WO1997024725A1 - High performance universal multi-port internally cached dynamic random access memory system, architecture and method - Google Patents
High performance universal multi-port internally cached dynamic random access memory system, architecture and method Download PDFInfo
- Publication number
- WO1997024725A1 WO1997024725A1 PCT/IB1996/000794 IB9600794W WO9724725A1 WO 1997024725 A1 WO1997024725 A1 WO 1997024725A1 IB 9600794 W IB9600794 W IB 9600794W WO 9724725 A1 WO9724725 A1 WO 9724725A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- data
- dram
- serial
- buffers
- interface
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C7/00—Arrangements for writing information into, or reading information out from, a digital store
- G11C7/10—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers
- G11C7/1075—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers for multiport memories each having random access ports and serial ports, e.g. video RAM
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F12/00—Accessing, addressing or allocating within memory systems or architectures
- G06F12/02—Addressing or allocation; Relocation
- G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
- G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
- G06F12/0893—Caches characterised by their organisation or structure
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F13/00—Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
- G06F13/14—Handling requests for interconnection or transfer
- G06F13/16—Handling requests for interconnection or transfer for access to memory bus
Definitions
- the present invention relates to dynamic random access memory technology (DRAM), being more specifically directed to novel DRAM system architectures that eliminate current system bandwidth limitations and related problems and provide significantly enhanced system performance and at reduced cost, enabling substantially universal usage for many applications as a result of providing unified memory architecture.
- DRAM dynamic random access memory technology
- bus-based architecture where a single system bus interconnects the CPU, the main memory and the I/O resources as shown in later-described Fig 1, (the terms 'main memory' and 'system memory' as herein used, being so used interchangeably).
- Fig 1 the terms 'main memory' and 'system memory' as herein used, being so used interchangeably.
- main memory generally implemented with DRAM
- an arbitration takes place for access to the system bus.
- the amount of concurrent activity in the system is limited by the overall capacity of the external bus.
- a typical networking equipment such as switches, routers, bridges, hubs, etc. interconnect multiple networks such as ATM, SONET, Token Ring, FDDI, Ethernet, Fiber Channel, etc. as shown in later-described Fig 2.
- a typical design includes a high performance CPU and a large amount of main memory generally implemented with the use of a traditional DRAM as represented in later-described Figs 3 and 4. Data from various networks is transferred to the main memory in the form of packets (a packet is a collection of bytes), processed by the CPU and then, in general, forwarded to their respective destination networks.
- All the networks mentioned above provide different means of transporting data from one point to another. They differ in hardware, software and data transfer speeds. Interconnect equipment is required to allow the users on one of these networks to communicate with the users on another network with different protocol, seamlessly.
- the network interfaces are implemented with a network interface controller (also commonly known as network controller), unique for each type of interface. Thus Ethernet has a different network interface than for Fiber Channel or ATM (Figs 3 and 4).
- system bus is 32 bit wide (4 bytes);
- the packet size is 1024 bytes.
- a packet is sent by a user on, for example, the Ethernet network to a user, for example, on the FDDI network.
- This packet is received by the interconnect equipment Ethernet interface controller and is analyzed by the controller chip, with only the relevant information content stored in its conventional local FIFO (First In First Out) memory, for subsequent transfer to the main memoiy.
- FIFO First In First Out
- arbitration takes place among all the active resources to acquire the system memory bus.
- data is then forwarded to the system memoiy using the 32 bit wide system bus interface. As there are 1024 bytes in the packet and 4 bytes are transferred to the main memory per transfer, 256 such transfers are required to move the packet.
- the number of arbitrations may be smaller if burst transfer capability is provided by the network controller. As an example, a 16 byte burst transfers capability for every acquisition, need minimum 64 arbitration cycles).
- this packet is stored in the main memoiy, it is processed by the CPU (primarily the header information) and redirected towards the FDDI port, in this example. Now the reverse process takes place.
- the data is picked up by the FDDI interface controller from the main memory and transferred to the chip internal FIFO memory. This also requires 256 transfers and a corresponding number of arbitrations. Data is then concurrently transferred from the FDDI controller chip to its network.
- FDDI operates at 100 mega bits per second, Ethernet at 10/100 mega bits per second, ATM at nearly 600 mega bits per second, Token ring at 16 mega bits per second, and Fiber Channel at 800 mega bits per second.
- the number of network interfaces increases or higher speed interfaces are added the amount of time available to each resource, including CPU, decreases, thus clipping the peak performance of the system. It also forces the designers to go for even higher performance CPUs and associated expensive components, thus driving up the cost.
- the number of networks which can be connected via this prior art type of system bus also remains low due to these severe limitations, and this problem becomes worse as more and more networks at higher and higher speeds are added to accommodate, for example, the Internet related expansion activity.
- the first task requires a large number of frequent data transfers, called 'BitBlt', from one place in the memory to another; but this requirement tends to be bursty in nature. This consumes a considerable portion of the system bandwidth and thus has necessitated the use of separate memory to store graphics data, as shown in later- described Fig 5, thereby adversely affecting the system cost.
- 'BitBlt' a large number of frequent data transfers
- VRAM Video DRAM
- SAM Serial Access Memory
- a 256K x 8 VRAM also has an additional port which is 8 bits wide to stream refresh data to the CRT continuously.
- the 'SAM' buffer has a fix connection with the external display interface.
- the CPU accesses the VRAM via the system data interface and a screen image is stored or updated in the VRAM. Then the screen data of one entire row is moved to the 'SAM' buffer in one access. This data is then subsequently transferred to the display via the SAM I/O interface, identical in width to the system interface.
- VRAMs provide an acceptable solution for the case where the design has to interact with only one graphics source/destination. They are, however, more expensive than traditional DRAMs due to the larger number of additional pins and the bigger silicon die, and the architecture provides a very rigid structure. Expandability to more interfaces with more devices is severely limited as the number of pins grows significantly. The 'SAM' connection to the external I/O interface is fixed and the data port size is also predetermined. This approach also does not solve the problem of speeding the huge data movement requirements. Thus the VRAMs provide an acceptable solution but only because of lack of any better altemative, until the present invention.
- VRAMs (specifically the 3 port version) were also occasionally touted towards networking applications but have rarely been so used due to their before-stated rigid I/O structure, veiy limited capability to interconnect number of resources (to be
- VRAMs are both the graphics and the main memory components; but the added cost of the components neutralizes the gains achieved by having a common memory.
- Another potential solution is to use the before-described RDRAM, which has a smaller number of pins per chip than VRAM, thus resulting in lower power consumption, smaller real estate and relatively lower cost. But unfortunately because of their block-oriented protocol and interface limitations, it is highly inefficient with non-localized main memory accesses, and thus does not render itself well to the concept of 'Unified Memory Architecture'.
- Use of RDRAM also poses a host of significant electrical engineering design challenges related to the emissions, noise interference and PCB layout, making the design task very difficult.
