Method and apparatus for accessing high-capacity test data via a high-speed interface
By enabling high throughput testing on existing IO channel circuits, the role of ATE is simplified and the high-speed IO standard is used to solve the high cost and low test coverage of SOC devices, achieving smaller, cheaper test equipment and high-speed IO standards Reliability, suitable for high performance computing and automotive ISO specification compliance.
Patent Information
- Application Number
- CN201910461487.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-10-30
- Filing Date
- 2019-05-30
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2039-05-30
AI Technical Summary
Existing test system-on-chip (SOC) devices require expensive test equipment (ATE), and as the chip size increases, the memory requirements of each IO channel increase, resulting in the additional test IO channel being unavailable and the test channel speed does not increase. The system Level structure testing has high latency and high cost, making it difficult to achieve high performance computing and automotive ISO specification compliance.
By enabling high throughput testing on existing IO channel circuits, simplifying the role of ATE to provide electrical energy and thermal control, leveraging high-speed IO standards such as PCIe to reduce the requirements of each IO channel and memory, and achieve smaller and cheaper ATEs. Suitable for system-level testing and field testing, cross-platform portability, supporting ISO 26262 standard fault-tolerant time intervals.
It realizes smaller and cheaper testing equipment, reduces testing complexity and capital costs, improves test coverage and reliability, is suitable for high-performance computing and automotive ISO specification compliance, and meets the requirements of on-site structural testing.
Smart Images

Figure CN111124768B_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the benefit of U.S. Provisional Patent Application Serial No. 62 / 580,265, filed on November 1, 2017, the entire contents of which are incorporated herein by reference. Background Art
[0003] Industry practices for testing systems on a chip (SOC) require expensive test equipment (ATE), whose cost is directly proportional to the number of channels, speed, and operating memory available for each test input-output (IO) channel. As chip size grows, the memory requirements for each IO channel also increase.
[0004] In many cases, additional test IO channels are not available, and test channel speeds will not increase. As 2.5D and 3D chips become more common, the IO channels available for test are further reduced. Scan compression schemes from various electronic design automation (EDA) tool vendors aim to achieve lower test data requirements, but they come at the cost of poor test coverage or vendor-specific design customizations.
[0005] Correlating ATE with system-level test failures is another challenge. Performing structural testing at the system level is expensive, which often makes the process impractical. Some efforts have been made to implement structural testing at the system level, but these tests use proprietary test interfaces and are not widely applicable. Existing solutions tend to incur high latency and therefore are expensive to use with multiple fault models or test procedures.
[0006] Automation and high performance computing (HPC) applications require testing to be performed in the field to help ensure safety and reliability. Structural testing provides high test coverage compared to functional mode and is best suited to meet the requirements of these applications. Existing solutions for in-field structural testing are limited by long execution times and / or data storage requirements. Summary of the invention
[0007] An embodiment of a system is disclosed that improves upon the challenges of earlier approaches. Utilizing IO channels in functional mode enables high throughput testing on existing IO channel circuits. Earlier approaches relied on dedicated test circuits in IO channels, the cost of which increases with the throughput requirements of the test.
[0008] The role of ATE is reduced to providing power and thermal control. Earlier approaches utilized ATE to provide test data and evaluate test results. The disclosed embodiments enable smaller and cheaper ATE.
[0009] The disclosed embodiments are applicable to platforms including ATE, system level test (SLT), board and field test. This portability enables diagnostic and debug correlation between platforms. Earlier approaches had specific design features for each platform, making correlation challenging.
[0010] The disclosed embodiments may be used to test automotive ISO specification compliance for fault tolerant time interval (FTTI) and improve reliability of HPC products.
[0011] BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS
[0012] To easily identify the discussion of any particular element or act, one or more of the most significant digits in a reference number refers to the figure number that first introduces the element.
[0013] Figure 1 A system architecture 100 is shown according to one embodiment.
[0014] Figure 2 An ATE architecture 200 is shown according to one embodiment.
[0015] Figure 3 A DUT logic element 300 is shown according to one embodiment.
