WO2015051711A1 - 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备 - Google Patents

一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备 Download PDF

Info

Publication number
WO2015051711A1
WO2015051711A1 PCT/CN2014/087627 CN2014087627W WO2015051711A1 WO 2015051711 A1 WO2015051711 A1 WO 2015051711A1 CN 2014087627 W CN2014087627 W CN 2014087627W WO 2015051711 A1 WO2015051711 A1 WO 2015051711A1
Authority
WO
WIPO (PCT)
Prior art keywords
computer
cache
nand
slc
files
Prior art date
Application number
PCT/CN2014/087627
Other languages
English (en)
French (fr)
Inventor
张维加
Original Assignee
张维加
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 张维加 filed Critical 张维加
Priority to US15/028,028 priority Critical patent/US20160253093A1/en
Publication of WO2015051711A1 publication Critical patent/WO2015051711A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601Interfaces specially adapted for storage systems
    • G06F3/0602Interfaces specially adapted for storage systems specifically adapted to achieve a particular effect
    • G06F3/061Improving I/O performance
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • G06F12/0802Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
    • G06F12/0866Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches for peripheral storage systems, e.g. disk cache
    • G06F12/0868Data transfer between cache memory and other subsystems, e.g. storage devices or host systems
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F12/00Accessing, addressing or allocating within memory systems or architectures
    • G06F12/02Addressing or allocation; Relocation
    • G06F12/08Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
    • G06F12/0802Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
    • G06F12/0806Multiuser, multiprocessor or multiprocessing cache systems
    • G06F12/0811Multiuser, multiprocessor or multiprocessing cache systems with multilevel cache hierarchies
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F13/00Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F13/14Handling requests for interconnection or transfer
    • G06F13/20Handling requests for interconnection or transfer for access to input/output bus
    • G06F13/28Handling requests for interconnection or transfer for access to input/output bus using burst mode transfer, e.g. direct memory access DMA, cycle steal
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F13/00Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F13/38Information transfer, e.g. on bus
    • G06F13/42Bus transfer protocol, e.g. handshake; Synchronisation
    • G06F13/4282Bus transfer protocol, e.g. handshake; Synchronisation on a serial bus, e.g. I2C bus, SPI bus
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601Interfaces specially adapted for storage systems
    • G06F3/0628Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0629Configuration or reconfiguration of storage systems
    • G06F3/0631Configuration or reconfiguration of storage systems by allocating resources to storage systems
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601Interfaces specially adapted for storage systems
    • G06F3/0628Interfaces specially adapted for storage systems making use of a particular technique
    • G06F3/0655Vertical data movement, i.e. input-output transfer; data movement between one or more hosts and one or more storage devices
    • G06F3/0661Format or protocol conversion arrangements
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/06Digital input from, or digital output to, record carriers, e.g. RAID, emulated record carriers or networked record carriers
    • G06F3/0601Interfaces specially adapted for storage systems
    • G06F3/0668Interfaces specially adapted for storage systems adopting a particular infrastructure
    • G06F3/0671In-line storage system
    • G06F3/0683Plurality of storage devices
    • G06F3/0685Hybrid storage combining heterogeneous device types, e.g. hierarchical storage, hybrid arrays
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00Arrangements for program control, e.g. control units
    • G06F9/06Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44Arrangements for executing specific programs
    • G06F9/4401Bootstrapping
    • G06F9/4411Configuring for operating with peripheral devices; Loading of device drivers
    • G06F9/4413Plug-and-play [PnP]
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2212/00Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
    • G06F2212/21Employing a record carrier using a specific recording technology
    • G06F2212/214Solid state disk
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2212/00Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
    • G06F2212/21Employing a record carrier using a specific recording technology
    • G06F2212/214Solid state disk
    • G06F2212/2146Solid state disk being detachable, e.g.. USB memory
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2212/00Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
    • G06F2212/21Employing a record carrier using a specific recording technology
    • G06F2212/217Hybrid disk, e.g. using both magnetic and solid state storage devices
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2212/00Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
    • G06F2212/22Employing cache memory using specific memory technology
    • G06F2212/221Static RAM
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2212/00Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
    • G06F2212/28Using a specific disk cache architecture
    • G06F2212/283Plural cache memories

