Patent Application
20200200819
The present disclosure relates generally to the field of electronic device testing systems and more specifically to the field of electronic device testing equipment for testing devices under test (DUTs).
Automated test equipment (ATE) can be any testing assembly that performs a test on a semiconductor device or electronic assembly. ATE assemblies may be used to execute automated tests that quickly perform measurements and generate test results that can then be analyzed. An ATE assembly may be anything from a computer system coupled to a meter, to a complicated automated test assembly that may include a custom, dedicated computer control system and many different test instruments that are capable of automatically testing electronics parts and/or semiconductor wafer testing, such as system-on-chip (SOC) testing or integrated circuit testing. ATE systems both reduce the amount of time spent on testing devices to ensure that the device functions as designed and serve as a diagnostic tool to determine the presence of faulty components within a given device before it reaches the consumer.
The ATE body 100 tests the electrical functions of the DUTs 112A-112N connected to the ATE body 100 through hardware bus adapters plugged into the hardware bus adapter sockets of the ATE body 100. Accordingly, the tester processor 101 is programmed to communicate the test programs needed to be run to the DUTs using the protocol unique to the hardware bus adapters. Meanwhile, the other hardware components built into the ATE body 100 communicate signals with each other and with the DUTs according to test programs operating in the tester processor 101.
The test program run by the tester processor 101 may include a function test which involves writing input signals created by the pattern generator 102 to the DUTs, reading out the written signals from the DUTs and using the comparator 106 to compare the output with the expected patterns. If the output does not match the input, the tester processor 101 will identify the DUT as being defective. For example, if the DUT is a memory device such as a DRAM, the test program will write data generated by the pattern generator 102 to the DUT using a Write Operation, read data from the DRAM using a Read Operation and compare the expected bit pattern with the read pattern using the comparator 106.
In conventional systems, the tester processor 101 needs to contain the functional logic blocks to generate the commands and test patterns used in testing the DUTs, such as the pattern generator 102 and the comparator 106, programmed in software directly on the processor. However, in some instances certain functional blocks such as the comparator 106 may be implemented on a field programmable gate array (FPGA), which is an application specific integrated circuit (ASIC) type semiconductor device that can program logic circuits according to a user's demand.
The FPGAs used in conventional systems rely on the tester processor 101 to transfer the commands and test patterns to the FPGA, which the FPGA in turn relays over to the DUTs. Because the tester processor, and not the FPGA, is responsible for generating the commands and test patterns, the number and type of DUTs that can be tested with a given ATE body is limited by the processing capabilities and programming of the tester processor. Where the tester processor generates all the commands and test patterns, bandwidth constraints on the system bus 130 connecting the tester processor to the various hardware components, including any FPGA devices and hardware bus adapter sockets, also places an upper limit on the number of DUTs that can tested simultaneously.
Also, in conventional systems, the communication protocol used to communicate with the DUTs is fixed because the hardware bus adapter cards that plug into the ATE body 100 are single purpose devices that are designed to communicate in only one protocol and cannot be reprogrammed to communicate in a different protocol. For example, an ATE body configured to test PCIe devices will have hardware bus adapter cards plugged into the body that support only the PCIe protocol. In order to test DUTs supporting a different protocol, e.g., SATA the user would ordinarily need to replace the PCIe hardware bus adapter cards with bus adapter cards supporting the SATA protocol. Unless the PCIe hardware bus adapter cards are physically substituted with cards supporting the other protocol, such a system can only test DUTs that support the PCIe protocol. Thus, on the test floor, critical time is consumed replacing hardware bus adapter cards when DUTs running a different protocol from the one that the existing adapter cards support need to be tested.
Another drawback of current tester systems is that the test systems do not presently support testing of DUTs (e.g., SSDs) that support protection modes and odd sector sizes. For example, most of the solid state drives (SSDs) available on the market are formatted for standard sector sizes, e.g., 4K native, 512 native. A 4K native is a drive with sectors that contain 4096 (4K) bytes (4096 logical bytes per sector). A 512 native is a drive with sectors that contain 512 bytes (512 logical bytes per sector). Most modern hard drives use one of the two sector sizes: 512 bytes per sector or 4096 bytes per sector. However, certain vendors also support non-standard sector sizes, e.g., 520 bytes per sector (BPS) or 528 bytes per sector (BPS) especially for drives to be used in enterprise-grade systems. Moreover, certain vendors include additional bytes of metadata in each sector to support parity checking, checksums, and/or cyclic redundancy checksums (CRCs) (e.g., 512+8 BPS where 512 bytes are reserved for data and 8 bytes are reserved for metadata). In other words, the additional metadata included within each sector can enable a protection mode for the SSD. Conventional ATE is limited because it does not support testing of drives that support non-standard sector sizes. Furthermore, conventional ATE also does not support testing of SSDs that enable protection modes, e.g., SSDs that contain additional metadata per sector for supporting parity checking and/or checksums.
Accordingly, a need exists for a tester architecture that can address the problems with the systems described above. What is needed is a test architecture whereby the command and test pattern generation functionality can be transferred onto the FPGA, so that the processing load on the tester processor and the bandwidth requirements on the system bus can be kept at a minimum. This would be more efficient than prior configurations where the tester processor bore the entire processing load and the system bus conveyed test data and commands for all the DUTs connected to the ATE body. Further, what is needed is a test architecture whereby the communicative protocol engine can be programmed on FPGA devices so that the protocol used to communicate with the DUTs is reconfigurable.
In addition, what is needed is a tester system whereby there is support at both the software and the firmware level for testing DUTs (e.g. SSDs) with all combinations of odd sector sizes and all combinations of protection modes. For example, SSDs may have non-standard sector sizes of 520 BPS, 4104 BPS, 528 BPS and 4224 BPS. Furthermore, SSDs may also have an extra 8 or more bytes of metadata per sector to enable certain protection modes. What is needed is a tester system that is able to seamlessly and efficiently test SSDs with non-standard sector sizes and with various protection modes enabled.