[0016] Figure 4 A testing system 400 is shown according to one embodiment.
[0017] Figure 5 A hardware controller 304 is shown according to one embodiment.
[0018] Figure 6 is a block diagram of a computing system 600 within which the techniques described herein may be implemented or performed. Detailed Description
[0019] “FTTI” stands for Fault Tolerant Time Interval. The ISO 26262 standard defines FTTI as the time span during which a fault (1.42) or multiple faults in a system (1.129) may exist before a hazardous (1.57) event occurs.
[0020] "IEEE 1500" refers to a scalable standard architecture for enabling test reuse and integration of embedded cores and associated circuits. It abandons addressing analog circuits and focuses on facilitating effective testing of the digital aspects of systems on a chip (SoC). IEEE Std 1500 features serial and parallel test access mechanisms (TAMs) and a rich instruction set suitable for testing cores, SoC interconnects, and circuits. In addition, IEEE Std 1500 defines features that support core isolation and protection.
[0021] “ISO 26262” refers to the international standard for the functional safety of electrical and / or electronic systems in production vehicles, defined by the International Organization for Standardization (ISO).
[0022] "MBIST" refers to the memory built-in self-test logic, which generates a set of March algorithms that can be applied to detect memory failures.
[0023] "PCIe" refers to Peripheral Component Interconnect Express, a high-speed serial computer expansion bus standard designed to replace the older PCI, PCI-X, and AGP bus standards. PCIe has many improvements over older standards, including higher maximum system bus throughput, lower I / O pin count and smaller physical footprint, better performance scaling for bus devices, more detailed error detection and reporting mechanisms (Advanced Error Reporting, AER), and native hot-plug capabilities. The latest PCIe standard devices provide hardware support for I / O virtualization.
[0024] Embodiments of the system are disclosed, which enable wafer-level and system-level testing of systems-on-chip (SOCs) based on high-speed IO (e.g., PCIe, USB 2.0 and above, IEEE IEEE1394 / 1394b, Serial ATA, NVLink TM, etc.). By reducing the IO channels and memory required for each IO channel of the traditional method, the requirements for expensive test equipment can be improved. Therefore, the ATE architecture and design can be simplified to achieve a smaller form factor and reduce the capital cost of ownership and maintenance. By testing the SOC directly on the system-level platform, the complexity of test insertion can also be reduced.
[0025] The system is based on existing high-speed input and output (HSIO), such as the PCIe standard, making it portable across platforms, including ATE, system-level test (SLT), board and field test. High throughput enables field testing within the fault-tolerant time interval (FTTI) in compliance with the ISO 26262 standard.
[0026] Figure 1 A system architecture 100 is shown according to one embodiment. System architecture 100 includes endpoint 102, endpoint 104, endpoint 106, switch 108, root complex 110, DDR SRAM 112, processor 114, processor 116, and frontside bus 118. Root complex 110 mediates communications with peripheral devices ("endpoints") on frontside bus 118. A well-known example of root complex 110 is a southbridge.
[0027] System architecture 100 is scalable to different platforms. Devices under test (DUTs) are added, for example, as PCIe endpoint 106, or, for dual-site testing, as endpoint 102 and endpoint 104 coupled via switch 108 in this example. Test data and test results are stored by system memory (e.g., DDR SRAM 112) accessible by the CPU (e.g., processor 114 and processor 116 coupled to root complex 110 via front-side bus 118). The CPU manages the entire test process, loading the system memory with test data and triggering each DUT to perform a "self-test". Figure 3 A controller that can be implemented within the DUT is shown.
[0028] The system architecture 100 controls the front side bus 118 and performs DMA of the test data, executes the test, stores the response in the system memory, and triggers the CPU when the test is complete. The CPU compares the test response with the expected result to confirm the pass / fail status.
[0029] The system architecture 100 can test multiple devices serially or simultaneously. The switching device appears as a hot plug event, and the device is enumerated to become visible to the CPU on the bus, after which the above process can be performed.