Definitions

  • This product belongs to the field of computer equipment and is a computer acceleration device based on the new USB protocol of multi-channel SLC NAND array and DRAM buffer.
  • changing the CPU changing the hard disk not only needs to accurately connect various data lines and sockets in the chassis, but also needs to export the data of the old hard disk and reinstall the system and each Class software, the average user does not. And the cost is still high, and compatibility with the motherboard is also a big problem.
  • the bottleneck of the entire data processing is on the hard disk. As long as this bottleneck can be opened, the information transmission will embark on the "highway.”
  • Solid state drives are used to replace mechanical hard drives.
  • Solid state disk Solid State Disk
  • a hard disk made of a solid state electronic memory chip array is composed of a control unit and a storage unit.
  • the interface specification and definition, function and usage of the SSD are exactly the same as those of the ordinary hard disk, and the shape and size of the product are also identical to those of the ordinary hard disk.
  • the flash-based solid-state hard disk is the main category of solid-state hard disk. Its internal structure is very simple.
  • the main body of the solid-state hard disk is actually a PCB board.
  • the most basic accessory on this PCB board is the control chip, the cache chip (some low-end hard disks are not available). Cache chip) and flash chip for storing data. Except for the main control chip and the cache chip, most of the rest of the PCB is in the NAND Flash memory chip.
  • SSDs do not have the motors and rotating media of ordinary hard disks, so they are quick to start and have excellent shock resistance.
  • Solid-state hard drives do not use magnetic heads, and disk reads and writes are fast and have low latency. Read and write speeds can generally reach 100M per second or more. To be precise, the expected speed is around 138M per second, which is calculated as follows: Flash without any acceleration measures
  • Flash write sequence can be divided into three phases: Command phase (Command), set the start and end of the write process by command word, and set the page address; Data phase (Data), load data into Flash
  • the programming phase Program
  • the data in the data buffer is actually programmed into the Flash array.
  • the typical write cycle is a minimum of 25ns, and the programming time is between 220 US and 500 US , so the write time is 275us and the maximum bandwidth is 61.4Mbps.
  • entry-level SSDs don't reach this speed.
  • the old computer upgrade mainly depends on the feasibility, and cost performance, these two aspects. Feasibility: The first is the compatibility issue. Early motherboards did not support solid state drives. Specifically, motherboards from the 1990s to 2010 were basically unable to support SSDs. In terms of feasibility: first for the vast majority For the old computer, installing SSD is ineffective, because the computer did not support SATA 2 protocol 11 years ago, let alone SATA3. The maximum support speed of the motherboard interface is 100M per second for ordinary IDE or SA TA hard disk protocol. Get accelerated with SSD. It is almost impossible to replace the motherboard. Secondly, it is still inconvenient. The average user is not good at replacing the hard disk.
  • Changing the hard disk means changing the whole system, copying all the files, reinstalling various drivers, and consuming at least one or two days. Furthermore, the S SD setting is complicated, and it can only exceed the normal hard disk speed under Win7 or Win8. XP does not recognize the Trim command of SSD, 4k alignment and ACHI. Not only can't you speed up, you can't use your computer, and in most cases it will blue screen and crash.
  • the entry level of the SSD 32G-64G is about 500 yuan, but 64G basically has no space after installing Win7 system and Office.
  • the price of the entry-level 128G is already close to $1,000.
  • the cost of upgrading an old computer is not worth it.
  • the second is the short life, solid state drives are generally MLC flash memory, and its life is very short without proper maintenance. And the maintenance of Trim, 4k alignment and other general maintenance measures will not be available to customers.
  • the Fast Disk is a PCI-E interface expansion card with one or two MLC NAND flash memory as a Mini PCI-E lx expansion card, through the PCI-E bus.
  • the system I/O controller performs data exchange.
  • the flash memory module used by the Fast Disk is NAND, not NOR. This is because NAND is better than NOR in accessing data performance and has better cost performance.
  • ReadyBoost and ReadyDrive functions can be provided. These functions will directly improve the performance of the system in terms of booting, hibernation, installing programs, copying files, loading games and other tasks related to disk operations. According to official data, the Fast Disk can speed up the downtime by 20%, while reducing the number of hard disk revolutions to save power.
  • ReadyBoost determines that the cache in the flash memory is more suitable for random read than the cache in the hard disk, it will randomly read data from the flash disk media.
  • the hard disk will read a large amount of data in batches at a time, and temporarily store it in the fast disk for the system to call; the data that needs to be written is also temporarily stored in the fast disk, and then accumulated after a certain amount of accumulation.
  • this on-demand read/write mechanism improves the system Performance is very helpful. During this period, the hard disk is idle, and the capacity of the fast disk is larger, and the longer the idle time of the hard disk is, thereby reducing the number of mechanical rotations and power consumption, and prolonging the battery life of the notebook battery.
  • ReadyDrive is in fact Microsoft's name for a hybrid hard drive (a hard drive with internal flash components).
  • the biggest temptation is that the data stored in it is "right to wait” _ because for flash memory, there is no need to start the head or wait for the head to rotate to the proper position.
  • the Hybird hard drive starts, sleeps, sleeps faster, and consumes less power. Because when the operating system reads and writes the cache, the drive itself can temporarily stop working without consuming any power. After resumed from hibernation, the laptop can immediately read data from the cache and start working without waiting for the drive's head to start up as usual.
  • the user can set the module to provide Read yBoost, Ready Drive, or both through the software interface.
  • the present invention provides a method for manufacturing a computer cache device, which is used for improving the running speed of an existing computer, and achieving a simple and reliable upgrade purpose.
  • the cache device manufactured by the method is increased. With the added durability of the cache and random read and write speeds, the write optimization is achieved, the multi-level hierarchical structure of the cache is realized, and a simpler USB interface is used.
  • an external hardware device specifically designed for accelerating a computer based on multi-channel parallel computing SLC NAND flash memory is proposed.
  • the present invention adopts the following scheme:
  • a plug-and-play USB interface (generalized USB interface, including ordinary USB, mini USB, microUSB)
  • Electronic device including the main control chip and SLC NAND flash memory module (or MLC NAND analog SLC working state, such as iSLC NAND flash memory is improved based on MLC NAND, is managed by specific flash memory Algorithm, reprogramming MLC NAND's 2-Bit Per Cell to 1 -Bit Per
  • the device which makes MLC NAND work closer to SLC NAND). It usually includes multiple SLC modules in parallel computing and multiple master ICs, or a multi-channel master IC to achieve a RAID-like effect.
  • the working principle of the device includes two aspects: First, the device is connected to the computer through the USB interface, and the cache file is created for the computer in the device memory, the common files of the cache system and the application program, and the scattered files frequently read and written by the pre-reading are utilized. High-speed random access to device memory and fast read and write speeds reduce computer system access to hard drives (including NAND-based SSDs), providing acceleration and improved I/O performance for computers.
  • the device and the DR AM cache are used as the agile level 1 cache, which can be implemented in two ways: 1.
  • the DRAM cache is provided in the device as a mapping table and a data buffer, for example, every 1 MB of DRAM cache. Map 1 GB of SLC NAND; 2.
  • In the cache call the computer's memory, and divide a part of the SLC NAND in the device to form a cache area of the high-speed cache.
  • the write credit is close to 8 times that of the read credit, so the write credit should be allocated to the DRAM layer, and the DR AM cache can be guaranteed.
  • the user's read operation is often more than the write operation, so it is reasonable to use DRAM as the first level cache and NAND as the second level. The above two methods can be used alone or in combination.
  • the device driver also improved the USB protocol, which hinders the traditional USB interface protocol.
  • the BOT protocol (Bulk-Only Transport) for fast data transmission is optimized, resource allocation optimization is performed on the USB transmission protocol, more system resources are allocated to the device, and multi-task transmission function similar to NCQ is supported, and the multi-QD is greatly improved. Random read and write rate.
  • the algorithm and architecture of the device also adopt the following design: 1. Providing intelligent compression and automatic background release to the system memory, thereby preventing the computer from increasing the reading and writing of the hard disk due to insufficient memory calling virtual memory; 2. Passing the user It is customary to carry out long-term monitoring and identification, and determine which data is to be used by the system.
  • SLC NAND flash module the CPU will directly obtain data from the device and then transfer it into the memory, thereby reducing the reading and writing of the hard disk;
  • the array module integrates two SLC NAND flash chips and uses dual-channel mastering to operate in dual-channel mode.
  • RAID 0 disk array 0
  • the performance bottleneck of solid-state memory is generally improved on the internal core interface by system-level or device-level parallel access.
  • the device virtualizes the application to pre-store all program files and program system environment files in the device.
  • the main program file When executed, it will be copied. Generate a virtual environment to execute, like the shadow system, all the operations involved are done in this virtual environment, and will not move the original system.
  • all the calling files are in the application directory, also It is under the SLC NAND flash module and will not be installed to the hard disk.
  • the equipment scheme is shown in the drawing of the specification.
  • the device can be used on computers with various Windows operating systems such as XP, Vista, Win7, Win8, and can be used by computers with a USB interface.
  • the DRAM cache is applied, with the current technical capabilities, it is generally possible to implement 1MB: 1GB of caching capability, and the DRAM buffer size is limited. Furthermore, using DRAM as a mapping table, the first mapping of the mapping table on the particle is loaded into the buffer before the self-test, and it is a more efficient speed-up method to speed up the writing back to the particle to update the speed. It is the reverse push repair algorithm after the power-down of the mapping table in the firmware. Otherwise, it will only cause a lot of hidden dangers, and the technical risk is large. Finally, the main disadvantage of the cache technology is that it needs to read the cache to build the index, which will introduce additional read transactions and increase the system's credit. The circuit is more complicated and consumes more power.
  • USB plug and play is the most convenient and simplest way to use, the most compatible, almost all computers have a USB interface. Any built-in interface is not convenient to use and will not be adopted by the public. So will the speed of the USB interface be affected? In fact, as long as the following classification discussion is made, it is obvious.
  • USB interface is generally USB2.0, and the speed bandwidth is 480M per second, which is equivalent to 60M per second for the highest data transmission. It doesn't look too high.
  • computers did not have SSDs.
  • the average data access speed of general mechanical hard disks was less than 20M per second, generally around 10M per second, which was much lower than the 60M bandwidth of USB2 mode.
  • you can optimize the USB protocol use the full speed as much as possible to speed up to nearly 6 times! (In the actual production sample described below, the speed under USB2 reaches 44-50M per second, as shown in Figure 2)
  • the speed of the shell ljUSB3 is actually greater than SATA.
  • USB 3.0 provides 5Gbps (625MB/s). Although the 6Gbps bandwidth of SATA III is larger, the conversion format is not the same, so it is only 600MB converted to MB/s. /S, in theory, will be better than USB