- e. is configurable to accommodate different data transfer rates of the I/O resources
- g. provides low pin count
- An objective of the invention accordingly, is to provide a new and improved dynamic random access (DRAM) system, architecture and method utilizing a novel multi-port internally cached DRAM structure that obviates cu ⁇ erit system bandwidth limitations and related problems, while providing significantly enhanced system performance at reduced cost, and which thereby enable substantially universal usage for myriads of applications.
- DRAM dynamic random access
- a further object is to provide such a novel system in which the transfer of blocks of data internal to the chip is an order of magnitude faster than traditional approaches, and with the facility to interconnect significantly higher numbers of resources with substantially enhanced performance and at notably lower cost.
- Still a further object is to provide a system configuration based on this novel architecture that works equally efficiently in both main memory functions and as graphics memory - a truly high performance unified memoiy architecture.
- the invention embraces for use in a system having a master controller such as a central processing unit (CPU) having parallel data ports and a dynamic random access memory (DRAM) each connected to and competing for access to a common system bus interface, an improved DRAM architecture comprised of a multi-port internally cached DRAM(AMPIC DRAM) comprising a plurality of independent serial data interfaces each connected between a separate external I/O resource and internal DRAM memory through corresponding buffers; a switching module interposed between the serial interfaces and the buffers; and a switching module logic control for the connecting of the serial interfaces to the buffers under a dynamic configuration by the bus master controller, such as said CPU, for switching allocation as appropriate for the desired data routability.
- AMPIC DRAM multi-port internally cached DRAM
- Fig 1 is a block diagram of a typical prior art single bus parallel architecture
- Fig 2 shows a typical prior art network configuration
- Figs 3 and 4 illustrate a typical prior art networking equipment employing DRAMs for use in a configuration such as that of Fig 2;
- Fig 5 is a block diagram of a prior art configuration of a graphics application with separate memories and using traditional DRAMs.
- Fig 6 is a similar diagram of a typical architecture of a graphics applications using VRAMs
- Fig 7 is a block diagram of a system architecture constructed in accordance with the present invention and embodying a multiple-port internally cached '(AMPIC) DRAM' of the invention
- Fig 8 is a similar view of a partial top level architecture of the 'AMPIC DRAM 1 of Fig 7, showing multiplex/crossbar switching between supplemental serial interfaces, buffers and the DRAM core;
- Fig 9 illustrates details of an illustrative serial data-multiplexer implementation in Fig 8.
- Fig 10 shows an example of the multiple serial interfaces configured as ports
- Fig 11 is a diagram of an exemplary 2-bit port of the AMPIC DRAM and associated control line
- Figs 12 and 13 are diagrams of examples of a serial data transfer format, with Fig 13 applied to the illustration of the 2 bits per port;
- Fig 14 is a block diagram of an example of a partial top level of a two-bank 'AMPIC DRAM' control module architecture with a later-described parallel row internal transaction intervention (called PRITI), without storage elements;
- PRITI parallel row internal transaction intervention
- Fig 15 shows the sequence of operations of the PRITl transfers with internal data transfer from the two banks
- Fig 16 is similar to Fig 14 but with two row wide sets of storage elements provided for the 'PRITl' capability;
- Fig 17 illustrates the 'PRITl' transfer, with two row wide sets of storage elements of Fig 16, showing the sequence of operations and internal data exchange between the two banks;
- Fig 18 is similar to Fig 17, but employs only one row wide set of storage elements
- Fig 19 presents an example of a useful pinout for an 'AMPIC DRAM' of the invention with an exemplary 9 serial interfaces
- Fig 20 illustrates an exemplary networking equipment architected with AMPIC DRAMS of the invention with a 32 bit wide system bus operating with the CPU;
- Fig 21 is a similar diagram for a graphics application
- Fig 22 is a similar diagram illustrating a four bank system configuration with each bank is connected to different network interfaces
- Fig 23 is similar to Fig 22 but uses two banks of 'AMPIC DRAMs' and two banks with traditional DRAMs;
- Fig 24 is also similar to Fig 22 but with two banks for graphics and two for other applications;
- FIG 25 is a diagram of still a further modification of an MPIC DRAM' architecture in which there are two internal banks and the before-mentioned 'PRITl' capability, wherein one bank is for main memoiy usage and the other bank for graphics or other applications.
- Fig 26 is a modification of the AMPIC DRAM system of before mentioned Fig 19, adapted for use with a so-called 'PARAS' interface and access, described in copending US patent application serial no. 08/320,058, filed October 7, 1994, and with a low pin count for the integrated memory architecture.
- This application discloses a method of and apparatus for improving the accessing capability of asynchronous and synchronous dynamic random access memory devices by a novel interfacing and accessing procedure in which the same pins are used for each of row, column and data accessing and in both the write and read cycles; such enabling effective increasing of the data bandwidth and addressing range in substantially the same size packages with fewer pins.
- Fig 27 is a block diagram of an example of a partial top level of a multi-bank 'AMPIC DRAM' control module architecture with the before-described parallel row internal transaction intervention (PRITl), with a one row wide set of storage elements.
- PRITl parallel row internal transaction intervention
- a CPU unit so labelled, using conventional primary parallel port data, is shown connected to a system bus, also connected with the main memory unit containing the later-described 'AMPIC DRAM' of the invention, and into which supplemental serial interface inputs (# 1 thru #n) are connected from respective input/output (I/O) resources #1 through #n.
- serial interfaces are thus provided on the 'AMPIC DRAM' to transport data between the I/O resources and the main memoiy.
- These serial interfaces are in addition to the primary parallel ports for the system bus interface for use by the central processing unit CPU or similar master controller device.
- the number of such serial interfaces is limited only by the device technology, pin count, power consumption and cost, etc.
- the serial data received or data to be transmitted via these interfaces # 1 through #n is stored inside the 'AMPIC DRAM' in small respective buffers #1 through #n, as more fully shown in Fig 8. For practical considerations, this could range from 64 bytes to 512 bytes, but in theory is limited by the layout of the sense amplifiers.
- a multiplexer and/or crossbar switch logic or combination connects the 'n' serial interfaces to the 'm' buffers.
- the connection between each serial interface and a buffer is dynamically configured by the CPU (or the current system bus master) and is altered as appropriate for data routability.
- FIG 9 A functional block diagram of one possible serial data interface implementation is shown in Fig 9, for the configuration of 4 serial interfaces and 4 buffers. There are, however, multiple ways to implement the desired architecture, though the basic idea remains the same.
- arbitration among the various active packet buffers and the CPU is performed. No arbitration, however, is required to receive or transmit data from/to the packet buffers via the serial interfaces.