[0030] Figure 2 ATE architecture 200 is shown according to one embodiment. ATE architecture 200 includes automatic test equipment 202, automatic test equipment 204, CPU 206, CPU 208, CPU 210, CPU 212, DUT 214, DUT 216, DUT 218, DUT 220, DRAM 222, DRAM 224, DRAM 226, and DRAM 228.
[0031] Figure 2 An exemplary ATE use case for the system architecture 100 is illustrated. The automatic test equipment 204 has multiple physical or virtualized CPUs with dedicated system memory (e.g., DRAM) that together drive the DUT. In this scenario, the role of the automatic test equipment 202 is reduced to providing power and temperature control. Testing and result comparison is performed by the controller of the CPU<->DUT (see Figure 3 ) is executed together with the logic executed on the CPU.
[0032] Figure 3 A DUT logic element 300 is illustrated according to one embodiment. The DUT logic element 300 includes a bus controller 302, a hardware controller 304, an IEEE 1500 306 component, a scan chain 308, and an MBIST 310 component.
[0033] The hardware controller 304 of the DUT 312 communicates with the bus controller 302 to acquire test data and control test execution. The hardware controller 304 also communicates with the JTAG / IEEE 1500 306 components to set the DUT 312 to various test configurations and trigger tests. For scan chain 308 testing, the hardware controller 304 provides high throughput direct access to the on-chip compressor / decompressor to load scan data and recycle data from the compressor.
[0034] Figure 4 A test system 400 according to one embodiment is illustrated. The test system 400 includes a CPU 402, a HDD 404, a root complex 110, a DDR SRAM 112, and a DUT 312. The DUT 312 includes a bus controller 302, a hardware controller 304, a unit 1 406, and a unit 2 408. The unit 1 406 includes a JTAG / IEEE 1500 component 410, a scan chain 412 component, and an MBIST component 414 element. The unit 2 408 generally includes similar components.
[0035] Depending on the idleness of unit 1 406 within the DUT 312, the DUT 312 communicates with the CPU to test unit 1 406, and other units (e.g., unit 2 408) remain functional. The functional state of unit 1 406 can be stored in an on-chip memory or in a system memory (e.g., DDR SRAM 112). The CPU loads the test data from a hard drive (e.g., HDD 404) into the system memory. The CPU prepares the DUT 312 for testing by executing logic for setup and triggers a hardware controller (e.g., hardware controller 304) to test a specific unit of the DUT 312. After the test is executed, the CPU restores the functional state of the unit under test and instructs the DUT 312 to resume operation.
[0036] Therefore, single CPU 402 can be used for testing the DUT of complex electronic system or circuit board.Each DUT 312 comprises the bus controller 302 that is joined (interface) to external bus port 314 and the internalized hardware controller 304 that is coupled to bus controller 302.Hardware controller 304 controls and coordinates the scan chain 308 and MBIST 310 components that are internalized in each functional unit in DUT.“Internalized” means that scan chain 308 and MBIST 310 do not go to the pin on the periphery that comprises functional unit, and hardware controller 304 is also inside that comprises DUT (hardware controller 304 interacts with the external device of DUT through external bus port 314).In some cases, DUT 312 can only comprise single functional unit, in which case, functional unit and DUT 312 can be the same. Typically, each DUT has multiple functional units, each of which includes local (internalized to the DUT) scan chains, local IEEE 1500 components, and (optionally) MBIST components if the functional unit has a memory array or facilitates testing an external memory array. The operation of each of these local devices is coordinated by a hardware controller 304 that is central to all the functional units in the DUT.
