux and bare board targets Supported host operating systems: Windows XP Professional (SP3), Win 7, 32-bit and 64-bit, Win 8 Red Hat Enterprise Linux 5.4, 32-bit and 64-bit, or later Ubuntu and (32-bit and 64-bit) SuSE 11, 32-bit External Use 8

10 Packet Analysis Tool DPAA Overview The tool only supports QorIQ devices which implement DPAA The tool user must have basic knowledge of DPAA architecture and applications. For details, search freescale.com for the QORIQDPAAWP keyword Queue Manager Buffer Manager Ethernet Interfaces Frame Manager SEC PME RMan SRIO Interfaces CoreNet Memory L2$ Core Core Core Core Core Core Core Core Core L2$ D$ Core I$ Core L2$ D$ Core I$ Core L2$ D$ Core I$ Core L2$ D$ I$ L2$ D$ I$ L2$ D$ I$ L2$ D$ I$ D$ I$ Acceleration of frame/packet processing Network protocols (Layer-1 to 4) Crypto and pattern-matching hardware accelerators Classification and distribution of data flows among cores and software partitions Load balancing through parse/classify/distribute Load spreading through queues shared among multiple consumers Abstract and manage efficiently intercore communications and the access to shared resources (NW interfaces, HW-Accelerators, Buffers, Queues) More sophisticated approach compared to basic BD/buffer list Scalability and Portability Across any mix of cores, accelerators and device boundaries Across device generations External Use 9

11 Packet Analysis Tool DPAA Trace Visibility into FM and QM activities via Nexus Trace FM Trace Optionally output by each FM engine: BMI, KeyGen, Parser, etc. The Packet Analysis Tool s trace configuration settings control which FM engines output trace and with what verbosity Timestamped internally by the FM clock The trace data contains: FD, FM port number, NIA, etc. QM Trace Optionally output by each QM enqueue and dequeue point The Packet Analysis Tool s trace configuration settings control which QM enqueue/dequeue points output trace and with what verbosity The trace data contains: FQID, channel, frame address, frame length, enqueue/dequeue flag, portal type and number, etc. External Use 10

12 Packet Analysis Tool DPAA Trace (continued) Traced Frames Only the tagged frames are traced. Tagged = FD[DD] bits set Rx flow the frames can be automatically tagged by the FM, as configured by the Packet Analysis Tool The Packet Analysis Tool s trace configuration settings control which frames are tagged by FM, to achieve trace data filtering in silicon Other flows the frames tagged by the instrumented software running on the cores External Use 11

13 Packet Analysis Tool DPAA Trace (continued) The DPAA Trace Data Can be collected from a running system Without interrupting it Without affecting its performance Can be collected non-intrusively to on-chip trace buffer (max 32kB for T4240 or B4860) Larger amounts of trace data can be collected to a buffer located in the external memory (DDR); the intrusiveness depends on the system s load and on the amount of trace data Timestamped in silicon so it can be used to precisely measure the timings External Use 12

14 Packet Latency Analysis External Use 13

15 Packet Latency Analysis Setup PC connected to T4240 board through a network switch PC running Packet Analysis Tool QorIQ T4240 device under test Run the cryptography test application using the crypto hardware accelerator (SEC) Run the debug agent External Use 14

16 Packet Analysis Tool - Create a New Configuration Press the toolbar button to create a New Configuration Use the default configuration and project settings New Configuration Button External Use 15

17 Packet Analysis Tool - Configuration A default configuration is added to a project. The configuration is displayed in the lower right of the workbench Project Navigation Configuration External Use 16

18 Packet Analysis Tool - Configure the Connection To setup the configuration, first configure the connection. Use the Connection toolbar to add a connection from scratch or to search for connections External Use 17

19 Packet Analysis Tool Debug Agent Connection For remote access (e.g. to systems deployed in the field) The debug agent s source code is provided with the tool Low intrusiveness the agent is used to: Configure the trace settings at the beginning of the test Retrieve the trace data at the end of the test Select Debug Agent for the connection type Specify the hostname or IP address and port number where the agent is located Debug Agent Connection Type QorIQ Linux Device Hostname and Port External Use 18