- the incoming packet buffer can be redefined as output packet buffer and data rerouted to its destination without even performing the intermediate step of transferring the data between the buffers and the core DRAM. This reduces the latency involved in reception of an incoming packet and its subsequent transmission to its destination. This is possible only because of the 'AMPIC DRAM' capability to assign any buffer to any serial interface via the Mux/Crossbar switch module.
- multiple one bit wide serial interfaces can also be configured as a narrow width bus (termed "port") of sizes such as 1, 2, 4 or 8, etc, but in theory there is no such limitation. It can be any number from 1 to 'n' permissible by the device technology and is also subject to implementation.
- SUBSTTTUTE SHEET (RULE 26) port is the same as a serial interface. This allows even faster data transfers and maintains flexibility at the same time, and is very useful in interfacing with resources operating with different bandwidth and data transfer requirements. It also follows that each packet buffer, Fig 8, has the capability to interface with a maximum of 'n' serial interfaces simultaneously, if defined as a port. The buffers get configured for the same port size identical to the port to which it is connected (also sometimes termed docked).
- Each port interface consists of one control and a set of serial data interfaces. As an example, if each serial port is only 1 bit wide, then one control line is used per one bit of data line. If two serial interfaces are configured as one port, then one control line is used for the two bit port and so on, as shown in Fig 11. To minimize the pin count, moreover, a serial interface control line is also configurable as a serial data interface line so long as the criterion of one control line per port is met. The association of each control line to its port is configurable. The purpose of the control lines is to control the data flow between the I/O resources and the serial ports. If pin count is not a concern for the manufacturer, one certainly could provide separate control pins. For some applications, furthermore, control pins are not even required, and the I/O resource parallel interface to the system bus is sufficient to exchange the control information.
- the data transfer format between an I/O resource and the 'AMPIC DRAM' serial port is such that each memoiy chip (part of the same external bank) receives and transmits data bits on its port simultaneously, as illustrated in Fig 12.
- each memoiy chip part of the same external bank
- Fig 12 The data transfer format between an I/O resource and the 'AMPIC DRAM' serial port is such that each memoiy chip (part of the same external bank) receives and transmits data bits on its port simultaneously, as illustrated in Fig 12.
- the chipO receives bitO
- chipl receives bit8, chip2 receives bitl ⁇ ; and chip3 receives bit24.
- all the bit numbers will be incremented by 1. This will continue until all 32 bits have been transferred, so that each chip received its 8 bits. Once completed, this process will be repeated for, as illustrated, in Fig 12, the next 32 bits, and so on.
- a total 8 bit interface is then provided to the I/O resource, which must provide two bits each, to each 'AMPIC DRAM' simultaneously.
- the ordering of the bits is such that chipO receives bitO and bitl, chipl receives bit8 and bit9, chip2 receives bitl ⁇ and bitl7; and chip3 receives bit24 and bit25 concurrently.
- all the bit numbers will be incremented by 2. This will continue until all 32 bits have been transferred, so that each chip received its 8 bits. Once completed, this process will be repeated for the next 32 bits, and so on.
- this architecture of the invention does not prevent the I/O resources, such as network controller chips, to share the parallel system bus, if so desired. It could be useful for tasks such as controller configuration and status management.
- the 'AMPIC DRAM' is provided with one master clock pin as shown in Fig 19, and every serial interface is configured to operate at a multiple or submultiple of this clock rate, allowing flexibility to accommodate a variety of resources. It is also possible to provide more than one independent clock instead of one master clock, the limitation being only the device technology, pin count, and cost constraints. It should be noted that the clock frequency assignment is a characteristic of the serial interface and not of the buffers. Thus, any of the 'm' buffers can be docked to any of the serial ports and operate at that port speed.
- the 'AMPIC DRAM' configurability of the invention allows switching of the serial interface from one buffer to another buffer without intemipting the transfers. This has a number of important applications in networking and graphics.
- One buffer can be loaded with row wide data in one access, while the other one is being used to transmit the information.
- split buffer transfers this is quite different from the present invention in that the external I/O interface with the buffer is always fixed and of the same width as the VRAM system data width.
- the 'AMPIC 's multiplexer/crossbar switch module of this invention totally eliminates all such limitations.
- SUBSTTTUTE SHEET (RULE 26) accesses to the accessed banks are not permitted. It should be noted that the transfer on the serial interfaces can also go on in parallel to this internal transfer. A similar concept, later contrasted from the features of the present invention, is disclosed in US Patent No. 5473566, issued on December 5, 1995.
- the process is identical for any number of banks. If there are 'm' internal banks connected via a row wide interface, the 'PRITl' module is capable of transferring data from a bank to more than one remaining bank, simultaneously. This is very useful when a broadcast packet is moved from one bank to all other internal banks. With this invention, no row wide registers or latches (also termed a row wide set of storage elements) are required to perform this operation, thus resulting in a very cost effective implementation.
- 'PRITl' The top level internal structure of the 'PRITl' module is shown in before-described Fig 14.
- 'PRITl' is loaded with the starting row addresses of each bank and the transfer count. After configured, it arbitrates to acquire the internal buses of both the banks.
- 'PRITl' module being
- SUBSTTTUTE SHEET (RULE 26) configured for a predetermined number of burst tiansfers once it acquires the right to access the rows, or release the bus after every transfer to allow other resources to share the DRAM core.
- An altemative embodiment of this invention involves using a row wide set of storage elements as labelled in Fig 27 (or any implementation capable of performing a logically equivalent task) to perform a data exchange operation.
- the said set of storage elements would contain 1024 storage elements.
- a row of an internal bank is accessed with read operation, and the retrieved data at the sense amplifiers for the bank (call it bank2) is stored in the row wide set of storage elements.
- Data is then retrieved from the other bank (bankl) and is then written to bank2. Subsequent to this operation, data from the storage elements is written to the bankl .
- the diagram of Fig 18 shows a suitable sequencing of such an operation.
- This implementation requires less circuit than the later-described approach with two sets of storage elements and still permits data exchange, though at the cost of somewhat slower execution.
- This approach is a universal approach for banks 1 to 'm'. This capability permits a massive information exchange in a very short time, an exceptionally useful tool for multimedia/graphics applications.
- This implementation obviously requires more circuit than the original approach due to the addition of a set of storage elements, but here it is not necessary to save the original data before new data is moved at its place.
- FIG. 16 Another modification of this invention uses two sets of storage elements as labelled in Fig 16 (or any circuit capable of performing a logically equivalent task) to perform a data exchange operation.
- a row in each of two internal banks is accessed simultaneously with read operation, and the retrieved data at the sense amplifiers is stored in the row wide set of storage elements, as indicated in Fig 16.
- Data so retrieved and then stored is subsequently written back to both the rows simultaneously.