In one embodiment, an automated test equipment (ATE) apparatus is presented. The apparatus comprises a computer system comprising a system controller, wherein the system controller is communicatively coupled to a tester processor and an FPGA wherein the system controller is operable to transmit instructions to the tester processor, and wherein the tester processor is operable to generate commands and data from the instructions for coordinating testing of a device under test (DUT). The DUT supports a plurality of odd sector sizes and a plurality of protection modes. The FPGA is communicatively coupled to the tester processor, wherein the FPGA comprises at least one hardware accelerator circuit operable to internally generate commands and data transparently from the tester processor for testing the DUT, and wherein the at least one hardware accelerator circuit is able to perform computations to calculate protection information associated with the plurality of protection modes and to generate repeatable test patterns sized to fit each of the plurality of odd sector sizes. Further, the tester processor is configured to operate in one of a plurality of functional modes, wherein each functional mode is configured to allocate functionality for generating commands and for generating data between the tester processor and the FPGA in a different manner.
In another embodiment, a method for testing using an automated test equipment (ATE) is disclosed. The method comprises transmitting instructions from a system controller of a computer system to a tester processor, wherein the system controller is communicatively coupled to a the tester processor and an FPGA, wherein the tester processor is operable to generate commands and data from the instructions for coordinating testing of a plurality of DUTs, wherein each of the plurality of DUTs supports a plurality of odd sector sizes and a plurality of protection modes. The method further comprises generating commands and data transparently from the tester processor for testing of a plurality of DUTs using a hardware accelerator circuit programmed within an FPGA, wherein the FPGA is communicatively coupled to the tester processor and wherein the hardware accelerator circuit is operable to test the plurality of DUTs, and wherein the hardware accelerator circuit is able to perform computations to calculate protection information associated with the plurality of protection modes and to generate repeatable test patterns sized to fit each of the plurality of odd sector sizes. The method additionally comprises operating the tester processor in one of a plurality of functional modes, wherein each functional mode is configured to allocate functionality for generating commands and for generating data between the tester processor and the FPGA in a different manner.
In another embodiment, a tester is disclosed. The tester comprises a system controller for controlling a test program for testing a plurality of DUTs, wherein each of the plurality of DUTs support a plurality of odd sector sizes and a plurality of protection modes. Further, the tester comprises a primitive comprising a plurality of tester slices, wherein each tester slice comprises a tester processor coupled to communicate with the system controller to receive instructions and data therefrom in accordance with the test program and at least one programmable instantiated tester block coupled to the tester processor, wherein the at least one programmable instantiated tester block is operable to generate test data for application to the plurality of DUTs in a way transparent to the test processor, further operable to receive and compare test data generated by the plurality of DUTs in a way transparent to the tester processor. The instantiated tester block is further yet operable to be programmed to communicate with the plurality of DUTs in a communication protocol compatible with the plurality of DUTs. Additionally, the at least one programmable instantiated tester block comprises a hardware accelerator circuit operable to perform computations to calculate protection information associated with the plurality of protection modes and to generate repeatable test patterns sized to fit each of the plurality of odd sector sizes. The tester also comprises a local memory coupled to the programmable instantiated tester block for storing test data therein and a device interface board for coupling the plurality of tester slices to the plurality of DUTs. Furthermore, each programmable instantiated tester block is operable in one of a plurality of functional modes, wherein each functional mode is configured to allocate functionality for generating commands and for generating data between the tester processor and the programmable instantiated tester block in a different manner.
The following detailed description together with the accompanying drawings will provide a better understanding of the nature and advantages of the present invention.
Embodiments of the present invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which like reference numerals refer to similar elements.
In the figures, elements having the same designation have the same or similar function.
Reference will now be made in detail to embodiments, examples of which are illustrated in the accompanying drawings. While the embodiments will be described in conjunction with the drawings, it will be understood that they are not intended to limit the embodiments. On the contrary, the embodiments are intended to cover alternatives, modifications and equivalents. Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a thorough understanding. However, it will be recognized by one of ordinary skill in the art that the embodiments may be practiced without these specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail as not to unnecessarily obscure aspects of the embodiments.
Some regions of the detailed descriptions which follow are presented in terms of procedures, logic blocks, processing and other symbolic representations of operations on data bits within a computer memory. These descriptions and representations are the means used by those skilled in the data processing arts to most effectively convey the substance of their work to others skilled in the art. In the present application, a procedure, logic block, process, or the like, is conceived to be a self-consistent sequence of steps or instructions leading to a desired result. The steps are those requiring physical manipulations of physical quantities. Usually, although not necessarily, these quantities take the form of electrical or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated in a computer system.
It should be borne in mind, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise as apparent from the following discussions, it is appreciated that throughout the present invention, discussions utilizing the terms such as “accessing,” “adding,” “adjusting,” “analyzing,” “applying,” “assembling,” “assigning,” “calculating,” “capturing,” “combining,” “comparing,” “collecting,” “creating,” “debugging,” “defining,” “depicting,” “detecting,” “determining,” “displaying,” “establishing,” “executing,” “generating,” “grouping,” “identifying,” “initiating,” “modifying,” “monitoring,” “moving,” “outputting,” “performing,” “placing,” “presenting,” “processing,” “programming,” “querying,” “removing,” “repeating,” “resuming,” “sampling,” “simulating,” “sorting,” “storing,” “subtracting,” “suspending,” “tracking,” “transforming,” “unblocking,” “using,” or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.
The description below provides a discussion of computers and other devices that may include one or more modules. As used herein, the term “module” or “block” may be understood to refer to software, firmware, hardware, and/or various combinations thereof. It is noted that the blocks and modules are exemplary. The blocks or modules may be combined, integrated, separated, and/or duplicated to support various applications. Also, a function described herein as being performed at a particular module or block may be performed at one or more other modules or blocks and/or by one or more other devices instead of or in addition to the function performed at the described particular module or block. Further, the modules or blocks may be implemented across multiple devices and/or other components local or remote to one another. Additionally, the modules or blocks may be moved from one device and added to another device, and/or may be included in both devices. Any software implementations of the present invention may be tangibly embodied in one or more storage media, such as, for example, a memory device, a floppy disk, a compact disk (CD), a digital versatile disk (DVD), or other devices that may store computer code.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present invention. As used throughout this disclosure, the singular forms “a,” “an,” and “the” include plural reference unless the context clearly dictates otherwise. Thus, for example, a reference to “a module” includes a plurality of such modules, as well as a single module, and equivalents thereof known to those skilled in the art.