20 Packet Analysis Tool Select Measurement Scenarios Below the Connection settings, select the scenario to be used during the measurement session Use the toolbar to Add or Remove Scenarios The DPAA All Scenario is selected by default. Recommended for starting your analysis External Use 19

21 Packet Analysis Tool What is a Scenario? A scenario defines what will be measured Defines a DPAA trace configuration Defines which target systems the trace configuration is compatible with The scenario is a user-editable Python script The DPAA trace configuration API used by the script is designed to hide complexity The tool ships with a set of predefined scenarios. The users can modify them or add new ones. External Use 20

22 Predefined Scenarios Scenario Name DPAA All DPAA All, Enable FM Timestamps DPAA FM1 DPAA FM2 DPAA QM Decode Only Description Overview of all of the DPAA activity FM and QM, all portals, etc. Same as above and it enables the FM s timestamp counter, which is used to timestamp the FM trace, for measuring the FM latencies Overview of FM1 activity, FM1 marks the frames for debug and generates trace Overview of FM2 activity, FM2 marks the frames for debug and generates trace FM1 and FM2 mark the frames for debug but don t generate trace. Only QM generates trace Decodes an existing DPAA binary trace file External Use 21

23 Packet Analysis Tool Using the DPAA QM Scenario To start measuring, press the Launch button Select the DPAA QM scenario External Use 22

24 Packet Analysis Tool Trace Buffer Configuration Trace buffer types On-Chip up to 32kB buffer (T4240 and B4860) for nonintrusive trace collection DDR trace buffer located in the system s external memory, large capacity DDR use intrusiveness depends on the system s load and on the amount of trace data Use Uboot to reserve memory for it (see the user manual) Auto vs. manual data collection start One-shot vs. circular collection mode External Use 23

25 Packet Analysis Tool Data Measurement Below is a snapshot of the Workbench views that show the status updates as the data is collected. Use the toolbar in the Analysis Session view to Start, Pause, or Stop the data collection View Overall Session Status Start/Pause/Stop Data Collection View Detailed Progress and Buffer Status External Use 24

26 Packet Analysis Tool Flow Level Analysis Processing Path Entity representing a data flow Used to group stats for related frames (similar lifetime) Automatically identified by the Packet Analysis Tool by analyzing the trace data, no software instrumentation requirements Frames are tracked based on the address from trace Example: SwPortal1 => QM => SEC => QM => SwPortal1 External Use 25

27 Packet Analysis Tool Analysis View Data Flows From Trace Data Flow Level Latency Statistics Sum of total time spent on this path Total number of frames processed on this path Frame Processing Stages Path 2 Sequence = (#1: SwPortal5 => QM), (#2: QM), (#3: QM => SEC), (#4: SEC), (#5: SEC => QM), (#6: QM), (#7: QM => SwPortal5) External Use 26

28 Packet Analysis Tool Processing Path Compare Compare path stats to determine weight of each data flow External Use 27

29 Packet Analysis Tool DPAA Subsystem Analysis Compare DPAA engine (subsystem) level stats to search for bottlenecks Observe frame processing latency on SEC engine and the QM stage before SEC processing Updated for the currently selected processing path External Use 28

30 Packet Analysis Tool Frame Details View Frame Address Frame Processing Latencies (e.g. observe the increasing SEC latency) View Filter Frame Processing Details Frame Processing Path External Use 29

31 Packet Analysis Tool Frame Lifetimes View Traced Frames Frame Timeline Timeline Details DPAA Subsystems Engine Utilization Statistics External Use 30

32 Packet Analysis Tool FM Latency Analysis Same Timestamp Due to Buffering Inside FM Timestamps Assigned Internally, by FM Use local timestamps to compute FM Latency Start: Frame Was Received by FM2 End: Frame Is Enqueued by FM2 External Use 31

