WO1997035317A1 - Processeur a memoire dram integree - Google Patents
Processeur a memoire dram integree Download PDFInfo
- Publication number
- WO1997035317A1 WO1997035317A1 PCT/JP1996/000731 JP9600731W WO9735317A1 WO 1997035317 A1 WO1997035317 A1 WO 1997035317A1 JP 9600731 W JP9600731 W JP 9600731W WO 9735317 A1 WO9735317 A1 WO 9735317A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- dynamic ram
- memory
- data
- semiconductor integrated
- dynamic
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C11/00—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor
- G11C11/21—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements
- G11C11/34—Digital stores characterised by the use of particular electric or magnetic storage elements; Storage elements therefor using electric elements using semiconductor devices
-
- G—PHYSICS
- G11—INFORMATION STORAGE
- G11C—STATIC STORES
- G11C5/00—Details of stores covered by group G11C11/00
- G11C5/02—Disposition of storage elements, e.g. in the form of a matrix array
- G11C5/025—Geometric lay-out considerations of storage- and peripheral-blocks in a semiconductor storage device
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/76—Architectures of general purpose stored program computers
- G06F15/78—Architectures of general purpose stored program computers comprising a single central processing unit
- G06F15/7839—Architectures of general purpose stored program computers comprising a single central processing unit with memory
- G06F15/7842—Architectures of general purpose stored program computers comprising a single central processing unit with memory on one IC chip (single chip microcontrollers)
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/76—Architectures of general purpose stored program computers
- G06F15/78—Architectures of general purpose stored program computers comprising a single central processing unit
- G06F15/7839—Architectures of general purpose stored program computers comprising a single central processing unit with memory
- G06F15/7842—Architectures of general purpose stored program computers comprising a single central processing unit with memory on one IC chip (single chip microcontrollers)
- G06F15/7857—Architectures of general purpose stored program computers comprising a single central processing unit with memory on one IC chip (single chip microcontrollers) using interleaved memory
-
- 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
- G11C7/1072—Input/output [I/O] data interface arrangements, e.g. I/O data control circuits, I/O data buffers for memories with random access ports synchronised on clock signal pulse trains, e.g. synchronous memories, self timed memories
Definitions
- the present invention relates to a semiconductor integrated circuit device, and more particularly, to a semiconductor integrated circuit device including a data processing device for performing image processing and a memory device for storing image data or instructions.
- the method (1) is implemented using a DRAM with a high-speed page mode or a synchronous DRAM.
- Japanese Patent Application Laid-Open No. 7-160249 discloses a device using a synchronous DRAM.
- the frame buffer and the graphic interface are built in the one-chip, and the bit width of the internal bus is made 128 bits or the like. ing.
- DRAM and graphics controller Examples of embedded LSIs include the “Development of Graphics LSI with Built-in Frame Buffer” on page 17 of the April 10, 1995 issue of Nikkei Electronics and Nikkei Microdevices. "Logic and One-Chip Integration—DRAM at the Core of the System", March 1976, pages 44-65.
- the graphics LSI with a built-in frame buffer described in the Nikkei Electronics which removes 9 Mbits of a 16 Mbit general-purpose standard DRAM and incorporates a logic circuit such as a controller Things.
- a logic circuit such as a controller Things.
- the graphic controller with a built-in DRAM described in the Nikkei Microphone Device except that the DRAM is built-in.
- An object of the present invention is to realize an optimum layout of a semiconductor integrated circuit device having a built-in circuit and an image processor.
- Another object of the present invention is to enable a conventional test method to be used as it is for testing a memory of a semiconductor integrated circuit device incorporating logic and memory.
- Still another object of the present invention is to increase the depth of the memory address and realize a built-in image memory having a large capacity as viewed from the image processor.
- Another object of the present invention is to facilitate the control logic of a logic state machine of a semiconductor integrated circuit device incorporating logic and a memory.
- a semiconductor integrated circuit device having a built-in image memory and image processor is arranged in accordance with the flow of information.
- test bus for built-in memory is provided in the semiconductor integrated circuit device to output to the outside.
- a normal port and a test port are provided in the built-in memory.
- each of the image memories incorporated in the semiconductor integrated circuit device is constituted by a plurality of identical memory modules, and the same row address is allocated to each memory module. Further, when logic built in the semiconductor integrated circuit device accesses memory, the latency of read and write operations of the memory is made equal.
- FIG. 1 shows an example of a system using a semiconductor integrated circuit device according to the present invention.
- Fig. 2 shows a typical image manipulation.
- FIG. 3 is a block diagram of a side operation unit of an image processor built in the semiconductor integrated circuit device according to the present invention.
- FIG. 4 is a block diagram of a straight-line operation unit of an image processor built in the semiconductor integrated circuit device according to the present invention.
- FIG. 5 is a block diagram of a pixel operation unit of an image processor incorporated in the semiconductor integrated circuit device according to the present invention.
- FIG. 6 shows a connection relationship between an image processor and an image memory incorporated in the semiconductor integrated circuit device according to the present invention.
- FIG. 7 shows a basic timing diagram of reading and writing of a memory module built in the semiconductor integrated circuit device according to the present invention
- FIG. 8 shows a semiconductor integrated circuit device according to the present invention. A timing diagram is shown when the row address of the memory module built in is switched.
- FIG. 9 shows a case in which drawing over a plurality of banks occurs.