  • the 625MB/s of 3.0 is small, not to mention the 3Gbps (300MB/s) of SATA II. Let's discuss the two in terms of convenience. USB is an indispensable port for every computer. USB 3.0 is not only backward compatible, but also has the convenience of plug and play. It increases the power supply from 500mA to 900mA. It has considerable advantages.
  • USB has always had a very serious problem of low bandwidth utilization
  • the bandwidth of 2.0 is 480Mbps (60MB/S), but even if you actually use a USB flash drive with a transmission speed of more than 100MB/s, you can't use the full bandwidth.
  • the speed limit is only about half.
  • USB half-duplex transmission mode is like a walkie-talkie.
  • the other party can only listen to the voice. You must wait for the other party to finish before you can press the call button to send the message, which means that the half-duplex mode is provided.
  • the function of two-way data transmission, but the direction of data transmission is only one-way.
  • the BOT protocol is a single-threaded transmission architecture. It must wait for a complete data block to be sent before sending the next data. That is, no matter how wide the road is, it can only allow one. The car is driving on this road.
  • USB is upgraded to the 3.0 specification, although the extra five contacts are used, the full-duplex data transmission mode can be used to perform two-way data transmission, and the bandwidth is increased ten times compared with the previous generation. There are many, but the transmission architecture is still under the BOT, so we must optimize and accelerate.
  • the BOT's acceleration mode is well understood in the above metaphor: Since under the BOT architecture, there can only be one car on the road, one person is one, and the passenger car is filled with five people. It is also a large passenger transport with 50 passengers or one. After transporting a certain number of people, each time a large passenger transports on the road, it can reduce a lot of traffic.
  • the so-called USB Turbo mode is designed based on this principle, the data is organized into larger data blocks and then transmitted, and regardless of the storage medium, the processing capacity of the large file is always better than the small file. Using this method can significantly increase the data transfer speed.
  • Virtualization also virtualizes the system environment into a series of file forms that are launched at runtime. This will All required reads and writes of the program are transferred to the directory where the program is located, that is, the external SLC NAND flash chip, and the system disk is no longer required to be read or written. For this device, it means that the hard disk of the accelerated computer will no longer run the program file or the system file called by the program, all running in the external SLC NAND flash chip. This process completely avoids the hard disk read and write in the program, otherwise the hard disk will still be read and written during the running of the application.
  • the device Compared with the traditional computer upgrade, the device has the following advantages: 1. Simple operation: Upgrading the old computer often requires disassembling and replacing the memory for the hard disk. If you want to speed up, you need to change the motherboard to change the CPU, and work hard for one or two days. It is also often bad, or blue screen, the compatibility between various interfaces is not clear to the average user. The most appropriate way is to use the porter to carry the computer to the computer city to upgrade the site, but the price is very high, the cat is a lot more tired, and often the parts are stolen. With this device, you only need to install the driver and plug it into the computer. It can be done automatically with a few clicks, as shown in Figure 3.
  • USB3 can be transferred from PCI-E or ExpressCard. 0, compared to the original USB3.0, these transferred USB3.0 speed is lower, the data transmission is about 150M per second. So the old computer can also use USB3.0).
  • Figure 1 Schematic of the device.
  • Figure 4 Accelerated memory console interface in use of the sample device.
  • FIG. 6 Schematic diagram of the triple buffer of the sample device.
  • FIG. 7 All the programs that have been virtualized in the sample device manage the startup icon through a control center.
  • the high-speed layer is two sets of DRAM caches (the DRAM cache is configured in a ratio of 64:1 in the NAND:DRAM ratio in the device, and the DRAM cache of the host computer is called in a ratio of 8:1 in the NAND:DRAM ratio, and the host is called.
  • Ramdisk storage cache image file is generated on the system disk, ⁇ shutdown, load and save to avoid loss of power loss), create and allocate cache files for computers in SLC NAND and DRAM, cache system and application common files And pre-reading frequently read and write scattered files as a cache.
  • the write operation buffer is configured to the DRAM.
  • Cache including web browsing, is a write operation
  • the read operation cache especially the random read operation, is configured to the NAND cache, including program and game loading, etc., which are read operations.
  • the console has an interactive interface. In the console, the user can manually complete the specified program preload, memory compression, focus on the acceleration program, and pre-install a browser specially written based on the device cache mechanism to implement the network application. Focus on acceleration (modern users are increasingly using browsers).
  • the algorithm and architecture of the device also adopt the following design: 1.
  • the device has a virtual Windows system environment, The application can be virtualized, so that all program files and system environment files required by the program are pre-stored in the device, and the slave device runs instead of running from the host to improve the cache hit rate; 2. Pre-stored algorithms, through long-term user habits Monitoring, judging the data that the system is about to use, pre-existing in the device; 3. Providing intelligent compression and automatic background release to the system memory.
  • the device driver also improves the USB protocol, optimizes the BOT protocol in the traditional USB interface protocol, and optimizes resource allocation on the USB transport protocol.
  • the present invention has produced mass-produced samples for practical use. It is divided into a high-end version and a low-end version, and the high-end version is described in the above preferred embodiment.
  • the low-end version takes care of both cost and performance.
  • the dual-channel dual-channel SLC is pre-installed on the sample as the primary cache.
  • NAND memory module 16GB cache area, 16MB DRAM on board in a ratio of 1000:1, and high-speed communication with USB3.0 interface, as a random storage area of the local system to achieve acceleration and improve cache performance.
  • USB3.0 interface the test reads 260MB per second
  • the write is 240MB per second, which has doubled the speed of the SSD
  • the 4K random read and write under USB2.0 also reaches 40-50MB per second. I/O and random read and write performance have far exceeded mechanical hard drives. See Figure 2, Figure 6.
  • the device In addition to the application of parallel technology to form a dual-channel SLC NAND cache and on-board DRAM, the device also calls a portion of the computer's DRAM memory (the user can decide how much to call, but the device will calculate the recommended value).
  • the device As a mapping table and a high-speed L1 cache, 8G SLC NAND is used for random data and frequent read and write file caching, and the remaining 8G SLC NAND is used as a virtualization program storage installation area.
  • the device virtualizes the application, thereby pre-storing all program files and system environment files required by the program in the device.
  • the USB protocol is automatically optimized, the BOT Turbo mode is implemented, and more resources are allocated to the device.
  • the USB transmission protocol is changed, only the single-line cache exchange is supported. Simultaneously handle multi-cache task reading and writing, similar to hard disk NCQ technology, and fully play the role of the device as a new memory of the system.
  • the USB3 reads 190 and writes 200. After optimization, both pairs exceed 250M per second. This work is very important.
  • the algorithm and architecture of the device include: 1. providing intelligent compression and automatic background release to the system memory; 2. determining the data to be used by the system by using the long-term monitoring and identification of the user, pre-existing in the device; Dual channel mode, advanced version uses SandForce master (in the past, this master is only used for high-end solid state drives), low-level version uses Silver Can IS903 master, array module integrates two 8G magnesium SLC NAND chips and uses dual channel Master control
  • the device also has a graphical interface console that provides intelligent automatic management and control, and can selectively load channels and partitions. See Figure 4, because the proposed product name is temporarily removed. The newly added external cache can be viewed and managed through the control panel. Other details are as follows.
  • the NAND chip and circuit diagram are shown in Figure 5. Shenjin technology, 4-layer USB differential impedance PCB, guarantees good USB signal transmission; Others: Power IC adopts DC/DC conversion, high-quality patch crystal oscillator, nickel-plated USB plug through 24 hours of salt spray test; Operating temperature: 0°C To +60 ° C; Storage temperature: -20 ° C to +70 ° C.
  • SandForce master in the past this master is only used for high-end solid state drives
  • the low-level version can also use the Silver Can IS903 master.
  • High-end samples have been introduced in the best implementation case above.
  • the Silver Can IS903 dual-channel main control chip is equipped with two 8G SLC NAND memory modules, which are dual-channel dual-channel solutions. Under the USB3.0 interface, its test read is 260M per second, and the write is 240M per second. It has exceeded the speed of the SSD and can be accelerated for the latest factory computer! Under the USB2.0 interface, the device reads the cache speed of 44MB per second (bottom).
  • the total cache speed reaches 60M per second, mainly the random read and write of 4K and 512K exceeds 50MB per second.
  • the computer is generally a mechanical hard disk, the 512K random data speed is only 10-15M per second, 4K random reading and writing is often only about 1MB per second, so the 512K cache speed is increased by 3 times, 4K is increased by tens of times, and the system acceleration effect is very obvious. If the old computer of the USB2.0 mechanical hard disk is transferred to USB3 through PCMCIA or ExpressCar d, the speed can be increased ten times.
  • the NAND memory module has a total of 64GB, adopts multiple hierarchical cache design, the bottom layer is an eight-channel SLC NAND cache, and the high-speed layer is two sets of DRAM caches (the DRAM cache is configured in a ratio of NAND:DRAM ratio of 64:1 in the device, and the same is followed. NAND: The ratio of DRAM is 8:1.
  • the D RAM cache of the host computer is called.
  • the host part is called and the mode is Ramdisk to store the cache.
  • the image file is generated on the system disk, and the file is loaded and saved after shutdown to avoid loss of power loss.
  • the low-end version constitutes a dual-channel SLC in addition to the application of parallel technology.
  • the device also calls 128M memory in the DRAM memory of the computer as a mapping table and high-speed L1 cache, with 8G SLC NAND as random data and frequent read and write file cache.
  • the installation area is stored for the virtualization program with the remaining 8G SLC NAND.
  • DRAM cache operations we used a very fast caching algorithm optimized for write operations, achieving very high I/O speeds of up to several gigabytes per second.
  • the second-level cache operation of SLC NAND the current algorithm of the sample is based on the traditional disk cache rewriting, but unlike the traditional cache, we made two major changes to the device. First, the previous cache algorithm itself.
  • the virtualization scheme adopted by the sample has a virtual Windows environment, and the user can directly use the thousands of commonly used programs pre-loaded in the device, or virtualize the native application, thereby pre-storing all program files and program system environment files.
  • Figure 7. The principle of virtualization has been explained earlier, mainly by using the sandbox virtualization technology.
  • the application is installed into the running, all the actions are recorded and processed in the cost of the file.
  • the main program file When the main program file is executed, it will come. ⁇ Generate a virtual environment to execute, like the shadow system, all the operations involved are done in this virtual environment, and will not move the original system.
  • all the calling files are in the application directory. That is, under the SLC NAND flash module, it will not be installed to the hard disk.) This process completely avoids the hard disk read and write in the program, effectively increasing the cache hit.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Software Systems (AREA)
  • Computer Security & Cryptography (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Memory System Of A Hierarchy Structure (AREA)