Automated Test Equipment (ATE) Support Framework for Solid State Device (SSD) Odd Sector Sizes and Protection Modes
Embodiments of the present invention allow test throughput to be improved by including operating system (OS) support within the tester for testing DUTs (e.g., SSDs) supporting different (odd or non-standard) sector sizes and protection modes without reconfiguring the hardware. More specifically, the calculations and computing required to support odd sector sizes (also known as non-standard sector sizes) and various protection modes may be performed by hardware accelerators implemented on FPGA devices. In other words, computation functionality formerly performed in software on a general-purpose tester processor is transferred to hardware accelerators implemented on FPGA devices. Another way is by increasing the number and types of devices under test (DUTs) that can be tested under prevailing hardware and time constraints, for example, by configuring the hardware so that DUTs supporting many different types of protection modes and sector sizes can be tested with the same hardware without needing to replace or substitute any hardware components. Embodiments of the present invention are directed to so improving test efficiency in the hardware of the automatic test equipment.
Referring to
In one embodiment, the system controller 201 may be a computer system, e.g., a personal computer (PC) that provides a user interface for the user of the ATE to load the test programs and run tests for the DUTs connected to the ATE 200. In one embodiment, the system controller 201 may be running the Windows operation system (OS). The Verigy Stylus software executing in the Windows environment is one example of test software normally used during device testing. It provides the user with a graphical user interface from which to configure and control the tests. It can also comprise functionality to control the test flow, control the status of the test program, determine which test program is running, and log test results and other data related to test flow. In one embodiment, the system controller can be connected to and control as many as 512 DUTs.
In one embodiment, the system controller 201 can be connected to the site module boards 230A-230N through a network switch, such as an Ethernet switch. In other embodiments, the network switch may be compatible with a different protocol such as TCP/IP, Fibre Channel, 802.11 or ATM, for instance.
In one embodiment, each of the site module boards 230A-230N may be a separate standalone board used for purposes of evaluation and development that attaches to custom-built load board fixtures, on which the DUTs 220A-220N are loaded, and also to the system controller 201 from where the test programs are received. In other embodiments, the site module boards may be implemented as plug-in expansion cards or as daughter boards that plug into the chassis of the system controller 201 directly. Alternatively, the site module boards may be housed within an enclosure of a primitive (as shown in
The site module boards 230A-230N can each comprise at least one tester processor 204 and at least one FPGA device. In one embodiment, the tester processor and its associated memory may be located on a separate board (not shown) affixed to the respective site module. This separate board may be called a Computer On Module (or COM) board. In other words, the FPGA may be located on a separate site module board while the tester processor (with an associated memory) is located on a COM board.
The tester processor 204 and the FPGA devices 211A-211M on the site module board run the test methods for each test case in accordance with the test program instructions received from the system controller 201. In one embodiment the tester processor can be a commercially available Intel x86 CPU or any other well-known processor. Further, the tester processor may be operating on the Ubuntu OS x64 operating system and running the Core Software, which allows it to communicate with the software running on the system controller, to run the test methods. In one embodiment, the tester processor 204 may be an x86 processor running the Linux OS or a modified version of the Linux OS. In one embodiment, the Linux OS running on the tester processor is able to receive commands and data from the Windows OS running on the system controller. The tester processor 204 controls the FPGA devices on the site module and the DUTs connected to the site module based on the test program received from the system controller.
The tester processor 204 is connected to and can communicate with the FPGA devices over bus 212. In one embodiment, tester processor 204 communicates with each of the FPGA devices 211A-211M over a separate dedicated bus. In one embodiment, for example in the standard or bypass mode, tester processor 204 can control the testing of the DUTs 220A-220N transparently through the FPGAs with minimal processing functionality allocated to the FPGA devices. In this embodiment, the data traffic capacity of bus 212 can be exhausted rapidly because all the commands and data generated by the tester processor need to be communicated over the bus to the FPGA devices. In other embodiments, the tester processor 204 can share the processing load by allocating functionality to control the testing of the DUTs to the FPGA devices, e.g., in protocol independent data accelerations (PIDA) or full acceleration (FA) modes as will be discussed further below. In these embodiments, the traffic over bus 212 is reduced because the FPGA devices can generate their own commands and data.
In one embodiment, each of the FPGA devices 211A-211M is connected to its own dedicated memory block 240A-240M. These memory blocks can, among other things, be utilized to store the test pattern data that is written out to the DUTs. In one embodiment, each of the FPGA devices can comprise two instantiated FPGA tester blocks 210A-210B with functional modules for performing functions including implementation of communicative protocol engines and hardware accelerators as described further herein. Memory blocks 240A-240 M can each contain one or more memory modules, wherein each memory module within the memory block can be dedicated to one or more of the instantiated FPGA tester blocks 210A-210B. Accordingly, each of the instantiated FPGA tester blocks 210A-210B can be connected to its own dedicated memory module within memory block 240A. In another embodiment, instantiated FPGA tester blocks 210A and 210B can share one of the memory modules within memory block 240A. In a different embodiment, each FPGA device can have multiple instantiated FPGA tester blocks, each with a respective memory block.
Further, each of the DUTs 220A-220N in the system can be connected to a dedicated instantiated FPGA tester block 210A-210N in a “tester per DUT” configuration, wherein each DUT gets its own tester block. This allows separate test execution for each DUT. The hardware resources in such a configuration are designed in a manner to support individual DUTs with minimal hardware sharing. This configuration also allows many DUTs to be tested in parallel, where each DUT can be connected to its own dedicated FPGA tester block and be running a different test program. In a different embodiment, each instantiated FPGA tester block may also be connected to and configured to test multiple DUTs.