[0037] Hardware controller 304 comprises the logic of the operation of the internalization assembly of the coordination function unit independently according to the test logic received via bus controller.System architecture 100 is designed to realize that the memory array or the position of its part tested by specific MBIST 310 is flexible in design.For example, the memory array tested by specific MBIST 310 can be inside the functional unit comprising MBIST 310, or the memory array can be inside the DUT comprising MBIST 310, but shared by the functional unit of DUT, or root complex 110 can be coupled to one or more DUTs to the shared memory array.Therefore, depending on realization, the MBIST of specific functional unit or DUT can utilize memory bus to local memory array or non-local memory array.But specific MBIST is internalized to specific functional unit in the return of its test input and control and test result.These architecture features can greatly reduce a plurality of IO test channel pins on complex electronic equipment, comprise and enable to test large shared memory array across multiple functional units and DUT in a distributed manner.
[0038] In one embodiment, hardware controller 304 comprises system test sequencer 502, LBIST master controller 504, JTAG controller 506, LBIST scan chain controller 508 and PCIe controller 510.PCe controller 510 is operated to obtain test data from system memory and write test result into system memory.LBIST master controller 504 communicates with system test sequencer 502 and LBIST scan chain controller 508 driving scan chain 412.System test sequencer 502 communicates read / write data, address and data length to LBIST master controller 504, and causes DMA read / write request to data.Read completion data is received and decoded by system test sequencer 502, and is sent to LBIST master controller 504 or JTAG controller 506.LBIST scan chain controller 508 drives the decompression / compression structure of scan chain 412, so that scan data is transferred to scan chain and data is recovered from test result shift register. The JTAG controller 506 accesses the IEEE 1500 component 410 network to program the IEEE 1500 registers. This is just one example of how the hardware controller 304 may be implemented.
[0039] Figure 6 6 is a block diagram of one embodiment of a computing system 600 in which one or more aspects of the present disclosure may be implemented. The computing system 600 includes a system data bus 632, a CPU 602, an input device 608, a system memory 604, a graphics processing system 606, and a display device 610. In an alternative embodiment, the CPU 602, portions of the graphics processing system 606, the system data bus 632, or any combination thereof may be integrated into a single processing unit. Additionally, the functionality of the graphics processing system 606 may be included in a chipset or in some other type of dedicated processing unit or co-processor.
[0040] As shown, the system data bus 632 connects the CPU 602, the input device 608, the system memory 604, and the graphics processing system 606. In an alternative embodiment, the system memory 604 can be directly connected to the CPU 602. The CPU 602 receives user input from the input device 608, executes programming instructions stored in the system memory 604, and operates on data stored in the system memory 604 to perform computing tasks. The system memory 604 typically includes dynamic random access memory (DRAM) for storing programming instructions and data. The graphics processing system 606 receives instructions sent by the CPU 602 and processes the instructions, for example, to implement various aspects of the disclosed embodiments, and / or renders and displays graphics (e.g., images, tiles, videos) on the display device 610.
[0041] As also shown, system memory 604 includes application programs 612, API 614 (application programming interface), and graphics processing unit driver 616 (GPU driver). Application programs 612 generate calls to API 614 to produce a desired set of computational results. For example, application programs 612 may send a program or its functionality to API 614 for processing within graphics processing unit driver 616.
[0042] The graphics processing system 606 includes a GPU 618 (graphics processing unit), an on-chip GPU memory 622, an on-chip GPU data bus 636, a GPU local memory 620, and a GPU data bus 634. The GPU 618 is configured to communicate with the on-chip GPU memory 622 via the on-chip GPU data bus 636, and communicate with the GPU local memory 620 via the GPU data bus 634. The GPU 618 can receive instructions sent by the CPU 602, process the instructions, and store the results in the GPU local memory 620. Subsequently, the GPU 618 can display certain graphics stored in the GPU local memory 620 on the display device 610. The test architecture disclosed herein can be used in the graphics processing system 606 to test the GPU 618, the GPU local memory 620, and the on-chip GPU memory 622, where those components can be in the same or different functional test partitions. The CPU 602 can be used in the test mode, as well as coordinating the hardware controllers 304 of each functional partition, and storing the test results in the system memory 604 for analysis.
[0043] GPU 618 includes one or more logic blocks 624 . Logic blocks 624 may implement embodiments of the systems and techniques disclosed herein, such as testing logic for hardware controller 304 .