Abstract

一种基于多通道SLC NAND与DRAM缓存的新USB协议计算机加速设备,包含主控芯片及SLC NAND模块,具备USB接口来连接计算机,在SLC NAND与DRAM中为计算机创建并分配缓存文件,缓存系统与应用程序的常用文件以及预读频繁读写的零散文件,作为高速缓存。同时,设备驱动还对USB协议进行改善,对于传统的USB接口协议中的BOT协议进行优化,并在USB传输协议上做资源分配优化。设备的算法与架构还采用如下设计:1. 设备将应用程序进行虚拟化处理,从而预存所有程序文件与程序所需系统环境文件在设备中;2. 多通道模式,阵列模组集成多块SLC NAND芯片并采用多通道主控;3. 通过对用户习惯长期监测,判断出系统即将使用的数据,预存在设备中;4. 对系统内存提供智能压缩与后台自动释放。

Description

一种基于多通道 SLC NAND与 DRAM缓存的新 USB协议 计算机加速设备
技术领域
[0001] 该产品属于计算机设备领域, 是一种基于多通道 SLC NAND阵列与 DRAM缓存 的新 USB协议的电脑加速设备。
背景技术
[0002] 计算机的更新换代非常快, 而产品型号众多, 设备种类繁多, 年代跨度大, 系 统平台复杂, 目前暂吋缺乏有效的通用型升级解决方案。
[0003] 1.为什么需要电脑加速型功能的产品
[0004] 技术的发展总把硬件甩在后面。 电影变高清, 系统出 Win8, 游戏的最低配置纷 纷变四核了, 微软新推出的 Office2013就要占掉 2G内存。 升级电脑成本很高, 一 般都要数千元。 目前, 这种升级是一个棘手的问题, 现有的解决方案一般是升 级电脑配新机器, 不但花费近万, 而且旧机器从此闲置成为占空间的电子垃圾 , 或者也有不少用户自行购买零配件来拆机换部件, 技术要求较高, 难度也很 大, 比如换 CPU, 换硬盘不但需要准确接驳机箱中的各种数据线与插口, 还需要 导出旧硬盘的数据并重装系统与各类软件, 一般用户根本不会。 而且成本依然 居高不下, 与主板的兼容性也存在很大问题。
[0005] 也有一些软件可以优化电脑系统, 比如 360优化大师, 加速球, 但是这些都没 有在实质上改善硬件能力, 只是清理电脑缓存垃圾等, 和很多人常说的电脑用 久了就重装一下系统可以变快很类似。 本身并没有增强计算机的性能。
[0006] 2.电脑加速的瓶颈在哪里?在于硬盘的速度, 尤其是小文件频繁读写和随机读 写
[0007] 近十年来, CPU和内存的性能提高了 100多倍, 但硬盘的性能只提高了两倍。
整个数据处理的瓶颈, 就在硬盘上。 只要能打通这个瓶颈, 信息传输就走上了" 高速公路"。
[0008] 正因为此, 固态硬盘才被用于取代机械硬盘。 固态硬盘 (Solid State Disk) 用 固态电子存储芯片阵列而制成的硬盘, 由控制单元和存储单元组成。 固态硬盘 的接口规范和定义、 功能及使用方法上与普通硬盘的完全相同, 在产品外形和 尺寸上也完全与普通硬盘一致。 基于闪存的固态硬盘是固态硬盘的主要类别, 其内部构造十分简单, 固态硬盘内主体其实就是一块 PCB板, 而这块 PCB板上最 基本的配件就是控制芯片, 缓存芯片 (部分低端硬盘无缓存芯片) 和用于存储 数据的闪存芯片。 除了主控芯片和缓存芯片以外, PCB板上其余的大部分位置都 是 NAND Flash闪存芯片了。
[0009] 固态硬盘没有普通硬盘的电机和旋转介质, 因此启动快、 抗震性极佳。 固态硬 盘不用磁头, 磁盘读取和写入速度快, 延迟很小。 读写速度一般可以达到 100M 每秒以上。 准确的说, 预期速度在 138M每秒左右, 其计算方式如下: 在不采取 任何加速措施的情况下, Flash
典型的读吋序由命令和数据部分组成, 读吋间为 78us, 因此典型的读带宽为 216 Mbps。 Flash的写吋序主要可以分为 3个阶段: 命令阶段 (Command) , 通过命令 字设置写入过程的幵始和结束, 并设置页面地址; 数据阶段 (Data) , 将数据加载 到 Flash
的数据缓冲中; 编程阶段 (Program) , 将数据缓冲中的数据真正编程到 Flash阵列 中。 典型的写周期最小为 25ns, 编程吋间在 220US-500US左右, 因此其写入吋间 为 275us, 最大带宽为 61.4Mbps。 对于一个典型的应用, 有 50%的读和 50%的写 , 则综合速率为 216x 50<¾+61.4x 50<¾=138.7Mbps。 当然, 入门级固态硬盘还达 不到此速度。
[0010] 虽然速度比起机械硬盘那还是快不少, 但是坏处也是很多的, 比如价格贵, 容 量小, 电池航程较短, 写入寿命有限等等。 关键是价格贵容量小, 确切的说是 不贵的就容量小, 五百元左右的入门级金士顿 SSD NOW只有 32G容量。 容量大 的就价格贵, 同样是 1TB的大小, 机械硬盘 200元左右, 固态硬盘至少五千。 因 此在新的出厂的电脑中, 固态硬盘也仍然没有取代机械硬盘。
[0011] 而且老电脑升级主要要考虑可行性, 以及性价比, 这两个方面。 可行性上: 首 先是兼容性问题。 早期的主板并不支持固态硬盘。 具体地说, 90年代到 2010年 期间的主板基本上都不能够支持固态硬盘。 就可行性而言: 首先对于绝大多数 老电脑来说, 安装 SSD是无效的, 因为 11年以前的电脑根本就不支持 SATA 2协 议, 更别提 SATA3了, 主板接口最大支持速度就是 100M每秒的普通 IDE或者 SA TA硬盘协议, 根本无法用 SSD获得加速效果。 而更换主板几乎不可能。 其次, 依然是不方便, 一般的用户并不擅长自己更换硬盘, 更换硬盘尤其意味着更换 整个系统, 拷贝所有的文件, 重装各种驱动, 消耗至少一两天的吋间。 再者, S SD设置复杂, 只有在 Win7或者 Win8下才可以超过普通硬盘速度, XP不识别 SSD 的 Trim指令、 4k对齐以及 ACHI。 不但不能加速, 还会无法使用电脑, 在大多数 情况下会蓝屏、 死机。
[0012] 就性价比而言: 第一就是价格高, 固态硬盘的入门级 32G-64G的价格都要五百 元左右, 可是 64G在安装 Win7系统与 Office之后就基本没有剩余空间了。 而入门 级 128G的价格就已经逼近千元。 升级旧电脑来说这个成本已经不值得。 第二就 是寿命短, 固态硬盘一般都是 MLC闪存, 其寿命在得不到正确保养下很短。 而 幵启 Trim, 4k对齐等保养措施一般客户不会。
[0013] 3.那么目前是否有其他的低成本更方便的技术方案来解决硬盘速度的瓶颈?
[0014] 当前也有用其他设备来加速电脑的尝试。 目前所知的就是英特尔的迅盘: 迅盘 是一块 PCI-E接口的扩展卡, 搭载有一块或两块 MLC NAND闪存, 作为一个 Mini PCI-E lx规格的扩展卡, 通过 PCI-E总线与系统 I/O控制器进行数据交换。 迅盘所 采用的闪存模块为 NAND, 而并非 NOR, 这是由于 NAND在存取数据的性能方面 要优于 NOR, 且具备更好的性价比。
[0015] 在系统的支持下, 可提供 ReadyBoost和 ReadyDrive功能, 这些功能将直接对系 统在启动、 休眠、 安装程序、 拷贝文件、 载入游戏等有关磁盘操作的任务上进 行性能提升。 官方资料表明, 迅盘可以使幵机速度加快 20%, 同吋减少硬盘转数 以节省功耗。
[0016] ReadyBoost功能简介:
[0017] 当 ReadyBoost确定闪存内的缓存比硬盘内的缓存更能满足随机读取需求吋, 它 便会从闪盘介质内随机读取数据。 硬盘会一次性的批量读出大量数据, 并暂吋 储存在迅盘中, 供系统随吋调用; 同吋需要写入的数据也先暂存在迅盘中, 等 积累到一定数量后再统一写入到硬盘中, 这种随用随取的读 /写机制对提高系统 性能很有帮助。 在这段吋间里, 硬盘处于闲置状态, 而且迅盘的容量越大, 硬 盘闲置的吋间越长, 从而减少机械转动次数和电量消耗, 延长笔记本电池的续 航吋间。
[0018] ReadyDriver功能简介:
[0019] ReadyDrive事实上就是微软对混合硬盘 (带有内部闪存部件的硬盘) 的称呼。
这种硬盘除了闪存显而易见的随机访问速度优势外, 最大的诱惑还是在于其中 保存的数据"立等可取" _因为对于闪存而言, 既不需要启动磁头, 也不用等待磁 头转动到合适的位置。 Hybird硬盘的启动、 休眠、 睡眠速度更快, 而且功耗更低 。 因为当操作系统读写缓存吋, 驱动器本身可以暂吋停止工作, 不消耗任何电 力。 而从休眠状态恢复运行吋, 笔记本电脑也能够马上从缓存中读取数据幵始 工作, 而不用像往常那样, 先得等待驱动器的磁头启动起来。
[0020] 在迅盘的驱动程序中可以看出, 使用者可以通过软件界面设定该模块提供 Read yBoost、 Ready Drive功育 ^, 还是两者兼具。
[0021] 但是, 迅盘依然不是一个有效的升级方案。 也正因为此, 现在已经不太有人提 起。 其失败的主要原因在于: 1.不能用于台式机, 也不能用于绝大多数笔记本
。 所有上网本以及多数笔记本电脑均不支持迅盘模块, 因为这不仅要求笔记本 电脑提供一个额外的 Mini PCI-E插槽, 同吋更重要的还要求笔记本电脑的 SATA 接口支持 ACHI功能; 2.安装复杂, 一般用户并不会拆机安装 mini PCI-E, 以至于 无法用于旧电脑升级; 3.效果不好。 PCI-E总线的速度本身被限制在 150M每秒以 下, 而英特尔的闪存则还远达不到这个速度, 实测在 35M每秒的随机读写速度, 对硬盘提升不大, 比固态硬盘还不如, 英特尔的迅盘体积受限, 无法加装缓存 或者并行模组, 或更多主控 IC; 4.价格昂贵。 4G的迅盘定价就在 100美元; 5.系 统兼容性差。 这一点本身就足以排除迅盘用于给旧电脑加速的可能了。 无论 Rea dydrive还是 Readyboost都只能用于 Windows Vista以上的操作系统, 而旧电脑绝大 多数都是 XP的操作系统, 也只能在 XP下流畅运行。
技术问题
[0022] 本发明提供了一种计算机缓存设备的制造方法, 用于为现有的计算机提高运行 速度, 实现简单可靠的升级目的, 与原有技术相比, 本方法制造的缓存设备增 加了缓存的耐用性与随机读写速度, 达到写优化, 实现了缓存的多级分层结构 , 并使用了更简便的 USB接口。
问题的解决方案
技术解决方案
[0023] 在本发明中, 一种基于多通道并行计算 SLC NAND闪存专门设计的用于加速电 脑的外接式硬件设备被提出。 为了有效提升旧电脑的性能, 并同吋满足简易的 安装使用能力与需求, 本发明采用了如下方案: 一种即插即用的 USB接口 (广 义的 USB接口, 包括普通 USB、 mini USB、 microUSB等) 的电子设备, 该电子 设备的包含有主控芯片以及 SLC NAND闪存模块 (或者以 MLC NAND模拟 SLC 工作状态, 如 iSLC NAND闪存是在 MLC NAND基础上改进的产品, 是通过特定 的闪存管理算法, 把 MLC NAND的 2-Bit Per Cell重新编程为 1 -Bit Per
Cell, 让 MLC NAND的工作方式变得更接近 SLC NAND) 。 一般包含并行计算的 多块 SLC模组以及多个主控 IC, 或一个多通道的主控 IC, 从而达到类似 RAID的 效果。 且该设备的工作原理包括两方面: 第一、 设备通过 USB接口与计算机连接 , 在设备存储器中为计算机创建缓存文件, 缓存系统与应用程序的常用文件以 及预读频繁读写的零散文件, 利用设备存储器的高速随机访问以及快速读写速 度, 减少计算机系统对硬盘 (包括基于 NAND的固态硬盘) 的访问, 为计算机提 供加速并提升 I/O性能。