33 Packet Loss Analysis External Use 32

34 Packet Loss Analysis Setup PC connected to T4240 board through a network switch PC running Packet Analysis Tool PC running the iperf traffic generator, in both server and client mode QorIQ T4240 device under test Run a packet reflector application JTAG probe connected to the device External Use 33

35 Packet Analysis Tool JTAG Probe Connection The Packet Analysis Tool needs to connect to the QorIQ target device in order to: Configure the DPAA trace settings Collect the trace data For lab use: JTAG probe (TAP) connected to the QorIQ device Backdoor access mechanism QorIQ device operating system agnostic External Use 34

36 Packet Analysis Tool JTAG Connection Autodiscovery Upon selecting the flashlight button in the Connection toolbar, the tool automatically discovers the available JTAG probes on the network Select a connection. Press the OK button to add it to the Configuration Filter Connections External Use 35

37 Packet Analysis Tool - Target Autodiscovery After adding the connection, click the Scan Probe link to automatically check the type and the availability of the target device. Discover Target After the scan is complete, the target s processor type and current status are displayed. Target is Available External Use 36

38 Packet Analysis Tool Traffic Generation Setup Run the packet loopback application on the QorIQ device For example, use the Freescale USDPAA Reflector application Run the server-side iperf instance, on the PC: iperf -s -i 1 Run the client-side iperf instance, on the PC: iperf -c <QorIQ device hostname> -i 1 -t 60 The packets will be sent from the iperf client instance to the QorIQ device The QorIQ device will reflect the packets to the iperf server iperf will report the bandwidth and packet loss statistics External Use 37

39 Packet Analysis Tool Using the DPAA QM Scenario To start measuring, press the Launch button Select the DPAA QM scenario External Use 38

40 Packet Analysis Tool Packet Loss Analysis iperf output [ ] [ ID] Interval Transfer Bandwidth Jitter Lost/Total Datagrams [ 4] sec 8.55 MBytes 71.8 Mbits/sec ms 2034/ 8136 (25%) iperf reported dropped packets The processing paths identified by the Packet Analysis Tool show where the packets are dropped The code servicing SwPortal1 drops packets External Use 39

41 Data Path Configuration Debug External Use 40

42 Data Path Configuration Debug PC connected to P4080 board through a network switch PC running Packet Analysis Tool PC pings the QorIQ DUT QorIQ P4080 device under test Run a packet reflector application JTAG probe connected to the device External Use 41

43 PCD Configuration Enable FM policer: three-color marking algorithm for frames matching the hash_ipv4_src_dst_dist9 distribution rule [ ] [ ] [ ] <policer name="policer_1g"> <algorithm>rfc2698</algorithm> <color_mode>color_blind</color_mode> <CIR> </CIR> <EIR> </EIR> <CBS> </CBS> <EBS> </EBS> <unit>packet</unit> </policer> <distribution name="hash_ipv4_src_dst_dist9"> <queue count="32" base="0xd00"/> <key> <fieldref name="ipv4.src"/> <fieldref name="ipv4.dst"/> </key> <action name="policer_1g" type="policer"/> </distribution> Policer action description <policy name="hash_ipsec_src_dst_spi_policy9"> <dist_order> <distributionref name="hash_ipv4_src_dst_dist9"/> <distributionref name="default_dist9"/> </dist_order> </policy> FQIDs for the distribution under analysis Policer action Use the fmc tool to apply the FM configuration External Use 42

44 Data Path Configuration Debug with Packet Analysis Tool Use the DPAA All scenario to capture both QM and FM trace data Identify the frames for which Parse Classify Distribute (PCD) is to be analyzed Step 2: Use the frame address and the FM enqueue timestamp to correlate with FM trace for deeper investigation Step 1: Use frame processing path info and FQID to identify frames of interest External Use 43