The architecture of the embodiment of the present invention depicted in
In one embodiment, new protocols can be downloaded and installed directly on the FPGAs via a simple bit-stream download from a cache on system controller 201 without any kind of hardware interactions. An FPGA will typically include a configurable interface core (or IP core) that is programmable to provide functionality of one or more protocol based interfaces for a DUT and is programmable to interface with the DUT. For example, the FPGAs 211A-211M in the ATE apparatus 200 will include an interface core that can be configured with the PCIe protocol to test PCIe devices initially and subsequently reconfigured via a software download to test SATA devices. Also, if a new protocol is released, the FPGAs can easily be configured with that protocol via a bit-stream download instead of having to physically switch all the hardware bus adapter cards in the system. Finally, if a non-standard protocol needs to be implemented, the FPGAs can nonetheless be configured to implement such a protocol.
In another embodiment, the FPGAs 211A-211M can be configured to run more than one communicative protocol, wherein these protocols also can be downloaded from system controller 201 and configured through software. In other words, each FPGA implements custom firmware and software images to implement the functionality of one or more PC based testers in a single chip. The required electrical signaling and protocol-based signaling is provided by on-chip IP cores in the FPGAs. As mentioned above, each FPGA is programmable with pre-verified interface or IP cores. This ensures compliance and compatibility according to a given interface standard. The programmable nature of the FPGA is utilized to optimize flexibility, cost, parallelism and upgradeabilty for storage testing applications from SSDs, HDDs and other protocol based storage devices.
For instance, instantiated FPGA tester block 210A can be configured to run the PCIe protocol while instantiated FPGA tester block 210B can be configured to run the SATA protocol. This allows the tester hardware to test DUTs supporting different protocols simultaneously. FPGA 211A can now be connected to test a DUT that supports both PCIe and SATA protocols. Alternatively, it can be connected to test two different DUTs, one DUT supporting the PCIe protocol and the other DUT supporting the SATA protocol, where each instantiated functional module (e.g., 210A, 210B) is configured with a protocol to test the respective DUT it is connect to.
In one embodiment, the interface or IP core in the FPGA may be acquired from a third party vendor but may require some customization to be compatible with the embodiments described herein. In one embodiment, the interface core provides two functions: a) wraps storage commands into a standard protocol for transmission over a physical channel; and 2) is the electrical signal generator and receiver.
The other major advantage of the architecture presented in
Referring to
As mentioned above, in one embodiment, the system controller 301 may be a computer system, e.g., a personal computer (PC) that provides a user interface for the user of the ATE to load the test programs and run tests for the DUTs connected to the ATE 300. Typically the system controller will run the Windows operating system. The Advantest Stylus is one example of test software normally used during device testing. It provides the user with a graphical user interface from which to configure and control the tests. It can also comprise functionality to control the test flow, control the status of the test program, determine which test program is running, and log test results and other data related to test flow.
Each of the device power supply boards 332A-332B can be controlled from one of the site modules 310A-310B. The software running on the tester processor 304 can be configured to assign a device power supply to a particular site module. In one embodiment, the site modules 310A-310B and the device power supplies 332A-332B are configured to communicate with each other using a high speed serial protocol, e.g., Peripheral Component Interconnect Express (PCIe), Serial AT Attachment (SATA) or Serial Attached SCSI (SAS), for instance.
In one embodiment, each site module is configured with two FPGAs as shown in
FPGAs 316 and 318 are connected to memory modules 308 and 304 respectively, where the memory modules perform a similar function to memory blocks 240A-240N in
FPGAs 316 and 318 can be connected to the DUTs 372A-372M on the load board 380 through buses 352 and 354 respectively. The load board 380 is a physical harness that allows a general purpose high speed connection at the site module end that is agnostic to the protocol used to communicate to the DUTs in on lines 352 and 354. At the DUT end, however, the load board needs to be designed so as to have connectors specific to the protocol being used by the DUT.
The DUTs 372A-372M, in one embodiment of the invention, are loaded on a load board 380 that is placed inside a thermal chamber 390 for testing. The DUTs 372A-372M and the load board 380 derive power from the device power supplies 332A and 332B.
The number of DUTs that can be connected to each FPGA is contingent on the number of transceivers in the FPGA and the number of I/O lanes required by each DUT. In one embodiment, FPGAs 316 and 318 can each comprise 32 high speed transceivers and buses 352 and 354 can each be 32 bits wide, however, more or less can be implemented depending on the application. If each DUT requires 8 I/O lanes, for example, only 4 DUTs can be connected to each FPGA in such a system.
In one embodiment, the communication protocol used to communicate between the tester processor 304 and the DUTs 372A-M can advantageously be reconfigurable. The communicative protocol engine in such an implementation is programmed directly into one or both of the FPGAs on the tester slice. The FPGA (e.g., 316 or 318) can therefore be configured to communicate with the DUTs in any protocol that the DUTs support. This advantageously eliminates the need for swapping out tester each time a DUT with a different protocol needs to be tested. In one embodiment, the protocols can be high speed serial protocols, including but not limited to SATA, SAS or PCIe, etc. The new or modified protocols can be downloaded and installed directly on the FPGAs via a simple bit-stream download from the system controller through the tester processor without any kind of hardware interactions. Also, if a new protocol is released, the FPGAs can easily be configured with that protocol via a software download.
In one embodiment of the present invention, each FPGA comprises a number of protocol engine modules, wherein each of the protocol engine modules within a FPGA device can be configured with a different communicative protocol. Accordingly, an FPGA device can be connected to test multiple DUTs, each supporting a different communicative protocol simultaneously. Alternatively, an FPGA device can be connected to a single DUT supporting multiple protocols and test all the modules running on the device simultaneously. For example, if an FPGA is configured to run both PCIe and SATA protocols, it can be connected to test a DUT that supports both PCIe and SATA protocols. Alternatively, it can be connected to test two different DUTs, one DUT supporting the PCIe protocol and the other DUT supporting the SATA protocol.
Referring to
Instantiated FPGA block 410 can comprise a protocol engine module 430, a logic block module 450, and a hardware accelerator block 440. The hardware accelerator block 440 can further comprise a memory control module 444, comparator module 446, a packet builder module 445, and an algorithmic pattern generator (APG) module 443.