[0024] 第二、 由于在 USB2.0的模式下, 贝 1」SLC的速度还会受到很大限制, 且 NAND的 读写操作不平衡, 写操作的幵销几乎为读幵销的八倍, 因此, 设备并且使用 DR AM缓存, 来作为敏捷型的一级缓存, 可以通过两种方式实现: 一、 在设备中自 带 DRAM缓存作为映射表和数据缓存区, 比如以每 1MB的 DRAM缓存映射 1 GB的 SLC NAND; 二、 在建立缓存吋调用计算机的内存, 划分一部分与设备中的 SLC NAND共同组成高速快取的 cache区。 因为闪存的读写幵销差异大, 写幵销接近 读幵销的 8倍, 因此应当将写幵销多分配到 DRAM层, 二就可以保障有足够的 DR AM缓存。 当然, 实际上用户的读操作往往多于写操作, 因此以 DRAM为一级缓 存而 NAND为二级是比较合理的。 以上的两种方式可以单用也可以并用。
[0025] 同吋, 设备的驱动还对 USB协议进行了改善, 对于传统的 USB接口协议中阻碍 快速数据传输的 BOT协议 (Bulk-Only Transport) 进行优化, 在 USB传输协议上 做了资源分配优化, 配置更多系统资源给设备, 并且支持类似 NCQ的多任务传 输功能, 大幅提升多 QD下的随机读写速率。
[0026] 设备的算法与架构还采用了如下设计: 1.对系统内存提供智能压缩与后台自动 释放, 从而避免计算机因内存不足调用虚拟内存而增大对硬盘的读写; 2.通过 对用户习惯进行长期监测识别, 判断出系统即将使用哪些数据, 预存在设备 SLC NAND闪存模块中, CPU将直接从设备中获取数据, 再将其转入内存中, 从而减 少对硬盘的读写; 3.双通道模式, 该阵列模组将 2块 SLC NAND闪存芯片集成在 一起, 并采用双通道主控, 从而能够以双通道模式工作。 作为一个逻辑磁盘组 , 将数据以分段的方式存储在这个逻辑磁盘组的不同物理磁盘上, 进行数据存 取吋, 阵列中的相关磁盘并行工作, 减低数据存取的吋间, 从而达到和 RAID 0 (磁盘阵列 0) 相同的加速效果, 使读写速度更快。 固态存储器的性能瓶颈一般 在内部的核心接口上, 可以采取系统级或设备级的并行访问来改进。
[0027] 另外很重要的一点是, 设备将应用程序进行虚拟化处理, 从而预存所有程序文 件与程序系统环境文件在设备中。 (虚拟化原理有多种, 主要是利用沙盒的虚 拟化技术, 先把应用程序安装到运行中所有的动作都记录起来并处理成本地的 文件, 当执行主程序文件吋, 它会临吋产生一个虚拟环境来执行, 类似影子系 统一样, 一切涉及的操作都是在这个虚拟环境中完成, 并不会去动原本的系统 。 这样处理后所有的调用文件都在应用程序的存放目录, 也就是 SLC NAND闪 存模块下, 而不会安装到硬盘。 ) 这一处理最彻底地避免了程序使用中的硬盘 读写。
[0028] 这样做是为了实现更快的程序运行、 更为简单的程序安装与操作、 更强大的系 统运行能力, 并具有广泛兼容性的系统程序运行能力。 使得该应用程序可以无 需安装, 直接以即插即用的形式在计算机运行, 并且是高速运行。 而后设备再 将该应用程序以数据形式导入到服务端。 这种做法还减少了系统服务项, 尤其 是大大减少了计划任务项, 扩展加载项与幵机启动项, 从而在增强系统应用功 能的同吋达到系统优化。
[0029] 设备方案见说明书附图 1. 该设备可以在 XP, Vista, Win7, Win8等各种 Windows操作系统的计算机上使用 , 并且只要有 USB接口的计算机就能使用。
[0030] 关于几个关键问题的说明如下:
[0031] 1. 为什么还要使用 SLCNAND的缓存与并行技术而非仅仅 DRAM缓存?
[0032] 首先, 如果仅仅应用 DRAM缓存的话, 以目前的技术能力, 一般可以实现 1MB : 1GB的缓存能力, DRAM缓存大小有限。 再者, 以 DRAM作为映射表, 在自检 前把颗粒上的映射表第一吋间加载到缓存里, 需要更新的吋候再回写回颗粒上 来提速是个比较有效率的提速方式, 但前提是固件内的映射表掉电后的反推修 复算法要做的好, 不然只会是造成大量丢盘的隐患, 技术风险较大。 最后, 缓 存技术的主要缺点是需要对缓存进行信息读取来构建索引, 从而会引入额外的 读事务而增大系统的幵销, 电路较为复杂, 功耗更大。 如果利用 F P GA (F i e l d P r o g r a mm i n gG a t e A r r a y) 或者 F 1 a s h自带 的部分缓存来加速读写, 则对整个系统来说缓存资源太少, 无法进行整体调度 , 会导致缓冲的频繁失效而增大系统的响应吋间。
[0033] 2. 为什么要采用外接 USB方式而不是内置的 SATA方式?
[0034] 很显然, USB即插即用是最方便也最简单的使用方式, 兼容性最高, 几乎所有 电脑都有 USB接口。 任何内置接口都不方便使用, 也不会被社会大众采纳。 那么 USB接口的速度会收到影响吗?实际上只要进行如下的分类讨论就很明显了。
[0035] 对于 2009年以前的电脑, USB接口一般为 USB2.0, 速度带宽 480M每秒, 相当 于最高数据传输为 60M每秒。 看起来不高。 但是, 09年以前的电脑没有固态硬盘 , 一般的机械硬盘的随机数据存取速度都小于 20M每秒, 一般在 10M每秒左右, 远低于 USB2模式的 60M带宽。 只要能够对 USB协议进行优化, 尽可能利用满速 度, 可以加速接近 6倍! (在下面介绍的实际生产样品中, 在 USB2下的速度就达 到了 44-50M每秒, 见附图 2所示)
而对于有 USB3的电脑, 贝 ljUSB3的速度实际上大于 SATA
3。 先以带宽来看, USB 3.0提供了 5Gbps (625MB/s) 的大小, 虽然 SATA III的 6Gbps账面带宽较大, 不过因为传输架构的换算方式并不相同, 因此换算成 MB/s吋仅为 600MB/S, 在理论值上会比 USB 3.0的 625MB/S来得小, 更不用说是 SATA II的 3Gbps (300MB/s) 了。 再以方便 性来讨论这两者, USB是每台计算机不可或缺的端口, USB 3.0不但可以向下兼 容、 拥有即插即用的便利性, 在供电方面更从 500mA增加到了 900mA, 而造就 了其相当大的优势。
[0037] 3. 为什么要修改 USB协议?
[0038] USB以往就一直存在相当严重的带宽利用率低下的问题, USB
2.0的带宽为 480Mbps (60MB/S) , 但就算实际使用传输速度高达 lOOMB/s以上 的优盘也无法用到完整的带宽, 速度上限仅有约一半的
33MB/S左右, 这是因为 USB的半双工传输模式与 BOT (Bulk Only Transfer) 传 输协议所导致的关系。 半双工的数据传输方式就像是对讲机, 当一方按下发话 钮吋另一方只能收听声音, 必须等待对方说完才可以按下发话钮进行发话, 也 就是说半双工的模式虽然提供了双向数据传输的功能, 但是数据的传输方向仅 有单向而已。 而 BOT协议则是一种单线程的传输架构, 必须等待一笔数据区块 完整送达之后才能够发送出下一笔数据, 也就是说不论这条马路有多宽, 但是 就只能够允许一台车子在这条马路上行驶, 这样的方式根本没有办法有效纾解 后方庞大的车流量, 而会造成资料区块 「塞车」 的情形发生。 而当 USB提升到 3. 0的规格, 虽然使用了额外的五个接点, 改为全双工的数据传输模式, 可以同吋 进行双向的数据传输, 带宽比起上一代也提升了有十倍之多, 不过其传输架构 还是在 BOT之下, 因此我们必须予以优化加速。
[0039] BOT的加速模式在上述的比喻下就很好理解了: 既然在 BOT的架构之下, 马路 上就只能够有一台车在行驶, 一个人幵是一台, 小客车坐满 5个人也是一台, 大 型客运坐满 50人也还是一台, 在要运送一定数目的人数吋, 每次都是大型客运 在路上行驶, 就可以减少许多车流量了。 所谓的 USB Turbo模式就是基于这样的 原理来设计的, 将数据整理成较大数据区块再进行传送, 而且不论何种储存媒 体, 对于大档案的处理能力总是比小档案要来得好, 因此使用这种方式可以显 著提升数据传输速度。
[0040] 4. 为什么要虚拟化系统程式?
[0041] 虚拟化将系统环境也虚拟化成一系列的文件形式, 在运行的吋候启动。 这就将 程序运行的所有所需读写都转移到程序所在目录, 也就是外接的 SLC NAND闪 存芯片中, 不会再需要系统盘的读写。 对于本设备而言, 意味着被加速电脑的 硬盘将不再运行程序文件或程序调用的系统文件, 所有这些都在外接的 SLC NAND闪存芯片中运行。 这一处理最彻底地避免了程序使用中的硬盘读写, 否则 在应用程序运行过程中不可避免对硬盘依然会有读写。
[0042] 同吋, 这样做是为了实现更快的程序运行、 更为简单的程序安装与操作、 更强 大的系统运行能力, 并具有广泛兼容性的系统程序运行能力。 使得该应用程序 可以无需安装, 直接以即插即用的形式在计算机运行, 并且是高速运行。 而后 设备再将该应用程序以数据形式导入到服务端。 这种做法还减少了系统服务项 , 尤其是大大减少了计划任务项, 扩展加载项与幵机启动项, 从而在增强系统 应用功能的同吋达到系统优化。
发明的有益效果
有益效果
[0043] 相比于传统的电脑升级, 本设备具备如下优点: 1.操作简单: 升级老电脑往往 需要拆机换内存换硬盘,如果要加快速度还要动手焊主板换 CPU, 忙活一两天还 常常倒腾坏, 或出现蓝屏, 各种接口之间的兼容性问题也绝非一般用户搞得清 楚的。 最妥当的办法是自己当搬运工将电脑抱到电脑城去现场升级, 但价格很 高, 猫腻很多, 常常被偷换部件。 用本设备只需要安装驱动插上电脑点击几下 就可以自动完成, 如附图 3。
[0044] 2.效果较好: 对于 USB2.0的普通机械硬盘的电脑, 程序启动运行速度可以提升 3-6倍, 对于 USB3.0的较新机械硬盘或混合硬盘的电脑, 程序启动运行速度可以 提升 10-20倍, 对于 USB3.0的固态硬盘的电脑, 程序启动运行速度可以提升 2-3倍 . (此外, 其实对于一般的电脑, 都可以从 PCI-E或者 ExpressCard转接出 USB3.0 , 相比于原装的 USB3.0, 这些转接出的 USB3.0速度较低, 数据传输大约在 150M 每秒。 因此老电脑也可以用上 USB3.0的) 。
[0045] 3.成本低廉, 便于普及。
对附图的简要说明
附图说明 [0046] 图 1.设备的原理图。
[0047] 图 2.样品设备的使用效果图, USB2下设备读取缓存速度 44MB每秒 (底部) , 经 DRAM写优化后缓存总速度达到了 60M每秒。
[0048] 图 3.样品设备的使用操作图, USB即插即用。
[0049] 图 4.样品设备的使用中加速内存控制台界面。
[0050] 图 5.样品设备的芯片与电路板图。
[0051] 图 6.样品设备的三重缓存示意图。
[0052] 图 7.样品设备中所有已经虚拟化的程序通过一个控制中心管理启动图示。
实施该发明的最佳实施例
本发明的最佳实施方式
[0053] 根据目前已知的市场设备与技术手段, 在合理的成本范围下, 目前能够应用的 最佳实施方案之一为:
[0054] 采用 USB 3.0或 3.1接口, SandForce主控, 搭载 1GB DRAM内存颗粒与 8片 8GB 的 SLC NAND颗粒, 八通道 SLC NAND存储模块共 64GB, 采用多重分层级缓存 设计, 底层为八通道 SLC
NAND缓存, 高速层为两组 DRAM缓存 (设备内按照 NAND:DRAM比例为 64: 1 的比例配置 DRAM缓存, 同吋按照 NAND:DRAM比例为 8: 1的比例调用宿主计 算机的 DRAM缓存, 调用宿主部分并拟态为 Ramdisk存放缓存,在系统盘生成镜 像文件, 幵关机吋载入与保存以免掉电丢失) , 在 SLC NAND与 DRAM中为计 算机创建并分配缓存文件, 缓存系统与应用程序的常用文件以及预读频繁读写 的零散文件, 作为高速缓存。 鉴于闪存的写操作幵销约为读操作幵销的 8倍这一 特点, 以及普通计算机用户读操作高于写操作这一现象, 将写入操作缓存尤其 是小文件写入操作缓存配置到 DRAM缓存, 包括网页浏览等属于写操作, 而读 取操作缓存尤其是随机读取操作配置到 NAND缓存, 包括程序与游戏载入等属于 读操作。 并带有交互界面的控制台, 在控制台中用户可手工完成指定的程序预 载, 内存压缩, 重点加速程序, 并预装有一个专门基于本设备缓存机制编写的 浏览器, 以实现网络应用的重点加速 (现代用户越来越多使用浏览器) 。
[0055] 设备的算法与架构还采用如下设计: 1.设备上带有虚拟的 Windows系统环境, 能够将应用程序进行虚拟化处理, 从而预存所有程序文件与程序所需系统环境 文件在设备中, 从设备运行而不是从主机运行, 以提高缓存命中率; 2.预存算 法, 通过对用户习惯长期监测, 判断出系统即将使用的数据, 预存在设备中; 3. 对系统内存提供智能压缩与后台自动释放。
[0056] 同吋, 设备驱动还对 USB协议进行改善, 对于传统的 USB接口协议中的 BOT协 议进行优化, 并在 USB传输协议上做资源分配优化。
[0057] 上所述仅为在目前已知的市场设备与技术手段下, 在合理的成本预算范围下, 本发明的最佳实施方式之一, 并不限定本专利的保护范围, 本领域技术人员在 本发明保护范围内的结构修改均应在本发明的保护范围之内。
本发明的实施方式
[0058] 本发明已经生产出批量量产的样品供实用。 分为高端版与低端版, 高端版如上 面最佳实施方式所述。 低端版则兼顾成本与性能, 样品上预装了作为主缓存的 双贴双通道的 SLC
NAND存储模块 16GB高速缓存区, 按照 1000: 1的比例板载 16MB的 DRAM, 并 用 USB3.0接口进行高速通信, 作为本地系统的随机存储区达到加速、 提升缓存 性能的目的。 在 USB3.0接口下, 其测试读取为 260MB每秒, 写入为 240MB每秒 , 已经双倍于固态硬盘的速度, USB2.0下的 4K随机读写也达到 40-50MB每秒, 其 I/O与随机读写性能都已经远远超过机械硬盘。 见附图 2、 附图 6所示。
[0059] 除了应用并行技术构成了双通道的 SLC NAND缓存以及板载的 DRAM之外, 设 备还调用计算机的 DRAM内存中一部分出来 (用户可以决定调用多少, 但是设 备会计算给出建议值) , 作为映射表与高速一级缓存区, 以 8G的 SLC NAND为 随机数据与频繁读写文件缓存, 以剩下的 8G的 SLC NAND为虚拟化程序存储安 装区。
[0060] 在 SLC NAND分区上, 带有便携式的 Windows虚拟环境, 设备将应用程序进行 虚拟化处理, 从而预存所有程序文件与程序所需系统环境文件在设备中。
[0061] 设备在与计算机连接后会自动进行 USB协议优化, 实现 BOT Turbo模式, 并分 配更多资源给设备, 改变 USB传输协议之后原本只支持单线的缓存交换, 变成可 同吋处理多重缓存任务读写, 类似硬盘 NCQ技术, 更充分发挥了设备作为系统 新内存的作用。 优化前 USB3下读 190, 写 200, 优化后双双超过 250M每秒, 可见 这一工作十分重要。
[0062] 设备的算法与架构包括了: 1.对系统内存提供智能压缩与后台自动释放; 2.通 过对用户习惯进行长期监测识别, 判断出系统即将使用哪些数据, 预存在设备 中; 3.双通道模式, 高级版本使用 SandForce主控 (在过去这种主控只用于高端 固态硬盘) , 低级版本使用银灿 IS903主控, 阵列模组集成 2块 8G的镁光 SLC NAND芯片并采用双通道主控;
[0063] 用户只需要插入电脑并安装驱动便可以幵启上述功能了, 见附图 3所示。
[0064] 设备还具有图形界面的控制台, 提供智能化的自动管理与控制, 可以选择性的 载入通道与分区。 见附图 4所示, 因需要暂吋隐去了拟定的产品名称。 新增加的 外接缓存可以通过控制面板査看与管理。 其他详细介绍如下。
[0065] 1.样品使用的两种缓存材料
[0066] 2块 8G的镁光 DDR SLC NAND芯片, SLC DDR同步闪存, 单通道 [SLC-8K:]。
样品设备的 SLC
NAND芯片与电路图见附图 5所示。 沉金工艺, 4层 USB差分阻抗 PCB, 保证 USB 信号传输良好; 其他: 电源 IC采用 DC/DC转换, 优质贴片晶振, 通过 24小吋盐 雾测试镀镍 USB插头; 工作温度: 0°C至 +60°C; 存储温度: -20°C至 +70°C。
[0067] 16MB DRAM高品质内存芯片颗粒, SOJ封装, 适应工业温度 (-40°C〜+85°C
[0068] 2.样品多通道硬件架构 (附图 5)
[0069] SandForce主控 (在过去这种主控只用于高端固态硬盘) , 低级版本也可使用 银灿 IS903主控。 高端样品已经在上面的最佳实施案例中介绍。 在低端样品上, 则采用银灿 IS903双通道主控芯片, 搭载两块 8G的 SLC NAND存储模块, 属于双 贴双通道方案。 在 USB3.0接口下, 其测试读取为 260M每秒, 写入为 240M每秒 , 已经超过固态硬盘的速度, 可以为最新出厂的计算机加速! 在 USB2.0接口下 , 其设备读取缓存速度 44MB每秒 (底部) , 经 DRAM写优化后缓存总速度达到 了 60M每秒, 主要是 4K与 512K的随机读写都超过了 50MB每秒, 而由于 USB2.0 的电脑一般为机械硬盘, 512K随机数据速度仅为 10-15M每秒, 4K随机读写往往 只有 1MB每秒左右, 所以 512K缓存速度提升了 3倍, 4K提升数十倍, 对于系统 加速效果十分明显。 如果 USB2.0的机械硬盘的老电脑通过 PCMCIA或 ExpressCar d转接到 USB3后, 则可以提速十倍。
[0070] 3.样品的三重缓存机制 (附图 6)
[0071] 高端版搭载 1GB DRAM内存颗粒与 8片 8GB的 SLC NAND颗粒, 八通道 SLC
NAND存储模块共 64GB, 采用多重分层级缓存设计, 底层为八通道 SLC NAND 缓存, 高速层为两组 DRAM缓存 (设备内按照 NAND:DRAM比例为 64: 1的比例 配置 DRAM缓存, 同吋按照 NAND:DRAM比例为 8: 1的比例调用宿主计算机的 D RAM缓存, 调用宿主部分并拟态为 Ramdisk存放缓存,在系统盘生成镜像文件, 幵关机吋载入与保存以免掉电丢失) 。 低端版除了应用并行技术构成了双通道 的 SLC
NAND缓存外, 以及 16MB的 DRAM—级缓存外, 设备还调用计算机的 DRAM内 存中 128M内存出来, 作为映射表与高速一级缓存区, 以 8G的 SLC NAND为随机 数据与频繁读写文件缓存, 以剩下的 8G的 SLC NAND为虚拟化程序存储安装区 。 在 DRAM缓存操作上, 我们使用了一种非常快的针对写操作优化的缓存算法 , 获得了很高的 I/O速度, 可以达到数 GB每秒。 在 SLC NAND构成的二级缓存操 作上, 目前该样品的算法则是基于传统的磁盘缓存改写的, 但是不同于传统缓 存的是我们做了针对设备的两大改变, 首先, 以往的缓存算法本身在实现的吋 候没有考虑到并行性, 所有请求全部是串行化处理。 然而我们的设备是多通道 的并行设备, 将串行 I/O改为并行 I/O则可以有效地提高其 I/O性能。 采用现代 多线程编程方法将串行 I/O改为并行 I/O, 采用细粒度的同步锁机制增加 I/O过 程的并行性, 从而改善 I/O性能。 另一方面, 以往的缓存算法在缓存磁盘数据的 吋候并未区分 I/O的类型, 它缓存所有请求, 不管是随机 I/O还是顺序 I/O。 事 实上我们的 SLC NAND部分缓存最擅长的随机读操作 I/O。 通过在 I/O过程中判 别其特征, 仅对较为随机的 I/O请求尤其是读操作请求采用 SLC NAND进行缓存 。 在 USB 3的设备上, 仅仅多通 SLC NAND部分就达到了数百 MB每秒的速度。
[0072] 4.样品采用的虚拟化方案 [0073] 样品具备虚拟的 Windows环境, 用户可以直接用设备中预载的数千款已经虚拟 化的常用程序, 或者将本机应用程序进行虚拟化处理, 从而预存所有程序文件 与程序系统环境文件在设备中, 见附图 7所示。 (虚拟化原理前面已经阐述了, 主要是利用沙盒的虚拟化技术, 先把应用程序安装到运行中所有的动作都记录 起来并处理成本地的文件, 当执行主程序文件吋, 它会临吋产生一个虚拟环境 来执行, 类似影子系统一样, 一切涉及的操作都是在这个虚拟环境中完成, 并 不会去动原本的系统。 这样处理后所有的调用文件都在应用程序的存放目录, 也就是 SLC NAND闪存模块下, 而不会安装到硬盘。 ) 这一处理最彻底地避免 了程序使用中的硬盘读写, 有效增加缓存命中度。
[0074] 以上所述乃是本发明的具体实施例及所运用的技术手段, 根据本文的揭露或教 导可衍生推导出许多的变更与修正, 若依本发明的构想所作的等效改变, 其所 产生的作用仍未超出说明书及附图所涵盖的实质精神吋, 均应视为在本发明的 技术范畴之内, 合先陈明。
[0075] 参考文献: [1].王芳, 李恪, 苏林, 耿立红. 空间太阳望远镜的星载固态存储 器研制 [ J ] . 电子学报, 2 0 0 4年第 3期: 4 7 2 - 4 7 5。
[0076] [2].王超, 张惠臻, 周学海, 马宏星. 异质存储系统中的高速缓存机制研究 [
J ] . 电子学报, 2 0 11年第 6期: 1267-1271。
工业实用性
[0077] 现今的计算机性能主要受制于 I/O性能。 根据目前的工业水平, 以及可预见的 未来的工艺增长, 可以预见到速度更快而成本更低的 SLC与拟 SLC闪存是能够广 泛量产的, 而将其与 DRAM缓存及并行多通道方案结合, 来充当计算机的多层 次缓存, 提升缓存速度的同吋, 起到保护闪存寿命的作用, 充分发挥其读写性 能尤其是随机读写性能, 将是一种重要的可广泛应用的新设备。 须知闪存的读 写幵销差异较大, 将其与 DRAM缓存及并行多通道方案结合后, 一方面能够最 大化其读速度, 另一方面也将写幵支尤其是小文件的频繁写幵支, 大量分配给 D RAM缓存承担, 最大化了写速度。 统计表明, 一般用户使用计算机, 读操作多 而写操作少, 因此这样的分级结构无需采用很大的 DRAM缓存便也足够。
[0078] 采用即插即用的 USB接口与 USB优化, 则可以在更方便使用的同吋保障性能不 被影响。 随着 USB带宽的继续上升, 其应用当更普及。 序列表自由内容
在此处键入序列表自由内容描述段落。