45 PCD Configuration Validation Using the FM Trace Step 2: Use the frame address in FM trace before the enqueue event to identify the FM PCD actions Step 3: Verify the PCD actions are expected for the applied configuration Step 1: Use the frame address and timestamp to identify the FM enqueue event of interest External Use 44

46 User Environment Integration External Use 45

47 Packet Analysis Tool User Environment Integration The Packet Analysis Tool can be easily integrated with Traffic generators Open-source network analysis tools (e.g. WireShark) Automated test frameworks Main integration goals: Provide insight into the internal performance and behavior of the QorIQ device Correlate the QorIQ performance and debug data with other analysis data: network packet statistics, test reports, validation criteria, etc. External Use 46

48 Packet Analysis Tool User Environment Integration Several options exist for integrating the Packet Analysis Tool into the users environments The hardware data retrieved by the Packet Analysis Tool can be correlated and analyzed in parallel with other data (e.g. WireShark s) The command line interface can be used to collect trace data while the users run their own debug or performance tests, possibly automated The binary trace data file(s) can be analyzed later using the GUI features External Use 47

49 Packet Analysis Tool WireShark Integration Full details regarding using the Packet Analysis Tool with Wireshark are included in application note AN4811 Two data sets need to be correlated: the log of network packets (analyzed by Wireshark) and the hardware trace (analyzed by the Packet Analysis Tool) One easy correlation method: use ICMP packet markers that can be easily found in both data sets (e.g. because they have certain sizes) At the beginning of the test, inject ICMP packet with size X At the end of the test, inject ICMP packet with size Y External Use 48

50 Packet Analysis Tool WireShark Correlated Analysis Matching Lengths Packet Marker Based Correlation External Use 49

51 Packet Analysis Tool WireShark Correlated Analysis (continued) After Correlation, compare Hardware stats with Network Traffic stats. See application note AN4811 for details External Use 50

52 Packet Analysis Tool - Command Line Interface All functionality provided by the GUI is also accessible using the CLI DDR trace collection example for Windows, Linux is similar: <PacketAnalysisTool>\cdde\bin$ python.bat dpaa_scripts\analysis\dpaatraceanalysis.py -config=dpaa_scripts\scenarios\dpaaall.py -buffer=ddr -bufferstart=0x buffersize=0x out=dpaatrace.txt profileout=dpaaprofile.txt -tap=gtap:jtag:1: system=p4080 -config specifies the scenario script -tap specifies the JTAG probe type and address or the debug agent s hostname/ip address and port -system specifies the QorIQ device type External Use 51

53 Packet Analysis Tool - Command Line Interface (continued) Timestamp Trace Event Index Decoded Trace Output Sample 1 : : Trace Client 11 : Verbose mode. Direct connect portal 0x0 (FM1). Debug Tag = 1. Enqueue Command Dispatched trace event. Portal was not halted in response to this queue operation. Order Restoration was not specified at enqueue. The frame enqueue was not deferred due to order restoration. Enqueue operation was not rejected. FQID = 0x107. Channel = 0x0. Frame Status/Command = 0x0. Frame Format = 0x0 (short, single buffer, simple). Frame offset = 0x80 (128), frame length = 0xaa (170). Complete LIODN Offset = 0x5c. Buffer Pool ID = 0x20. Address = 0xdbaa : : Trace Client 11 : Verbose mode. Software portal 0x30. Debug Tag = 1. Dequeue trace event. Portal was not halted in response to this queue operation. Order Restoration was not specified at enqueue. The frame enqueue was not deferred due to order restoration. Enqueue operation was not rejected. FQID = 0x107. Channel = 0x401. Frame Status/Command = 0x0. Frame Format = 0x0 (short, single buffer, simple). Frame offset = 0x80 (128), frame length = 0xaa (170). Complete LIODN Offset = 0x5c. Buffer Pool ID = 0x20. Address = 0xdbaa8500. External Use 52