- the diagram of Fig 17 shows exemplary sequencing of such an operation
- the 'PRITF approach of the invention therefore, is not restricted to two internal banks only, and is equally applicable for any multibank oiganization inside a DRAM chip. It is also possible to add 'PRITl' capability even in a traditional type DRAM without the rest of the 'AMPIC DRAM' architecture A more sophisticated 'PRITl', furthermore can also have the transfer bounda ⁇ es defined in terms of columns in addition to rows requiring additional registers to load the column addresses
- more than one row wide bus is implemented along with their own set of storage elements connecting multiple banks, thus allowing more than one parallel 'PRITl' transfers
- the maximum number of possible row wide buses without redundancy is 'm/2'
- W2' sets of stoiage elements are required to make 'm' separate simultaneous 'PRITl' tiansfei s, one for each bank, b. If the number of banks is large, then the banks can be subgrouped on separate buses.
- 4 banks can reside on one bus while the other four are on the second bus with their own above-described 'PRITl' transfer implementation, and these two sub groups are then connected via another bus with any of the above-described 'PRITl' transfer capabilities.
- the chip has somewhat different pin out to reflect its unique architecture.
- One possible pinout for a 2M x 8 chip with 9 serial interfaces is shown exemplarily in Fig 19, with the added pins requiring changes in the interface design of the 'AMPIC DRAM'-based main memory.
- a 'WAIT' signal is provided for the system bus interface, Fig 19, whenever an internal transfer between the buffers and the DRAM core is taking place.
- the CPU or other master controller
- the CPU can either use it to delay the start of access or in an altemative implementation, the access cycle can be extended to allow for the internal transfer to complete, before proceeding with this access.
- a master clock pin (“master clock"), as explained earlier, is also provided.
- a mechanism is required to differentiate between a regular DRAM core access versus either a configuration command or buffer transfer interaction.
- the approach in Fig 19 is to provide an additional control signal to signify a command or data access.
- the command instruction can be carried over the data lines as they are not used during 'RAS' cycle. This is specifically useful for internal transfer commands, where a DRAM core address need be provided along with the buffer ID. This scheme allows use of traditional signals 'RAS' and 'CAS' to provide the core DRAM address, where the data lines will have the buffer number or any other additional information/instruction. In fact it is possible to issue two commands; one, when 'RAS' goes active, and then when 'CAS' is asserted. There are multiple known ways to implement this access mechanism, such also being a function of the device technology and cost considerations.
- the 'AMPIC DRAM' of the invention has more pins than the traditional DRAMs due to serial ports, if the earlier proposed 'PARAS' type DRAM model of said pending application is used, one could have this DRAM with only a marginal increase in number of pins.
- serial interfaces/ports are provided between each network controller and the main memory.
- the data movement between the controllers and the main memory is primarily serial.
- the serial data received from a network controller, or data to be transmitted to a network controller is stored in a packet buffer, assigned to it by the system bus master.
- This discussion assumes that the nerwork controllers are capable of sourcing or receiving serial data streams in the format required by the new system architecture.
- a user on, for example, the Ethernet network is sending a packet of 1024 bytes to another user on, for example, the FDDI network.
- the data is to be received by the serial port on the 'AMPIC connected to the Ethernet controller. No arbitration is required and no main memoiy bandwidth is consumed on the transfers.
- each of the four 'AMPIC DRAMs' would receive 256 bytes
- it can be transferred entirely to the DRAM core in only one access after this packet buffer acquires the internal bus via arbitration.
- a row address is provided to the DRAM core, its sense amplifiers have all the data bits of this row available.
- the entire packet buffer can be stored in one access. If the size of the packet buffer is less than one row wide, then multiple, though still few, accesses are required.
- this packet is transferred to the 'AMPIC DRAM' core, it is processed by the CPU and redirected towards the FDDI port in this example. Now the reverse process takes place.
- the packet is transferred from the core to the appropriate packet buffer in a single access requiring arbitration.
- This data is subsequently transferred from the packet buffer to the FDDI contioller via the serial ports and then concurrently moved from the FDDI controller chip to its network.
- SUBSTTTUTE SHEET (RULE 26) arbitrate only once for its transfer, whereas 256 transfers and corresponding arbitrations are required in existing designs.
- the 'AMPIC DRAM' can also be configured as previously stated to provide the graphics screen data to the display at high speeds.
- serial interfaces per chip have been defined as a port and used to transfer the display data
- d. data retrieval rate per port is at 100 Mhz (can be faster than this rate).
- one external bank with a 32 bit wide system bus was used. Some applications, however, may use more than one external bank, such as 4 banks, 32 bit wide each of 'AMPIC DRAM', as shown in Fig 22.
- This architecture permits that different network interfaces can be connected to each bank, if so desired. This can increase the network interconnectivity to a much larger level within reasonable cost as compared to the prevailing solutions. As an illustration, if 9 serial interfaces were provided on each 'AMPIC DRAM' and each bank was connected to 4 network interfaces, then a total of 16 networks can be connected. This represents a major gain compared to the existing technology which generally peaks out at 4 to 5 interfaces.
- the 'AMPIC DRAM' can be used to provide both the graphics or display interface, as depicted in Fig 24, and to connect to other types of I/O resources such as, for example, video cameras, or a satellite interface or like.
- a system level solution based on the 'AMPIC DRAM' of the invention provides the configurable serial interfaces and 'PRITl' capability that certainly fills this void. It has more signal pins (but the number of power and ground pins may be smaller) than the one proposed by the RAMBUS, but less than VRAMs, and is equally efficient at both the
- Another alternate embodiment for such solution is to have two intemal banks in the chip.
- One may follow the 'AMPIC model for graphics or similar applications, while a second intemal bank, possibly larger, may resemble the traditional DRAM based memory, and with both banks sharing the 'PRITl' capability of the invention, as in Fig 25.
- This potent integration provides the best of both worlds; one bank appears like main memory, while the other bank appears as an optimized graphics memory.
- This chip architecture allows massive amounts of data transfers between the two internal banks with very little impact on the system bandwidth due to the 'PRITl' capability, and thus provides all the necessary capabilities needed to allow a common chip and one single bus, universally for all types of applications.
- the device has low pin count for the functionality provided.
- the architecture reduces the latency time between the reception of an incoming packet and its subsequent transmission.
- the system design interface is nearly identical to the existing DRAMs, thus minimizing the design cycle.