In one embodiment, logic block module 450 comprises decode logic to decode the commands from the tester processor, routing logic to route all the incoming commands and data from the tester processor 304 and the data generated by the FPGA devices to the appropriate modules, and arbitration logic to arbitrate between the various communication paths within instantiated FPGA tester block 410.
In one implementation, the communication protocol used to communicate between the tester processor and the DUTs can advantageously be reconfigurable. The communication protocol engine in such an implementation is programmed directly into the protocol engine module 430 of instantiated FPGA tester block 410. The instantiated FPGA tester block 410 can therefore be configured to communicate with the DUTs in any protocol that the DUTs support. The pre-verified interface or IP cores mentioned above, for example, can be programmed into the protocol engine module 430. This ensures compliance and compatibility according to a given interface standard. Further, the IP core allows the tester to achieve flexibility in that the IP core enables software-based changing of interfaces. Embodiments provide an ability to test multiple types of DUTs independent of the hardware bus adapter sockets. With such interface flexibility, new interfaces may be loaded into the IP core of a programmable chip thereby eliminating the need for the hardware bus adapter sockets (discussed in connection with
In one embodiment, for example, for storage/SSDs/HDDs, each FPGA comprises a configurable IC that connects to a SSD and that is programmable to provide storage based patterns through a storage specific interface such as SATA or SAS.
In one embodiment, an FPGA may be an SSD module-based tester that uses protocol-based communications to interface with a DUT or module. In one embodiment, the configurable interface core may be programmed to provide any standardized protocol-based communications interface. For example, in one embodiment, in the case of an SSD module-base test, the interface core may be programmed to provide standardized protocol-based communications interfaces such as SATA, SAS, etc.
Accordingly, from an electrical perspective, the FPGAs utilize an IP core instead of a hardware bus adapter socket. Enabled by software programming of the programmable chip resources of an FPGA, a given IP core may be easily reprogrammed and replaced with another IP core without changing the physical FPGA chip or other hardware components. For example, if a given FPGA-based tester currently supports SATA, all that would be required to be able to connect to a SAS DUT is for the FPGA to be reprogrammed to use a SAS IP core instead of the existing IP core configured for SATA.
This advantageously eliminates the need for hardware bus adapter cards and no protocol-specific hardware need be replaced to test DUTs with different protocol support. In one embodiment, the protocols can be high speed serial protocols, including but not limited to SATA, SAS or PCIe, etc. The new or modified protocols can be downloaded and installed directly on the FPGAs via a simple bit-stream download from the system controller through the tester processor without any kind of hardware interactions. Also, if a new protocol is released, the FPGAs can easily be configured with that protocol via a software download.
In
In one embodiment of the present invention, each of the protocol engine modules within a FPGA device can be configured with a different communicative protocol. Accordingly, an FPGA device can be connected to test multiple DUTs, each supporting a different communicative protocol simultaneously. Alternatively, an FPGA device can be connected to a single DUT supporting multiple protocols and test all the modules running on the device simultaneously. For example, if an FPGA is configured to run both PCIe and SATA protocols, it can be connected to test a DUT that supports both PCIe and SATA protocols. Alternatively, it can be connected to test two different DUTs, one DUT supporting the PCIe protocol and the other DUT supporting the SATA protocol.
The hardware accelerator block 440 of
The hardware accelerator block 440 can use comparator module 446 to compare the data being read from the DUTs to the data that was written to the DUTs in a prior cycle. The comparator module 446 comprises functionality to flag a mismatch to the tester processor 304 to identify devices that are not in compliance. More specifically, the comparator module 446 can comprise an error counter that keeps track of the mismatches and communicates them to the tester processor 304.
Hardware accelerator block 440 can connect to a local memory module 420. Memory module 420 performs a similar function to a memory module within any of the memory blocks 240A-240M. Memory module 420 can be controlled by both the hardware accelerator block 440 and the tester processor 304. The tester processor 304 can control the local memory module 420 and write the initial test pattern data to it.
The memory module 420 stores the test pattern data to be written to the DUTs and the hardware accelerator block 440 accesses it to compare the data stored to the data read from the DUTs after the write cycle. The local memory module 420 can also be used to log failures. The memory module would store a log file with a record of all the failures the DUTs experienced during testing. In one embodiment, the accelerator block 440 has a dedicated local memory module block 420 that is not accessible by any other instantiated FPGA tester blocks. In another embodiment, the local memory module block 420 is shared with a hardware accelerator block in another instantiated FPGA tester block.
Hardware accelerator block 440 can also comprise a memory control module 444. The memory control module 444 interacts with and controls read and write access to the memory module 420.
Finally, hardware accelerator block 440 comprises a packet builder module 445. The packet builder module is used by the hardware accelerator block in certain modes to construct packets to be written out to the DUTs comprising header/command data and test pattern data.
In certain embodiments, hardware accelerator block 440 can be programmed by the tester processor 304 to operate in one of several modes of hardware acceleration.
In bypass or standard mode, the hardware accelerator is bypassed and commands and test data are sent by the tester processor 304 directly to the DUT through path 472.
In hardware accelerator pattern generator mode, test pattern data is generated by the APG module 443 while the commands are generated by the tester processor 304. The test packets are transmitted to the DUT through path 474. This mode is also known as Protocol Independent Data Acceleration (PIDA) mode.
In hardware accelerator memory mode, the test pattern data is accessed from local memory module 420 while the commands are generated by the tester processor 304. The test pattern data is transmitted to the DUT through path 476. Routing logic 482 is needed to arbitrate between paths 472, 474 and 476 to control the flow of data to the DUT.
In hardware accelerator packet builder mode, the test pattern data may be generated by the APG module 443 and the packet builder module 445 is used to construct packets to be written out to the DUTs comprising header/command data and test pattern data. The test packets are transmitted to the DUT through path 474. This mode is also known as Full Acceleration (FA) mode.