Claims

权利要求书
一种即插即用的通用串行总线接口的电子设备, 该电子设备包含 有主控芯片以及 SLC NAND闪存模组 (或者以 MLC
NAND模拟 SLC工作状态, 如 iSLC NAND闪存是在 MLC NAND基 础上改进的产品, 是通过特定的闪存管理算法, 把 MLC
NAND的 2-Bit Per Cell重新编程为 1 -Bit Per Cell, 让 MLC NAND的 工作方式变得更接近 SLC NAND) , 且该设备的工作原理具备两 个核心: 一、 设备通过 USB接口与计算机连接, 在 SLC NAND模 块中为计算机创建缓存文件, (缓存文件可以包括: 系统与应用 程序的常用文件、 预读频繁读写的零散文件、 随机数据等) , 利 用 SLC NAND闪存的高速随机访问以及快速读写速度, 减少计算 机系统对硬盘 (包括基于 NAND的固态硬盘) 的访问, 为计算机 提供加速并提升 I/O性能; 二、 设备并且使用 DRAM缓存, 可以通 过如下两种方式中的任意一种或两种实现: 一、 在设备中自带 DR AM缓存作为映射表和数据缓存区, 比如以每 1MB的 DRAM缓存映 射 1GB的 SLC NAND; 二、 在建立缓存吋调用计算机的内存, 划 分一部分与设备中的 SLC NAND共同构成高速快取的 cache, 从而 可以利用 DRAM与 SLC NAND不同的特点进行任务分配。
根据权利要求 1的一种设备, 其特征在于, 设备还对 USB协议进行 改善, 例如对于传统的 USB接口协议中阻碍快速数据传输的 BOT 协议进行优化, 在 USB传输协议上做了系统资源分配的优化, 提 供更多资源给 USB设备。
根据权利要求 1的一种设备, 其特征在于, 设备的算法与架构还采 用了如下设计: 设备将应用程序进行虚拟化处理, 从而预存所有 程序文件与程序系统环境文件在设备中 _虚拟化原理有多种, 主要是利用沙盒的虚拟化技术, 先把应用程序安装到运行中所有 的动作都记录起来并处理成本地的文件, 当执行主程序文件吋, 它会临吋产生一个虚拟环境来执行, 类似影子系统一样, 一切涉 及的操作都是在这个虚拟环境中完成, 并不会去动原本的系统, 这样处理后所有的调用文件都在应用程序的存放目录, 也就是 SL C NAND闪存模块下, 而不会安装到硬盘, 这一处理最彻底地避 免了程序使用中的硬盘读写。
根据权利要求 1的一种设备, 其特征在于, 设备的算法与架构还采 用了如下设计: 通过对用户习惯进行长期监测识别, 判断出系统 即将使用哪些数据, 预存在根据权利要求 1的该种设备中, CPU将 直接从设备中获取数据, 再将其转入内存中, 从而减少对硬盘的 读写。
根据权利要求 1的一种设备, 其特征在于, 设备的算法与架构还采 用了如下设计: 多通道模式, 阵列模组集成多块 SLC NAND闪存 芯片并采用多通道主控一具备可选的阵列模组, 该阵列模组将 多块 SLC NAND闪存或 3D V-NAND芯片集成在一起, 并采用多通 道主控, 从而能够以双通道或多通道模式工作, 例如, 多个物理 芯片组成一个阵列, 作为一个逻辑磁盘组, 将数据以分段的方式 存储在这个逻辑磁盘组的不同物理磁盘上, 进行数据存取吋, 阵 列中的相关磁盘并行工作, 减低数据存取的吋间, 从而达到和 RAI D O (磁盘阵列 0) 相同的加速效果, 使读写速度更快。
根据权利要求 1的一种设备, 其特征在于, 设备的算法与架构还采 用了如下设计: 具备两重缓存, 除了 SLC NAND闪存模块之外还 具有 MLC NAND闪存模块, 以 SLC
NAND闪存模块为一级缓存, MLC NAND闪存模块为二级缓存。 根据权利要求 1的一种设备, 其特征在于, 设备的算法与架构还采 用了如下设计: 对系统内存提供智能压缩与后台自动释放, 从而 避免计算机因内存不足调用虚拟内存而增大对硬盘的读写。
根据权利要求 1的一种设备, 其特征在于, 设备和远程服务器通信 , 用户通过设备将程序安装到远程服务器, 然后由远程服务器的 云计算来分担本地宿主计算机的程序运行、 游戏运行等计算任务 , 在本地计算机上显示用户界面: 为实现该目的, 设备还可能包 含了 3G网络模块或更高速的网络模块, 用于和远程服务器传输数 据, 而无需借用宿主计算机的网络, 从而可以在网络环境不好的 地区实现云计算功能; 如果只是将用户界面以及键盘, 鼠标动作 和屏幕的更新信息, 通过 3G模块用 3G网络在本地计算机和服务器 之间传递.传递的是键盘, 鼠标动作和屏幕的更新信息, 是图片增 量变化的那部分信息, 一般情况下,这种变化的信息只有几 K到几 十 K而已, 3G网络是没有问题的。
[权利要求 9] 根据权利要求 1的一种设备, 其特征在于, 该设备还为本地计算机 提供了安全验证机制的升级, 通过将该设备的硬件指纹 (闪存、 显卡、 通讯模块等均有硬件指纹) 和用户自行设置的密码组合作 为加密机制, 为电脑提供类似 U盾的文件防护、 计算机锁等功能。
[权利要求 10] 根据权利要求 1的一种设备, 其特征在于, 设备还提供即插即用的 操作系统, 可以通过设置 BIOS从 USB接口启动该设备预装在非易 失存储区内的操作系统。
PCT/CN2014/087627 2013-10-13 2014-09-28 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备 WO2015051711A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US15/028,028 US20160253093A1 (en) 2013-10-13 2014-09-28 A new USB protocol based computer acceleration device using multi I/O channel SLC NAND and DRAM cache