Abstract
Description
Claims
Priority Applications (11)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE69610714T DE69610714T2 (en) | 1995-12-29 | 1996-08-12 | SYSTEM, ARCHITECTURE AND METHOD WITH HIGH PERFORMANCE FOR A UNIVERSAL MULTIPORT DYNAMIC DIRECT ACCESS MEMORY WITH INTERNAL CACHE MEMORY |
DK96925929T DK0870303T3 (en) | 1995-12-29 | 1996-08-12 | System, architecture and high performance approach for a universal multiport dynamic storage with random access with inte |
KR1019980705020A KR100328603B1 (en) | 1995-12-29 | 1996-08-12 | General purpose multi-port internal cached DRAM system, architecture and method |
CA002241841A CA2241841C (en) | 1995-12-29 | 1996-08-12 | High performance universal multi-port internally cached dynamic random access memory system, architecture and method |
IL12513596A IL125135A (en) | 1995-12-29 | 1996-08-12 | Dram architecture |
EP96925929A EP0870303B1 (en) | 1995-12-29 | 1996-08-12 | High performance universal multi-port internally cached dynamic random access memory system, architecture and method |
JP50793497A JP3699126B2 (en) | 1995-12-29 | 1996-08-12 | High performance universal multiport internal cache dynamic random access memory system, structure and method |
AT96925929T ATE197101T1 (en) | 1995-12-29 | 1996-08-12 | HIGH PERFORMANCE SYSTEM, ARCHITECTURE AND METHODS FOR A UNIVERSAL MULTIPORT DYNAMIC RANDOM ACCESS MEMORY WITH INTERNAL CACHE MEMORY |
AU65295/96A AU721764B2 (en) | 1995-12-29 | 1996-08-12 | High performance universal multi-port internally cached dynamic random access memory system, architecture and method |
HK99103219A HK1018342A1 (en) | 1995-12-29 | 1999-07-27 | High performance universal multi-port internally cached dynamic random access memory system, architecture and method |
GR20010400081T GR3035261T3 (en) | 1995-12-29 | 2001-01-17 | High performance universal multi-port internally cached dynamic random access memory system, architecture and method |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US08/581,467 | 1995-12-29 | ||
US08/581,467 US5799209A (en) | 1995-12-29 | 1995-12-29 | Multi-port internally cached DRAM system utilizing independent serial interfaces and buffers arbitratively connected under a dynamic configuration |
Publications (2)
Publication Number | Publication Date |
---|---|
WO1997024725A1 true WO1997024725A1 (en) | 1997-07-10 |
WO1997024725A9 WO1997024725A9 (en) | 1997-10-09 |
Family
ID=24325313
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/IB1996/000794 WO1997024725A1 (en) | 1995-12-29 | 1996-08-12 | High performance universal multi-port internally cached dynamic random access memory system, architecture and method |
Country Status (15)
Country | Link |
---|---|
US (2) | US5799209A (en) |
EP (1) | EP0870303B1 (en) |
JP (1) | JP3699126B2 (en) |
KR (1) | KR100328603B1 (en) |
CN (1) | CN1120495C (en) |
AT (1) | ATE197101T1 (en) |
AU (1) | AU721764B2 (en) |
CA (1) | CA2241841C (en) |
DE (1) | DE69610714T2 (en) |
DK (1) | DK0870303T3 (en) |
GR (1) | GR3035261T3 (en) |
HK (1) | HK1018342A1 (en) |
IL (1) | IL125135A (en) |
TW (1) | TW318222B (en) |
WO (1) | WO1997024725A1 (en) |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO1999005604A1 (en) * | 1997-07-28 | 1999-02-04 | Nexabit Networks, Llc | Multi-port internally cached drams |
FR2779843A1 (en) * | 1998-06-16 | 1999-12-17 | Busless Computers | Serial multi port memory component comprising RAM memory bank assemblies for use in computer |
WO2000030421A1 (en) * | 1998-11-12 | 2000-05-25 | Nexabit Networks, Inc. | Electronic circuit board |
WO2000058840A1 (en) * | 1999-03-26 | 2000-10-05 | Nexabit Networks, Inc. | Ampic dram system |
AU760297B2 (en) * | 1997-04-30 | 2003-05-08 | Canon Kabushiki Kaisha | Memory controller architecture |
CN100390755C (en) * | 2003-10-14 | 2008-05-28 | 中国科学院计算技术研究所 | Computer micro system structure comprising explicit high-speed buffer storage |
CN1573723B (en) * | 2003-05-15 | 2010-05-05 | 三星电子株式会社 | Method and apparatus for communication via serial multi-port |
WO2011094218A2 (en) * | 2010-01-29 | 2011-08-04 | Mosys, Inc. | Hierarchical multi-bank multi-port memory organization |
Families Citing this family (75)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6118776A (en) * | 1997-02-18 | 2000-09-12 | Vixel Corporation | Methods and apparatus for fiber channel interconnection of private loop devices |
JPH10283088A (en) * | 1997-04-02 | 1998-10-23 | Oki Electric Ind Co Ltd | Serial communication circuit |
JP3733699B2 (en) * | 1997-06-20 | 2006-01-11 | ソニー株式会社 | Serial interface circuit |
US5918074A (en) * | 1997-07-25 | 1999-06-29 | Neonet Llc | System architecture for and method of dual path data processing and management of packets and/or cells and the like |
US6108758A (en) * | 1997-08-29 | 2000-08-22 | Intel Corporation | Multiple masters in a memory control system |
US6067595A (en) * | 1997-09-23 | 2000-05-23 | Icore Technologies, Inc. | Method and apparatus for enabling high-performance intelligent I/O subsystems using multi-port memories |
KR100261218B1 (en) * | 1997-12-08 | 2000-07-01 | 윤종용 | Pin assignment method of semiconductor memory device & semiconductor memory device inputing packet signal |
US6622224B1 (en) * | 1997-12-29 | 2003-09-16 | Micron Technology, Inc. | Internal buffered bus for a drum |
US6085290A (en) * | 1998-03-10 | 2000-07-04 | Nexabit Networks, Llc | Method of and apparatus for validating data read out of a multi port internally cached dynamic random access memory (AMPIC DRAM) |
US6138219A (en) * | 1998-03-27 | 2000-10-24 | Nexabit Networks Llc | Method of and operating architectural enhancement for multi-port internally cached dynamic random access memory (AMPIC DRAM) systems, eliminating external control paths and random memory addressing, while providing zero bus contention for DRAM access |
US6003121A (en) * | 1998-05-18 | 1999-12-14 | Intel Corporation | Single and multiple channel memory detection and sizing |
US6112267A (en) * | 1998-05-28 | 2000-08-29 | Digital Equipment Corporation | Hierarchical ring buffers for buffering data between processor and I/O device permitting data writes by processor and data reads by I/O device simultaneously directed at different buffers at different levels |
US6122680A (en) * | 1998-06-18 | 2000-09-19 | Lsi Logic Corporation | Multiple channel data communication buffer with separate single port transmit and receive memories having a unique channel for each communication port and with fixed arbitration |