The site module can comprise a general purpose connector 481. Because the protocol engine module 430 can be configured to run any number of various communicative protocols, a general purpose high speed connector 481 is required on the site module. Accordingly, if the protocol implemented on the protocol engine module 430 needs to be changed, no accompanying physical modification needs to be made on the site module. The site module connects to the DUT using load board 380 that can connect to the general purpose connector on the site module end, but is specific to the protocol being implemented on the DUT end. DUTs supporting different communicative protocols will require different configurations. Accordingly, the load board needs to be switched out and replaced if the protocol is reprogrammed to accommodate DUTs requiring a different configuration.
It should be noted that DUTs connected to any given tester slice (e.g., 340A-340N) or any given primitive are capable of running a different protocol than one or more other DUTs connected to the same respective tester slice or primitive. For example, the primitive 1010 may be connected to and used to test a plurality of DUTs 1020, each of which can run a different protocol, e.g., SATA, ATA, SCSI, etc. In one embodiment, primitive 1010 may be connected to and used to test primarily SSD drives.
In one embodiment, a Windows based host controller 1110 may be communicatively coupled to several different primitives, e.g., 1130A and 1130B. The Windows based host controller 1110 is capable of displaying graphical user interfaces to users for displaying information and for accepting user input. A communication backplane 1120 is coupled between the host controller 1110 and one or more primitives 1130A and 1130B. Each primitive comprises a number of tester slices (e.g., tester slices 340A-340N). In one embodiment, each primitive may comprise a number of COM boards 1140 (comprising a tester processor and associated memory) that are coupled to a number of FPGA tester slices 1150 via drivers 1160. (As mentioned previously, in one embodiment, the tester processor may be located on a COM board that is a separate board from the tester slices comprising the FPGAs). The tester slices, in one embodiment, are coupled to DUTs (e.g. SSDs) via a device interface board 1165. A different DIB can be used to support SSDs of different form factors and connection types. There can be several primitives within a rack, where each primitive is coupled to and controlled by the host controller. This allows a large volume of DUTs to be tested concurrently.
In one embodiment, the COM boards 1140 may contain an instance of the embedded OS platform in accordance with the present invention. The COM boards 1140 may control the tester slices 1150 within the primitives. The embedded OS platform is configured to communicate on one side with the host controller 1110 and on the other side the various FPGA tester slices 1150 that are populated within a primitive. In one embodiment, the embedded OS platform can be a heavily modified version of the Linux OS. There can be as many as 32 instances of embedded OS platforms within the system 1105 and controlled by the host controller 1110. The various OS platforms provide functionality to communicate with the FPGA tester boards 1150 also to communicate directly with the DUTS 1170.
Embodiments of the present invention support testing of drives that may have various protection modes enabled. For example, certain SSDs may have an extra 8 bytes of metadata per sector (e.g., 512+8, 4096+8). It should be noted that the values provided in the table in
Embodiments of the present invention also support testing of drives with non-standard sector sizes. For example, as shown in the table of
Embodiments of the present invention advantageously allow all combinations of protection modes and all combinations of variable sector sizes for the various DUTs to be flexibly supported in all the various hardware acceleration modes. For example, in PIDA mode, the commands are generated by the tester processor on the COM board but the test pattern data is generated by the FPGAs. In this instance, the embedded OS allows commands to be generated by the tester processor that support the various protection modes and odd sector sizes. Furthermore, the FPGAs can generate the data for the protection modes and various odd sector sizes much faster than conventional tester systems. Similarly, in full acceleration mode, the FPGA generates both the commands and data to support the various protection modes and odd sector sizes at much faster rates than conventional tester systems. For example, the calculations and computations to determine the various fields of the protection information (as shown in
In providing the variable sector size support, the embedded OS allows a “C” based tester scripting language to be used that can alter the sector size for the various SSDs being tested. Therefore, any sector size (from 512 to 4k, for instance) can be applied for testing across the various DUTs 1170. This sector size configuration can originate from the host computer and be applied to all primitives and subsequently to any or all DUTs 1170. With respect to protection mode support, any combination of protection mode support can be applied across the DUTs 1170 and can originate from host computer 1110. Protection mode, as mentioned above, influences the manner in which metadata is generated by the software or firmware. In particular, application of protection mode configuration alters the makeup of the metadata bytes that follow or are otherwise associated with a data packet. Any of the combinations of protection modes along with any combination of sector size can be applied across the primitives 1130A, 1130B from the host controller 1110. In other words, the firmware of the system allows hardware acceleration to be applied to all combinations of protection modes along with any combination of odd sector size. Applying hardware acceleration allows the SSDs to be tested faster because the FPGAs serialize the data and transmit it to the DUTs at much higher speeds than the tester processor itself, e.g., in PIDA and FA modes, the FPGA generates all the data and transmits it to the DUTs. Further, the firmware also allows the protection information for the protection mode to be calculated at high speeds using the FPGAs. Additionally, because the FPGAs are located in close proximity to the DUTs, the data is transmitted to and received from the DUTs faster.
Conventional tester systems, by comparison, do not have a mechanism to be able to test drives supporting protection modes and odd sector sizes. Furthermore, conventional tester systems were not able to test drives supporting protection modes and odd sector sizes at high speeds using firmware acceleration. Conventional tester systems only had the capability of testing drives that were formatted for standard sector sizes, e.g., 512 BPS.
In one embodiment, the tester framework 1105 also supports testing SSDs for the various sector sizes and protection modes they support at high speeds without needing to swap devices. For example, DUTs 1170 in Figure may support several different sector sizes and protection modes. The DUTs may, for instance, support 512+0, 512+8 (8 bytes of protection information metadata), and also an odd-sector size of 528. The tester framework 1105 cycles through the various sector sizes and protection modes and tests the DUTs for all possible configurations using the hardware acceleration modes. For example, the tester framework may first test the DUTs for 512 BPS. The tester framework may be configured to then re-format the drives and test them for 512+8 BPS, wherein the computations to generate the protection information and to generate the data for testing is performed by the FPGAs in one of the hardware acceleration modes, e.g. PIDA or FA. Subsequently, the tester framework may again re-format the drives and generate data at the FPGA level (in one of the acceleration modes) to test the SSDs for the non-standard sector size of 528. In this way, embodiments of the present invention can test all connected DUTs for all supported sector sizes and protection modes at high speeds without the need for disconnecting or replacing the devices.