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201310475462.4 2013-10-13
CN201310475462.4A CN103500076A (zh) 2013-10-13 2013-10-13 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备

Publications (1)

Publication Number Publication Date
WO2015051711A1 true WO2015051711A1 (zh) 2015-04-16

Family

ID=49865292

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2014/087627 WO2015051711A1 (zh) 2013-10-13 2014-09-28 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备

Country Status (3)

Country Link
US (1) US20160253093A1 (zh)
CN (1) CN103500076A (zh)
WO (1) WO2015051711A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113064843A (zh) * 2021-03-24 2021-07-02 深圳市时创意电子有限公司 一种固态硬盘的制造方法与固态硬盘
US11344818B2 (en) 2018-10-04 2022-05-31 Acer Incorporated Computer system, game loading method thereof and computer readable storage medium

Families Citing this family (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11614893B2 (en) 2010-09-15 2023-03-28 Pure Storage, Inc. Optimizing storage device access based on latency
CN103500075A (zh) * 2013-10-11 2014-01-08 张维加 一种基于新材料的外接的计算机加速设备
CN103500076A (zh) * 2013-10-13 2014-01-08 张维加 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备
CN104035897B (zh) * 2014-06-12 2018-04-24 上海新储集成电路有限公司 一种存储控制器
CN104298620A (zh) * 2014-10-10 2015-01-21 张维加 一种耐擦写低能耗的外接计算机加速设备
CN104298474A (zh) * 2014-10-13 2015-01-21 张维加 一种基于服务端与外部缓存系统的外接式计算设备加速方法与实现该方法的设备
CN104598844A (zh) * 2014-12-31 2015-05-06 湖南国科微电子有限公司 一种防止固态硬盘系统文件损坏的掉电保护方法及装置
CN105867832B (zh) * 2015-01-19 2020-07-24 张维加 一种面向用户与应用的计算机与智能设备加速方法和装置
US10148755B2 (en) * 2015-08-27 2018-12-04 Dell Products L.P. System and method to redirect USB mass storage devices in high latency VDI environments
KR101923661B1 (ko) * 2016-04-04 2018-11-29 주식회사 맴레이 플래시 기반 가속기 및 이를 포함하는 컴퓨팅 디바이스
US10437798B2 (en) 2016-07-19 2019-10-08 Sap Se Full system simulator and memory-aware splay tree for in-memory databases in hybrid memory systems
US10474557B2 (en) 2016-07-19 2019-11-12 Sap Se Source code profiling for line-level latency and energy consumption estimation
US10387127B2 (en) 2016-07-19 2019-08-20 Sap Se Detecting sequential access data and random access data for placement on hybrid main memory for in-memory databases
US11977484B2 (en) * 2016-07-19 2024-05-07 Sap Se Adapting in-memory database in hybrid memory systems and operating system interface
US10452539B2 (en) * 2016-07-19 2019-10-22 Sap Se Simulator for enterprise-scale simulations on hybrid main memory systems
US10698732B2 (en) 2016-07-19 2020-06-30 Sap Se Page ranking in operating system virtual pages in hybrid memory systems
US10540098B2 (en) 2016-07-19 2020-01-21 Sap Se Workload-aware page management for in-memory databases in hybrid main memory systems
US10783146B2 (en) 2016-07-19 2020-09-22 Sap Se Join operations in hybrid main memory systems
CN106951194A (zh) * 2017-03-30 2017-07-14 张维加 一种新结构的计算机设备
US11010379B2 (en) 2017-08-15 2021-05-18 Sap Se Increasing performance of in-memory databases using re-ordered query execution plans
CN107957852B (zh) * 2017-10-13 2021-08-13 记忆科技(深圳)有限公司 一种提升固态硬盘性能一致性的方法
CN108646980A (zh) * 2018-04-27 2018-10-12 江苏华存电子科技有限公司 一种有效利用内存带宽的方法
US10909012B2 (en) * 2018-11-12 2021-02-02 H3 Platform, Inc. System having persistent memory
KR20200076491A (ko) * 2018-12-19 2020-06-29 에스케이하이닉스 주식회사 메모리 시스템 및 그것의 동작 방법
CN111858016A (zh) 2019-04-29 2020-10-30 阿里巴巴集团控股有限公司 计算作业处理方法、系统、移动设备及加速设备
KR20210068699A (ko) * 2019-12-02 2021-06-10 삼성전자주식회사 스토리지 장치, 스토리지 시스템 및 스토리지 장치의 동작 방법
CN114205344B (zh) * 2020-08-31 2023-02-28 华为技术有限公司 媒体文件传输的方法及装置
CN112115077B (zh) * 2020-08-31 2022-04-19 瑞芯微电子股份有限公司 一种dram内存驱动优化方法和装置
US11847324B2 (en) 2020-12-31 2023-12-19 Pure Storage, Inc. Optimizing resiliency groups for data regions of a storage system
US11614880B2 (en) 2020-12-31 2023-03-28 Pure Storage, Inc. Storage system with selectable write paths
CN117555492B (zh) * 2024-01-12 2024-03-29 合肥开梦科技有限责任公司 存储器控制器、存储器管理方法及存储装置

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20100042773A1 (en) * 2008-08-12 2010-02-18 Phison Electronics Corp. Flash memory storage system and data writing method thereof
CN101981555A (zh) * 2008-06-30 2011-02-23 英特尔公司 用于多级缓存利用的设备和方法
CN103500076A (zh) * 2013-10-13 2014-01-08 张维加 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN100403249C (zh) * 2006-06-19 2008-07-16 威盛电子股份有限公司 磁盘阵列及其数据存取方法
CN100458751C (zh) * 2007-05-10 2009-02-04 忆正存储技术(深圳)有限公司 并行闪存控制器

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101981555A (zh) * 2008-06-30 2011-02-23 英特尔公司 用于多级缓存利用的设备和方法
US20100042773A1 (en) * 2008-08-12 2010-02-18 Phison Electronics Corp. Flash memory storage system and data writing method thereof
CN103500076A (zh) * 2013-10-13 2014-01-08 张维加 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11344818B2 (en) 2018-10-04 2022-05-31 Acer Incorporated Computer system, game loading method thereof and computer readable storage medium
CN113064843A (zh) * 2021-03-24 2021-07-02 深圳市时创意电子有限公司 一种固态硬盘的制造方法与固态硬盘

Also Published As

Publication number Publication date
US20160253093A1 (en) 2016-09-01
CN103500076A (zh) 2014-01-08

Similar Documents

Publication Publication Date Title
WO2015051711A1 (zh) 一种基于多通道slc nand与dram缓存的新usb协议计算机加速设备
US9417794B2 (en) Including performance-related hints in requests to composite memory
TWI614752B (zh) 經由記憶體通道關閉之節電技術
CN107391397B (zh) 支持近存储器和远存储器访问的存储器通道
US8392670B2 (en) Performance management of access to flash memory in a storage device
KR102114109B1 (ko) 데이터 저장 장치
KR20120132820A (ko) 스토리지 디바이스, 스토리지 시스템 및 스토리지 디바이스의 가상화 방법
US10067676B2 (en) Hybrid memory drives, computer system, and related method for operating a multi-mode hybrid drive
US9164804B2 (en) Virtual memory module
WO2015051694A1 (zh) 一种基于新材料的外接的计算机加速设备
TWI526828B (zh) 資料存取方法及使用此方法的記憶體控制器與儲存裝置
CN112115067A (zh) 闪存物理资源集合管理装置及方法及计算机可读取存储介质
CN111796759A (zh) 多平面上的片段数据读取的计算机可读取存储介质及方法
WO2016058560A1 (zh) 一种基于服务端与外部缓存系统的外接式计算设备加速方法与实现该方法的设备
CN109799959A (zh) 一种提高开放通道固态盘写并行性的方法
US11775188B2 (en) Communications to reclaim storage space occupied by proof of space plots in solid state drives
US20130173855A1 (en) Method of operating storage device including volatile memory and nonvolatile memory
CN111813703A (zh) 数据储存装置及逻辑至物理地址映射表的更新方法
KR20200114086A (ko) 컨트롤러, 메모리 시스템 및 그것의 동작 방법
CN104298620A (zh) 一种耐擦写低能耗的外接计算机加速设备
CN114077557A (zh) 执行读取/写入操作的方法和托管客户端装置的计算系统
KR102425470B1 (ko) 데이터 저장 장치 및 그것의 동작 방법
US20210240642A1 (en) Data storage device with an exclusive channel for flag checking of read data, and non-volatile memory control method
US9208073B2 (en) Firmware storage and maintenance
US20240118829A1 (en) Relocation of data in a storage device based on access information

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 14852814

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 15028028

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 14852814

Country of ref document: EP

Kind code of ref document: A1