US6237130B1 (en) * | 1998-10-29 | 2001-05-22 | Nexabit Networks, Inc. | Chip layout for implementing arbitrated high speed switching access of pluralities of I/O data ports to internally cached DRAM banks and the like |
US6272567B1 (en) * | 1998-11-24 | 2001-08-07 | Nexabit Networks, Inc. | System for interposing a multi-port internally cached DRAM in a control path for temporarily storing multicast start of packet data until such can be passed |
US6389494B1 (en) | 1998-12-30 | 2002-05-14 | Emc Corporation | System for interfacing a data storage system to a host utilizing a plurality of busses for carrying end-user data and a separate bus for carrying interface state data |
US7073020B1 (en) | 1999-01-04 | 2006-07-04 | Emc Corporation | Method for message transfer in computer storage system |
US7117275B1 (en) | 1999-01-04 | 2006-10-03 | Emc Corporation | Data storage system having separate data transfer section and message network |
US6467018B1 (en) * | 1999-01-04 | 2002-10-15 | International Business Machines Corporation | Method and apparatus for addressing individual banks of DRAMs on a memory card |
US6345345B1 (en) * | 1999-01-26 | 2002-02-05 | Advanced Micro Devices, Inc. | Data communications device and associated method for arbitrating access using dynamically programmable arbitration scheme and limits on data transfers |
US6412032B1 (en) * | 1999-09-30 | 2002-06-25 | Rockwell Automation Technologies, Inc. | Interface for industrial controller network card |
DE19951046A1 (en) * | 1999-10-22 | 2001-04-26 | Siemens Ag | Memory component for a multi-processor computer system has a DRAM memory block connected via an internal bus to controllers with integral SRAM cache with 1 controller for each processor so that memory access is speeded |
US6628662B1 (en) | 1999-11-29 | 2003-09-30 | International Business Machines Corporation | Method and system for multilevel arbitration in a non-blocking crossbar switch |
US7010575B1 (en) | 2000-03-31 | 2006-03-07 | Emc Corporation | Data storage system having separate data transfer section and message network having bus arbitration |
US7007194B1 (en) | 2000-06-29 | 2006-02-28 | Emc Corporation | Data storage system having point-to-point configuration |
US6779071B1 (en) | 2000-04-28 | 2004-08-17 | Emc Corporation | Data storage system having separate data transfer section and message network with status register |
US6651130B1 (en) | 2000-04-28 | 2003-11-18 | Emc Corporation | Data storage system having separate data transfer section and message network with bus arbitration |
KR20010106079A (en) * | 2000-05-19 | 2001-11-29 | 강 크리스토퍼 | Pipelined and shared memory switch |
US6349058B1 (en) * | 2001-02-16 | 2002-02-19 | Microchip Technology Incorporated | Electronic circuit and method for storing configuration and calibration information in a non-volatile memory array |
US6919592B2 (en) | 2001-07-25 | 2005-07-19 | Nantero, Inc. | Electromechanical memory array using nanotube ribbons and method for making same |
US7566478B2 (en) | 2001-07-25 | 2009-07-28 | Nantero, Inc. | Methods of making carbon nanotube films, layers, fabrics, ribbons, elements and articles |
US6706402B2 (en) | 2001-07-25 | 2004-03-16 | Nantero, Inc. | Nanotube films and articles |
US7259410B2 (en) | 2001-07-25 | 2007-08-21 | Nantero, Inc. | Devices having horizontally-disposed nanofabric articles and methods of making the same |
US6643165B2 (en) | 2001-07-25 | 2003-11-04 | Nantero, Inc. | Electromechanical memory having cell selection circuitry constructed with nanotube technology |
US6924538B2 (en) | 2001-07-25 | 2005-08-02 | Nantero, Inc. | Devices having vertically-disposed nanofabric articles and methods of making the same |
US6574130B2 (en) | 2001-07-25 | 2003-06-03 | Nantero, Inc. | Hybrid circuit having nanotube electromechanical memory |
US6911682B2 (en) | 2001-12-28 | 2005-06-28 | Nantero, Inc. | Electromechanical three-trace junction devices |
US6835591B2 (en) | 2001-07-25 | 2004-12-28 | Nantero, Inc. | Methods of nanotube films and articles |
US6988161B2 (en) * | 2001-12-20 | 2006-01-17 | Intel Corporation | Multiple port allocation and configurations for different port operation modes on a host |
US7176505B2 (en) | 2001-12-28 | 2007-02-13 | Nantero, Inc. | Electromechanical three-trace junction devices |
US6784028B2 (en) | 2001-12-28 | 2004-08-31 | Nantero, Inc. | Methods of making electromechanical three-trace junction devices |
US7335395B2 (en) | 2002-04-23 | 2008-02-26 | Nantero, Inc. | Methods of using pre-formed nanotubes to make carbon nanotube films, layers, fabrics, ribbons, elements and articles |
DE10253918A1 (en) * | 2002-11-19 | 2004-06-17 | Infineon Technologies Ag | Storage system, in particular for network broadcasting applications such as video / audio applications, and method for operating a storage system |
US7099983B2 (en) * | 2002-11-25 | 2006-08-29 | Lsi Logic Corporation | Multi-core communications module, data communications system incorporating a multi-core communications module, and data communications process |
US7560136B2 (en) | 2003-01-13 | 2009-07-14 | Nantero, Inc. | Methods of using thin metal layers to make carbon nanotube films, layers, fabrics, ribbons, elements and articles |
US7571287B2 (en) * | 2003-03-13 | 2009-08-04 | Marvell World Trade Ltd. | Multiport memory architecture, devices and systems including the same, and methods of using the same |
US20040199727A1 (en) * | 2003-04-02 | 2004-10-07 | Narad Charles E. | Cache allocation |
JP2004355351A (en) * | 2003-05-29 | 2004-12-16 | Hitachi Ltd | Server device |
US7587521B2 (en) * | 2005-06-23 | 2009-09-08 | Intel Corporation | Mechanism for assembling memory access requests while speculatively returning data |
US8332598B2 (en) | 2005-06-23 | 2012-12-11 | Intel Corporation | Memory micro-tiling request reordering |
US7765366B2 (en) * | 2005-06-23 | 2010-07-27 | Intel Corporation | Memory micro-tiling |
US7558941B2 (en) * | 2005-06-30 | 2009-07-07 | Intel Corporation | Automatic detection of micro-tile enabled memory |
US8253751B2 (en) | 2005-06-30 | 2012-08-28 | Intel Corporation | Memory controller interface for micro-tiled memory access |
JP2007334564A (en) * | 2006-06-14 | 2007-12-27 | Matsushita Electric Ind Co Ltd | Unified memory system |
JP5018074B2 (en) * | 2006-12-22 | 2012-09-05 | 富士通セミコンダクター株式会社 | Memory device, memory controller and memory system |