54 Packet Analysis Tool - Command Line Interface (continued) ######################################################################## # DPAA TRACE-BASED PROFILE # (Note: the reported latency is computed based on the timestamp field # in Nexus messages generated at QM enqueue and dequeue points) ######################################################################## Profiler Output Sample NUMBER OF TRACED FRAMES: 88 NUMBER OF BYTES IN TRACED FRAMES: 9760 Path Level Stats PROCESSING PATH 1: 4 frames(4.55%), 434 bytes(4.45%) Processing sequence: (#1: FM1 => QM), (#2: QM), (#3: QM => SwPortal48) (#1: FM1 => QM) 0 enqueue rejects (0.00%) (#2) QM Scheduling latency(min/max/avg): (54/56/54) [platform cycles] PROCESSING PATH 2: 4 frames(4.55%), 320 bytes(3.28%) Processing sequence: (#1: FM1 => QM), (#2: QM), (#3: QM => SwPortal49) (#1: FM1 => QM) 0 enqueue rejects (0.00%) (#2) QM Scheduling latency(min/max/avg): (54/56/54) [platform cycles] External Use 53

55 Packet Analysis Tool - Command Line Interface (continued) [ ] Profiler Output Sample (cont.) Frame Level Details ######################################################################## # DPAA TRACED FRAMES DETAILS: ######################################################################## Frame 0 (frame address: 0xdbaa8500, frame length: 170 bytes, tracked on processing path 1) Processing sequence: (FM1 => QM), (QM => SwPortal48) Timestamp/Latency_to_prior_TS: Operation / -: enqueue by FM1 (FQID 263, CID 0) / 56: dequeue by SwPortal48 (FQID 263, CID 1025) Frame 1 (frame address: 0xe , frame length: 60 bytes, tracked on processing path 2) Processing sequence: (FM1 => QM), (QM => SwPortal49) Timestamp/Latency_to_prior_TS: Operation / -: enqueue by FM1 (FQID 263, CID 0) / 54: dequeue by SwPortal49 (FQID 263, CID 1025) External Use 54

56 Packet Analysis Tool CLI Python Scripting The tool is based on Python scripts which use a set of components The DPAA debug hardware has a very rich set of features The trace configuration API available from the Python scripts abstracts the DPAA debug resources and groups them into sets The configuration API is documented in the user manual The result is a very user-friendly and powerful programming interface External Use 55

57 Packet Analysis Tool CLI Trace Configuration Example of code (part of a scenario) which: Enables the output of QM trace for all the QM enqueue and dequeue points Configures the FMs to assign debug tags for all frames on Rx flow Disables the output of FM trace def setdpaaconfiguration(self, dpaa): # user defined DPAA API-based trace configuration comes here # example: all_dbg_chains = dpaa.getalldbgchains() # initialize: match all frames on the first debug chain and set maximum trace verbosity firstdbgchain = all_dbg_chains.find("dbgchain1") firstdbgchain.cnd(1) firstdbgchain.acttrace(firstdbgchain.trace_verb_high) #disable trace generation on FM1 and FM2 allfms = dpaa.getallfm() fm1 = allfms.find("fm1") fm1.acttrace(fm1.trace_verb_none) if allfms.size() > 1: fm2 = allfms.find("fm2") fm2.acttrace(fm2.trace_verb_none) return External Use 56

58 Packet Analysis Tool CLI Analysis Data Access The Python interface also provides access to the QorIQ DPAA analysis data displayed in the tool s graphical user interface or text output files This makes it possible to integrate the tool with automated test frameworks For example, this can be used to test if the QorIQ device enqueues the frames to the expected frame queues Full details are provided in application note AN Network Performance Investigation External Use 57

59 Packet Analysis Tool Frame Queue Validation Test Setup PC connected to T4240 board through a network switch PC running Packet Analysis Tool PC running the iperf traffic generator, in both server and client mode QorIQ T4240 device under test Run a packet loopback application JTAG probe connected to the device External Use 58