As mentioned above, conventional tester systems also did not have any ability to test drives formatted for non-standard sector sizes. Embodiments of the present invention allow drives formatted for non-standard sector sizes (e.g., 520+0, 4104+0, etc.) both with and without hardware acceleration. In other words, the embedded OS in the tester processor 304, for example, is able to generate commands and test patterns to test drives formatted for odd sectors in standard (or bypass) mode. In other words, embodiments of the present invention can test drives formatted for odd sectors bypassing all of the FPGAs. Furthermore, embodiments of the present invention can also test drives formatted for odd sector sizes in any of the three hardware acceleration modes (e.g., hardware accelerator pattern generator or PIDA mode, hardware accelerator memory mode, and full acceleration mode).
Embodiments of the present invention in both standard mode and in the three hardware acceleration modes need to be able to perform the significant computations to generate repeatable test data for non-standard sector sizes (that are not a power of 2). The data generated for testing needs to be repeatable so that it can be written to the sector in a different order than it is read. In other words, if the data is not read in the same order as it is written, it is nevertheless important for it to be verifiable. Accordingly, the test patterns written into the odd sector sizes need to be repeatable.
Embodiments of the present invention are configured to perform the calculations to generate repeatable random or pseudo-random data for arbitrary sector sizes. Both the tester processor (for standard mode) and the FPGAs (for the hardware acceleration modes) will be programmed with an algorithm that generates random numbers using the sector number as a seed. In other words, the random data that is generated will be a function of the sector address. Using the sector number as a seed, the algorithm will generate a reproducible sequence of any length, e.g., the length of the odd sector size. When the data needs to be read back, it can be regenerated again (using the sector address as a seed) so that it can be compared with the data read back from the DUTs in order to ensure the integrity of the DUTs.
Flowchart 500 will be described with continued reference to exemplary embodiments described above in reference to
Referring now to
At block 502, the user also loads the test program into the system controller 301 through the graphical user interface. The test program defines all the parameters of the test that needs to be run on the DUTs. At block 504, the system controller transmits instructions to the tester processor on the site module 310A. This step includes the transmission of the bit files for the protocol engines to be programmed onto the FPGAs. The system controller can comprise routing logic to route instructions for a particular test program to the tester processor connected to the DUT controlled by the test program.
At block 506, after receiving instructions from the system controller, the tester processor 304 can determine the hardware acceleration mode for running the tests on the DUTs connected to site module 310A.
In one embodiment, the tester processor 304 can operate in one of four different hardware acceleration modes. Each functional mode is configured to allocate functionality for generating commands and test data between the tester processor 304 and the FPGAs 316 and 318. In one embodiment, the tester processor can be programmed to operate in bypass or standard mode, wherein all the commands and test data for testing the DUTs is generated by the tester processor 304 and the FPGAs 316 and 318 are bypassed.
In another embodiment, the tester processor 304 can be programmed to operate in hardware accelerator pattern generator mode (or PIDA mode), wherein pseudo-random or random data to be used in the testing of the DUTs is generated by the FPGAs 316 and 318 and the comparing is also done by the FPGAs, but the tester processor handles the command generation.
In yet another embodiment, the tester processor 304 can be programmed to operate in hardware accelerator memory mode, wherein the test pattern is pre-written onto the memory module connected to each FPGA 316 and 318 by the tester processor during initial set-up. The FPGAs in this mode access the dedicated memory device to retrieve the test data to be written to the DUTs, read the test data from the DUTs and compare the read data with the data written on the memory device. In this mode, each of the FPGAs control the memory device in response to read and write operations from the DUTs. The tester processor, however, is still responsible for the command generation in this mode.
In still another embodiment, the tester processor 304 can be programmed to operate in hardware accelerator packet builder mode (or Full Acceleration mode), wherein the data and basic read/writ/compare commands are generated by the FPGAs 316 and 318.
At block 508, the tester processor branches out to the mode under which the test will be run.
It should be noted that the FPGA 1035 of
Flowchart 600 will be described with continued reference to exemplary embodiments described above in reference to
Referring now to
At block 606, the tester processor communicates with instantiated FPGA block 410 and downstream port 480 to route the test packets containing the test pattern data to the DUTs. The bypass mode is a pass through mode, wherein, with some limited exceptions, the commands and data pass transparently through the instantiated FPGA block 410 directly to the DUTs. The DUTs are directly controlled by the tester processor 304 in bypass mode. While the instantiated FPGA block can comprise logic to route the packets through to the downstream port, it is not involved in either the command generation (also referred to as “signaling”) or the data generation.
At block 608, the tester processor 304 communicates with downstream port 480 to initiate a read operation from the DUTs of the data that was previously written to the DUTs at block 606. At block 610, the tester processor compares the data read from the DUTs to the data written at block 606. If there is any mismatch between the data written at block 606 and the data read at block 610, a flag is sent by the tester processor 304 to the system controller 301 at block 612. The system controller will then flag the mismatch to the user.
In bypass mode, tester processor 304 is constrained in the number of DUTs it can support because its processing capabilities can be maximized quickly from generating all the commands and test data for the DUTs. Also, the number of DUTs that can be supported by site module 310A is further limited by the bandwidth constraints on system buses 330 and 332. In bypass mode, the bandwidth of buses 330 and 332 is exhausted relatively quickly because of the large volume of data that is transmitted by the tester processor 304 over to the DUTs. Thus, other modes with more hardware acceleration are made available, wherein the FPGA devices have more functionality to generate test data and commands.