KR101067081B1 (en) | 2007-01-12 | 2011-09-22 | 파나소닉 주식회사 | Plasma display device, and method for driving plasma display panel |
CN101216751B (en) * | 2008-01-21 | 2010-07-14 | 戴葵 | DRAM device with data handling capacity based on distributed memory structure |
JP5599969B2 (en) | 2008-03-19 | 2014-10-01 | ピーエスフォー ルクスコ エスエイアールエル | Multi-port memory and computer system including the multi-port memory |
JP5449686B2 (en) * | 2008-03-21 | 2014-03-19 | ピーエスフォー ルクスコ エスエイアールエル | Multiport memory and system using the multiport memory |
JP5588100B2 (en) * | 2008-06-23 | 2014-09-10 | ピーエスフォー ルクスコ エスエイアールエル | Semiconductor device and data processing system |
US8713248B2 (en) * | 2009-06-02 | 2014-04-29 | Nokia Corporation | Memory device and method for dynamic random access memory having serial interface and integral instruction buffer |
US9003206B2 (en) * | 2009-12-23 | 2015-04-07 | Bae Systems Information And Electronic Systems Integration Inc. | Managing communication and control of power components |
CN102193865B (en) * | 2010-03-16 | 2015-03-25 | 联想(北京)有限公司 | Storage system, storage method and terminal using same |
US8718806B2 (en) | 2011-09-02 | 2014-05-06 | Apple Inc. | Slave mode transmit with zero delay for audio interface |
US9514069B1 (en) | 2012-05-24 | 2016-12-06 | Schwegman, Lundberg & Woessner, P.A. | Enhanced computer processor and memory management architecture |
EP3008946B1 (en) | 2013-06-11 | 2018-08-08 | Seven Networks, LLC | Offloading application traffic to a shared communication channel for signal optimization in a wireless network for traffic utilizing proprietary and non-proprietary protocols |
CN104717152B (en) * | 2013-12-17 | 2019-07-19 | 深圳市中兴微电子技术有限公司 | A kind of method and apparatus realizing interface caching and dynamically distributing |
CN106293635B (en) * | 2015-05-13 | 2018-10-30 | 华为技术有限公司 | Instruction block processing method and processing device |
US9965211B2 (en) | 2016-09-08 | 2018-05-08 | Cisco Technology, Inc. | Dynamic packet buffers with consolidation of low utilized memory banks |
KR20180092476A (en) * | 2017-02-09 | 2018-08-20 | 에스케이하이닉스 주식회사 | Storage device and operating method thereof |
CN109582226A (en) * | 2018-11-14 | 2019-04-05 | 北京中电华大电子设计有限责任公司 | A kind of high speed storing access logical construction and its control method |
TW202141290A (en) | 2020-01-07 | 2021-11-01 | 韓商愛思開海力士有限公司 | Processing-in-memory (pim) system and operating methods of the pim system |
US11385837B2 (en) | 2020-01-07 | 2022-07-12 | SK Hynix Inc. | Memory system |
US11315611B2 (en) * | 2020-01-07 | 2022-04-26 | SK Hynix Inc. | Processing-in-memory (PIM) system and operating methods of the PIM system |
CN115065572B (en) * | 2022-02-28 | 2023-09-29 | 西安电子科技大学 | CAN FD controller for vehicle-mounted electronic system |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0340668A2 (en) * | 1988-05-06 | 1989-11-08 | Hitachi, Ltd. | Multi-processor system having a multi-port cache memory |
EP0471932A2 (en) * | 1990-07-27 | 1992-02-26 | International Business Machines Corporation | Virtual multi-port ram |
US5450355A (en) * | 1993-02-05 | 1995-09-12 | Micron Semiconductor, Inc. | Multi-port memory device |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5581773A (en) * | 1992-05-12 | 1996-12-03 | Glover; Michael A. | Massively parallel SIMD processor which selectively transfers individual contiguously disposed serial memory elements |
US5490112A (en) * | 1993-02-05 | 1996-02-06 | Micron Technology, Inc. | Multi-port memory device with multiple sets of columns |
JPH06251166A (en) * | 1993-02-25 | 1994-09-09 | Toshiba Corp | Image processing device |
US5442747A (en) * | 1993-09-27 | 1995-08-15 | Auravision Corporation | Flexible multiport multiformat burst buffer |
US5457654A (en) * | 1994-07-26 | 1995-10-10 | Micron Technology, Inc. | Memory circuit for pre-loading a serial pipeline |
-
1995
- 1995-12-29 US US08/581,467 patent/US5799209A/en not_active Expired - Lifetime
-
1996
- 1996-08-12 IL IL12513596A patent/IL125135A/en not_active IP Right Cessation
- 1996-08-12 DK DK96925929T patent/DK0870303T3/en active
- 1996-08-12 KR KR1019980705020A patent/KR100328603B1/en not_active IP Right Cessation
- 1996-08-12 EP EP96925929A patent/EP0870303B1/en not_active Expired - Lifetime
- 1996-08-12 AU AU65295/96A patent/AU721764B2/en not_active Ceased
- 1996-08-12 AT AT96925929T patent/ATE197101T1/en not_active IP Right Cessation
- 1996-08-12 CN CN96180069A patent/CN1120495C/en not_active Expired - Fee Related
- 1996-08-12 JP JP50793497A patent/JP3699126B2/en not_active Expired - Fee Related
- 1996-08-12 DE DE69610714T patent/DE69610714T2/en not_active Expired - Lifetime
- 1996-08-12 CA CA002241841A patent/CA2241841C/en not_active Expired - Fee Related
- 1996-08-12 WO PCT/IB1996/000794 patent/WO1997024725A1/en active IP Right Grant
- 1996-09-14 TW TW085111239A patent/TW318222B/zh not_active IP Right Cessation
-
1998
- 1998-07-06 US US09/110,929 patent/US6108725A/en not_active Expired - Lifetime
-
1999
- 1999-07-27 HK HK99103219A patent/HK1018342A1/en not_active IP Right Cessation
-
2001
- 2001-01-17 GR GR20010400081T patent/GR3035261T3/en not_active IP Right Cessation
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP0340668A2 (en) * | 1988-05-06 | 1989-11-08 | Hitachi, Ltd. | Multi-processor system having a multi-port cache memory |
EP0471932A2 (en) * | 1990-07-27 | 1992-02-26 | International Business Machines Corporation | Virtual multi-port ram |
US5450355A (en) * | 1993-02-05 | 1995-09-12 | Micron Semiconductor, Inc. | Multi-port memory device |
Non-Patent Citations (2)
Title |
---|
C. HAVELL: "Dram or vram: paying for performance", NEW ELECTRONICS INCORPORATING ELECTRONICS TODAY., vol. 27, no. 4, April 1994 (1994-04-01), LONDON GB, pages 29 - 30, XP000447249 * |
J. MAILLOUX: "Multiport DRAM trends", IRE WESCON CONFERENCE RECORD, vol. 34, November 1990 (1990-11-01), NORTH HOLLYWOOD US, pages 18 - 23, XP000227841 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
AU760297B2 (en) * | 1997-04-30 | 2003-05-08 | Canon Kabushiki Kaisha | Memory controller architecture |
WO1999005604A1 (en) * | 1997-07-28 | 1999-02-04 | Nexabit Networks, Llc | Multi-port internally cached drams |
JP2001511559A (en) * | 1997-07-28 | 2001-08-14 | ネグザビット・ネットワークス,リミテッド・ライアビリティー・カンパニー | Multi-port internal cache DRAM |
AU748133B2 (en) * | 1997-07-28 | 2002-05-30 | Nexabit Networks, Llc | Multi-port internally cached drams |