[0044] The GPU 618 may be provided with any number of on-chip GPU memories 622 and GPU local memories 620 , including zero, and may use the on-chip GPU memories 622 , GPU local memories 620 , and system memory 604 in any combination for memory operations.
[0045] On-chip GPU memory 622 is configured to include GPU programming 628 and on-chip buffer 630. GPU programming 628 may be transferred from graphics processing unit driver 616 to on-chip GPU memory 622 via system data bus 632. GPU programming 628 may include logic block 624.
[0046] The GPU local memory 620 typically includes less expensive off-chip dynamic random access memory (DRAM) and is also used to store data and programming used by the GPU 618. As shown, the GPU local memory 620 includes a frame buffer 626. The frame buffer 626 may, for example, store data for an image (e.g., a graphics surface), which may be used to drive the display device 610. The frame buffer 626 may include more than one surface, so that the GPU 618 may render one surface while driving the display device 610 using a second surface.
[0047] The display device 610 is one or more output devices capable of emitting a visual image corresponding to an input data signal. For example, a liquid crystal display or any other suitable display system may be used to construct the display device. The input data signal sent to the display device 610 is typically generated by scanning the contents of one or more image data frames stored in the frame buffer 626.
Claims
1. A device comprising: External bus port; A memory, wherein the memory is used to store test data; Hardware controller; as well as Multiple functional units, each functional unit includes: Internalized scan chains; Internalized IEEE 1500 components; Internalized MBIST components; and The hardware controller includes logic for independently coordinating operation of the internalized IEEE 1500 components on the scan chain and the internalized MBIST components on the memory array using direct memory access to obtain the test data for each functional unit through the external bus port.
2. The apparatus of claim 1, wherein the memory array tested by the MBIST component is inside a functional unit that includes the MBIST component.
3. device as claimed in claim 1, described memory array is tested by multiple MBIST assemblies in different functional units.
4. The apparatus of claim 1, the hardware controller comprising a test sequencer for providing the direct memory access over a PCI bus.
5. The apparatus of claim 1, wherein the internalized IEEE 1500 component is coupled to the internalized scan chain via JTAG.
6. The apparatus of claim 1, the memory array being external to the functional device.
7. The apparatus of claim 1, comprising a graphics processing unit.
8. The apparatus of claim 1, the external bus port comprising a Peripheral Component Interconnect Type E (PCIe) port.
9. The apparatus of claim 1, the external bus port comprising a Universal Serial Bus (USB) port.
10. The apparatus of claim 1, configured such that during testing of the apparatus using automatic test equipment, use of the external bus port is limited to temperature control and power.
11. A system comprising: Interface with external automatic test equipment; Central processing unit; System memory; as well as Multiple devices, each device includes: Hardware controller; logic for storing test data to the system memory; and Multiple functional units, each functional unit includes: Internalized scan chains; Internalized IEEE 1500 components; Internalized MBIST components; and The hardware controller includes logic for independently coordinating operation of the internalized IEEE 1500 components on the scan chain and the internalized MBIST components on the memory array; and The hardware controller is controlled by the central processing unit independently of the automatic test equipment such that test data is obtained from the system memory via direct memory access during testing and control provided to the automatic test equipment through the interface during testing is limited to temperature control.
12. The system of claim 11, wherein the memory array tested by the MBIST component is inside a functional unit that includes the MBIST component.
13. The system of claim 11, wherein the memory array is tested by multiple MBIST components in different functional units.
14. The system of claim 11, the hardware controller comprising a test sequencer, a LBIST master controller, a JTAG controller, and a LBIST scan chain controller for providing the direct memory access through a PCI bus.
15. The system of claim 11, the internalized IEEE 1500 component coupled to the internalized scan chain via JTAG.
16. The system of claim 11, the memory array being external to the functional device.
17. The system of claim 11, the device comprising a graphics processing unit.
18. The system of claim 11, the interface comprising a Peripheral Component Interconnect Type E (PCIe) port.
19. The system of claim 11, the interface comprising a universal serial bus (USB) port.