Flowchart 700 will be described with continued reference to exemplary embodiments described above in reference to
Referring now to
The mode is considered “hardware accelerated” because the functionality for generating data can be done much faster in hardware by the algorithmic pattern generator of the FPGA device than in software by the tester processor. Also the “tester per DUT” architecture allows the DUT to be directly connected to its own dedicated instantiated FPGA tester block generating test pattern data for the DUT as shown in
In one embodiment, the algorithmic pattern generator module 443 can be programmed to generate data on the fly. The APG module can generate incremental patterns, pseudo-random patterns or some type of constant pattern. The APG module can also have certain gating capabilities to generate test patterns with stripes, diagonal stripes or alternating patterns. In one embodiment, the APG module can use finite state machines, counters or linear feedback shift registers, among other things, to generate test patterns.
In some implementations, the APG module can be provided a starting seed as an initial value to generate more complex random patterns. As mentioned above, for example, using the sector number as a seed, the APG module will generate a reproducible sequence of any length, e.g., the length of the odd sector size. When the data needs to be read back, it can be regenerated again (using the sector address as a seed) so that it can be compared with the data read back from the DUTs in order to ensure the integrity of the DUTs.
At step 706, the instantiated FPGA block 410 communicates with the downstream port 480 to route the test pattern data to the DUTs in accordance with the commands and packet headers generated by the tester processor. The instantiated FPGA block 410, at step 708, communicates with the downstream port to read the test pattern data from the DUTs in accordance with commands generated by the tester processor. The comparator module 446 of the hardware accelerator block 440 is then used to compare the read data to the data written to the DUTs at block 710. The APG module 443 is designed in a way such that the comparator module can perform read operations on it with the same parameters that were used to generate the pseudo-random data and receive the same data that was written to the DUTs at block 704. The APG module 443 regenerates the data that was written to the DUTs on the fly and communicates it to the comparator module 446. Any mismatches are either logged on memory module 420 by the memory control module 444 or communicated by the instantiated FPGA block to the tester processor at block 712. The tester processor subsequently flags mismatches to the system controller at block 714 after receiving the error log.
Flowchart 800 will be described with continued reference to exemplary embodiments described above in reference to
Referring now to
At block 800 of the hardware accelerator pattern memory mode, the tester processor 304 generates commands and packet headers for the packets to be routed to the DUTs. The tester process retains the functionality for signaling in this mode. At block 802, the tester processor initializes the local memory module 420 of the instantiated FPGA tester block 410 with test patterns to be written out to the DUTs. One advantage of the hardware accelerator memory mode is that the test patterns generated by the tester processor may constitute real random data as opposed to pseudo-random data generated by the APG module 443 in the hardware accelerator pattern generator mode. Both the tester processor and the instantiated FPGA tester block have read and write access to the local memory module 420. However, the tester processor only accesses memory module 420 during initial set-up. During the accelerator mode, the tester processor does not access the memory module because the additional processing load on the tester processor 304 and the additional data load on the system buses 330 and 332 slows the acceleration down considerably.
At block 804, the instantiated FPGA tester block reads the test pattern data to be routed to the DUTs from the memory module 420. Because the memory module 420 is dedicated to the FPGA tester block or shared with just one other FPGA tester block, there is a high bandwidth connection between the two resulting in fast read operations. The logic block module 450 comprises functionality for routing the data generated and adding it to the packets to be written out to the DUTs.
After the data has been added to the packets, at block 806, the instantiated FPGA tester block communicates with the downstream port 480 to route the test pattern data to the DUTs in accordance with the commands and packet headers generated by the tester processor. The instantiated FPGA block 410, at step 808, communicates with the downstream port to read the test pattern data from the DUTs in accordance with commands generated by the tester processor. The comparator module 446 of the hardware accelerator block 440 is then used to compare the read data to the data written to the DUTs at block 810. Any mismatches are either logged on memory module 420 or communicated by the instantiated FPGA block to the tester processor at block 812. The tester processor subsequently flags mismatches to the system controller at block 814 after receiving the error log.
Flowchart 900 will be described with continued reference to exemplary embodiments described above in reference to
Referring now to
At block 902 of the hardware accelerator packet builder mode, the tester processor 304 generates commands to be communicated to the instantiated FPGA block 410 to generate its own packets. The tester processor retains functionality for only the non read/write/compare commands in this mode. The functionality for commands such as read, write and compare operations are conveyed to the instantiated FPGA blocks. At block 904, the packet builder module 445 of the instantiated FPGA tester block builds packets with header and command information to be communicated to the DUTs. The packets comprise at least the command type, the block address of the device and the test pattern data.
The algorithmic pattern generator module 443 within the hardware accelerator block 440 generates the pseudo random test data to be written to the DUTs at block 906. The logic block module 450 comprises functionality for routing the data and commands generated by the instantiated FPGA block and consolidating them into packets to be written out to the DUTs.
At block 908, the instantiated FPGA tester block communicates with the downstream port 480 to route the test pattern data to the DUTs. The instantiated FPGA block 410, at step 910, communicates with the downstream port to read the test pattern data from the DUTs. The comparator module 446 of the hardware accelerator block 440 is then used to compare the read data to the data written to the DUTs at block 912. Any mismatches are either logged on memory module 420 or communicated by the instantiated FPGA block to the tester processor at block 914. The tester processor subsequently flags mismatches to the system controller at block 916 after receiving the error log.
At block 1310, a host controller is coupled to a tester processor and an FPGA. The host controller may be a Windows based operation system as discussed above. Further, the tester processor may be running the Linux or a modified version of the Linux OS. The FPGA is communicatively coupled to the tester processor and is operable to generate commands and data for testing a plurality of DUTs in accordance with one of the various acceleration modes discussed above.
At block 1312, an acceleration mode is selected for generating commands and data for testing a plurality of connected DUTs. The acceleration mode may be a standard or bypass mode where the tester process generates all the commands and data and the FPGA is bypassed. Alternatively, the acceleration mode may be a PIDA, FA, or hardware accelerator memory mode as discussed above.
At block 1314, a number of sector sizes and protection modes to be tested for each of the connected plurality of DUTs is determined. For example, each SSD may have multiple odd sector sizes that it supports and needs to be tested for each of the sector sizes.
At block 1316, the commands and data for testing the plurality of DUTs for the supported sector sizes and protection modes is generated in accordance with the selected acceleration mode.
The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as may be suited to the particular use contemplated.