FR2779843A1 (en) * | 1998-06-16 | 1999-12-17 | Busless Computers | Serial multi port memory component comprising RAM memory bank assemblies for use in computer |
WO2000030421A1 (en) * | 1998-11-12 | 2000-05-25 | Nexabit Networks, Inc. | Electronic circuit board |
WO2000058840A1 (en) * | 1999-03-26 | 2000-10-05 | Nexabit Networks, Inc. | Ampic dram system |
CN1573723B (en) * | 2003-05-15 | 2010-05-05 | 三星电子株式会社 | Method and apparatus for communication via serial multi-port |
CN100390755C (en) * | 2003-10-14 | 2008-05-28 | 中国科学院计算技术研究所 | Computer micro system structure comprising explicit high-speed buffer storage |
WO2011094218A2 (en) * | 2010-01-29 | 2011-08-04 | Mosys, Inc. | Hierarchical multi-bank multi-port memory organization |
WO2011094218A3 (en) * | 2010-01-29 | 2011-11-24 | Mosys, Inc. | Hierarchical multi-bank multi-port memory organization |
US8547774B2 (en) | 2010-01-29 | 2013-10-01 | Mosys, Inc. | Hierarchical multi-bank multi-port memory organization |
US9030894B2 (en) | 2010-01-29 | 2015-05-12 | Mosys, Inc. | Hierarchical multi-bank multi-port memory organization |
Also Published As
Publication number | Publication date |
---|---|
US6108725A (en) | 2000-08-22 |
KR19990076893A (en) | 1999-10-25 |
DE69610714T2 (en) | 2001-05-10 |
TW318222B (en) | 1997-10-21 |
US5799209A (en) | 1998-08-25 |
JP3699126B2 (en) | 2005-09-28 |
GR3035261T3 (en) | 2001-04-30 |
CN1120495C (en) | 2003-09-03 |
EP0870303B1 (en) | 2000-10-18 |
EP0870303A1 (en) | 1998-10-14 |
CA2241841C (en) | 1999-10-26 |
ATE197101T1 (en) | 2000-11-15 |
IL125135A (en) | 2002-12-01 |
KR100328603B1 (en) | 2002-10-19 |
AU6529596A (en) | 1997-07-28 |
IL125135A0 (en) | 1999-01-26 |
JP2000501524A (en) | 2000-02-08 |
CA2241841A1 (en) | 1997-07-10 |
HK1018342A1 (en) | 1999-12-17 |
AU721764B2 (en) | 2000-07-13 |
DE69610714D1 (en) | 2000-11-23 |
DK0870303T3 (en) | 2001-02-26 |
CN1209213A (en) | 1999-02-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
EP0870303B1 (en) | High performance universal multi-port internally cached dynamic random access memory system, architecture and method | |
WO1997024725A9 (en) | High performance universal multi-port internally cached dynamic random access memory system, architecture and method | |
KR100272072B1 (en) | High performance, high bandwidth memory bus architecture utilizing sdrams | |
US5537353A (en) | Low pin count-wide memory devices and systems and methods using the same | |
KR100494201B1 (en) | Memory Systems, I / O Subsystem Devices, and How to Operate Memory Devices | |
US5182801A (en) | Apparatus and method for providing fast data transfer between multiple devices through dynamic reconfiguration of the memory space of the devices | |
US6212597B1 (en) | Apparatus for and method of architecturally enhancing the performance of a multi-port internally cached (AMPIC) DRAM array and like | |
JP3383846B2 (en) | SAM with expandable data width for multiport RAM | |
US20020124149A1 (en) | Efficient optimization algorithm in memory utilization for network applications | |
US6920510B2 (en) | Time sharing a single port memory among a plurality of ports | |
US6138219A (en) | Method of and operating architectural enhancement for multi-port internally cached dynamic random access memory (AMPIC DRAM) systems, eliminating external control paths and random memory addressing, while providing zero bus contention for DRAM access | |
US5717904A (en) | Apparatus and methods for automatically controlling block writes | |
US5898623A (en) | Input port switching protocol for a random access memory | |
US5167029A (en) | Data processing system and associated process using memory cards having data modify functions utilizing a data mask and an internal register | |
KR100264301B1 (en) | Use of a processor bus for the transmission of i/o traffic | |
CN114442908B (en) | Hardware acceleration system and chip for data processing | |
JPH10143466A (en) | Bus communication system | |
US5969997A (en) | Narrow data width DRAM with low latency page-hit operations | |
US4618946A (en) | Dual page memory system having storage elements which are selectively swapped between the pages | |
US7324546B1 (en) | Network router integrated onto a silicon chip | |
EP0745242B1 (en) | Variable data processor allocation and memory sharing | |
JP3651672B2 (en) | Arithmetic processing system, especially arithmetic processing system for communication devices | |
KR100599504B1 (en) | A method of controlling single port memory for display unit | |
JP2000148697A (en) | Computer system | |
JPH05143561A (en) | Compound mode multiprocessing system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 96180069.0 Country of ref document: CN |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): AL AM AT AT AU AZ BB BG BR BY CA CH CN CU CZ CZ DE DE DK DK EE EE ES FI FI GB GE HU IL IS JP KE KG KP KR KZ LK LR LS LT LU LV MD MG MK MN MW MX NO NZ PL PT RO RU SD SE SG SI SK SK TJ TM TR TT UA UG US UZ VN AM AZ BY KG KZ MD RU TJ TM |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): KE LS MW SD SZ UG AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT |
|
DFPE | Request for preliminary examination filed prior to expiration of 19th month from priority date (pct application filed before 20040101) | ||
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
COP | Corrected version of pamphlet |
Free format text: PAGE 9/27,DRAWINGS,REPLACED BY A NEW PAGE BEARING THE SAME NUMBER;AFTER RECTIFICATION OF OBVIOUS ERRORS AS AUTHORIZED BY THE INTERNATIONAL SEARCHING AUTHORITY |
|
ENP | Entry into the national phase |
Ref document number: 2241841 Country of ref document: CA Ref document number: 2241841 Country of ref document: CA Kind code of ref document: A Ref document number: 1997 507934 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1019980705020 Country of ref document: KR Ref document number: PA/A/1998/005336 Country of ref document: MX |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1996925929 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1996925929 Country of ref document: EP |
|
REG | Reference to national code |
Ref country code: DE Ref legal event code: 8642 |
|
WWP | Wipo information: published in national office |
Ref document number: 1019980705020 Country of ref document: KR |
|
WWG | Wipo information: grant in national office |
Ref document number: 1996925929 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1019980705020 Country of ref document: KR |