60 Packet Analysis Tool Traffic Generation Setup Run the packet loopback application on the QorIQ device For example, use the Freescale USDPAA Reflector application Run the server-side iperf instance, on the PC: iperf -s -i 1 Run the client-side iperf instance, on the PC: iperf -c <QorIQ device hostname> -i 1 -t 60 The packets will be sent from the iperf client instance to the QorIQ device The QorIQ device will reflect the packets to the iperf server External Use 59

61 Packet Analysis Tool CLI Frame Queue Validation Follow the instructions from application note AN Network Performance Investigation - to modify the Python script which provides access to the analysis data: [ ] #list of FQIDs we expect the frames to be enqueued to MY_FQIDS = [9000, 9001, 9240, 9241,] [ ] if fqid not in MY_FQIDS: printbadframeenqueue( ) Run the Packet Analysis Tool using the CLI If any frames are enqueued to unexpected frame queues, the results will be displayed at the console For example: Frame at address 0xbc31e300 was enqueued to frame queue 9248 by DCP2 External Use 60

62 Packet Analysis Tool User Documentation The Packet Analysis Tool ships with a rich set of user documentation The User Manual documents: the GUI features the command line interface the trace configuration API mentioned above Application notes: AN Packet Latency Measurement AN SEC Latency Measurement AN DPAA Configuration Debug AN Network Traffic Analysis Using the Packet Analysis Tool and Wireshark AN Network Performance Investigation External Use 61

63 Packet Analysis Tool - Download Want to download the latest version of the tool? Search freescale.com for packet analysis OR Use the link below: PE_QORIQ_DPAA_PACKET&fsrch=1&sr=6 External Use 62

64 Packet Analysis Tool Summary Packet-centric system-level analysis of QorIQ device performance and behavior (unique in the industry, to our knowledge) Very easy to use, yet very powerful Remote device (deployed system) analysis Plug-and-play functionality - connect to a running system and analyze it without interrupting it, without instrumenting the software running on it, etc. Easy to integrate with other tools and into the user s debug and test environment Scripting interface for extending the functionality External Use 63

65 Introducing The QorIQ LS2 Family Breakthrough, software-defined approach to advance the world s new virtualized networks New, high-performance architecture built with ease-of-use in mind Groundbreaking, flexible architecture that abstracts hardware complexity and enables customers to focus their resources on innovation at the application level Optimized for software-defined networking applications Balanced integration of CPU performance with network I/O and C-programmable datapath acceleration that is right-sized (power/performance/cost) to deliver advanced SoC technology for the SDN era Extending the industry s broadest portfolio of 64-bit multicore SoCs Built on the ARM Cortex -A57 architecture with integrated L2 switch enabling interconnect and peripherals to provide a complete system-on-chip solution External Use 64

66 QorIQ LS2 Family Key Features SDN/NFV Switching Data Center Wireless Access Unprecedented performance and ease of use for smarter, more capable networks High performance cores with leading interconnect and memory bandwidth 8x ARM Cortex-A57 cores, 2.0GHz, 4MB L2 cache, w Neon SIMD 1MB L3 platform cache w/ecc 2x 64b DDR4 up to 2.4GT/s A high performance datapath designed with software developers in mind New datapath hardware and abstracted acceleration that is called via standard Linux objects 40 Gbps Packet processing performance with 20Gbps acceleration (crypto, Pattern Match/RegEx, Data Compression) Management complex provides all init/setup/teardown tasks Leading network I/O integration 8x1/10GbE + 8x1G, MACSec on up to 4x 1/10GbE Integrated L2 switching capability for cost savings 4 PCIe Gen3 controllers, 1 with SR-IOV support 2 x SATA 3.0, 2 x USB 3.0 with PHY External Use 65

67 See the LS2 Family First in the Tech Lab! 4 new demos built on QorIQ LS2 processors: Performance Analysis Made Easy Leave the Packet Processing To Us Combining Ease of Use with Performance Tools for Every Step of Your Design External Use 66

68 Freescale Semiconductor, Inc. External Use