- FIG. 10 shows the state of the four-stage pipeline processing of the image processor built in the semiconductor integrated circuit device according to the present invention.
- FIG. 11 shows a specific example of a memory module included in a semiconductor integrated circuit device according to the present invention.
- FIG. 12 shows a schematic configuration of a layer image of the semiconductor integrated circuit device according to the present invention.
- FIG. 13 shows an example of a layout of a memory module stored in a semiconductor integrated circuit device according to the present invention.
- FIG. 14 shows another example of a layout of a memory module incorporated in the semiconductor integrated circuit device according to the present invention.
- FIG. 15 shows a test mechanism of the semiconductor integrated circuit device according to the present invention.
- FIG. 16 shows a test function of a memory module built in the semiconductor integrated circuit device according to the present invention.
- FIG. 17 shows an example of a memory module switching circuit built in the semiconductor integrated circuit device according to the present invention.
- FIG. 18 shows the assignment of test control pins of the semiconductor integrated circuit device according to the present invention.
- FIG. 19 shows test terminal inputs and outputs during a logic test of the semiconductor integrated circuit device according to the present invention.
- FIG. 20 is an overall block diagram of the semiconductor integrated circuit device according to the present invention.
- FIG. 21 to FIG. 23 show input / output pins of the semiconductor integrated circuit device according to the present invention.
- FIG. 1 shows an example of a system using a semiconductor integrated circuit device SIC according to an embodiment of the present invention.
- the system shown in Fig. 1 uses a data processing system such as a personal computer or an amusement device. Form part of the system.
- the semiconductor integrated circuit device SIC is composed of an image processor GP, a command.
- Source data image memory hereinafter referred to as command memory
- VRAM a VRAM
- drawing memory a drawing / display memory
- the semiconductor integrated circuit device SIC is connected to a central processing unit CPU and a CRT control circuit DP.
- the central processing unit CPU accesses the image processor GP through the bus control circuit BC1.
- the output from the bus control circuit BC 1 passes through the CPU interface CIU, and the bus BUS 1 for accessing the drawing command DCF DCF and the bus for accessing the command and memory VRAM Divided into BUS2.
- the drawing command switch DCF is accessed from the CPU interface unit CIU, the command to be processed and the input data are read from the command memory VRAM, the side processing unit EDGE, and the straight line This is given to the drawing control unit DM that performs image processing calculations such as the calculation unit LINE and the pixel calculation unit DOT.
- the drawing command fetch section DCF issues an execution start command, fetches the command from the memory VRAM, and inputs necessary parameters to the side operation section EDGE, the linear operation section LINE, and the pixel operation section. Transfer to the DOT and start the side processing unit EDGE.
- the side processing unit EDGE calculates the coordinates stored in the input data and the drawing coordinates for each end point, and activates the line calculation unit LINE.
- the line calculation unit LINE calculates the coordinates of the input data and the drawing coordinates in units of one dot, and issues an instruction to the pixel calculation unit DOT that processes the data.
- the operation unit DOT fetches input data from the command / memory VRAM and after processing, draws it to either the drawing memory FB0 or the drawing memory FB I via the bus control unit BC3 and the switch SW. I do.
- the starting point for drawing is determined by the state after reset.
- the memory that is not drawing is read out by the display control unit DISP via the switch switch SW with the bus control unit BC4 and displayed.
- Data is transferred to the display processor DP via the output bus BUS3.
- the display processor DP converts the display data into a video signal and sends it to the display device CRT.
- the image memory FB0 and the image memory FBI are each composed of 2 Mbits of DRAM.
- FIG. 2 ( a ) shows the function of mapping a rectangular source image ABCD onto an arbitrary square A'B'CD '.
- the image processor GP uses a method that implements this mapping by executing the line copy multiple times.
- the line copy refers to the horizontal pixel row ⁇ ( ⁇ , ⁇ ) force, P l (Xpl, Ypl) of the source image as shown in FIG. 2 (b).
- An image operation in which an arbitrary straight line Q0 (Xq0, Yq0) in the function space is mapped to Ql (Xql, Yql).
- the image processor GP performs a side operation to find the start point Q0 and end point Q1 of the line copy, and a straight line operation to find a straight line connecting Q0 and Q1.
- the image processor GP can execute the modified splice processing in a maximum of 29 Mpixels by a macro command from an external data processing device.
- FIG. 3 shows a detailed block diagram of the side processing unit EDGE.
- the EDGE is composed of two 13-bit arithmetic units (Arithmetic Uni AUa and AUb, which have dedicated read and write buses, and a 13-bit register common to the two arithmetic units AUa and AUb.
- R l -Rn Evening (R l -Rn), 13-bit registers (Ra l -Ran, Rbl -Rbn) dedicated to each arithmetic unit AUa and AUb, and registers (R1-Rn, Ra l -Ran Rbl -Rbn) Address decoder 121 for selecting the operation, and a side operation unit sequencer 122 for controlling the arithmetic units AUa, AUb, etc. Is done.
- the side processing unit EDGE is a module that executes a side drawing algorithm.
- the side processing unit EDGE fetches a drawing command, drawing source data, and drawing parameters from the command memory VRAM.
- the command and parameters that have been flushed are stored in the internal registers in the side operation unit EGDE and pixel operation unit DOT.
- the side operation unit EGDE executes the side operation according to the rendered drawing command and the drawing parameter, and stores the side operation result in an internal register in the straight line operation unit LINE.
- FIG. 4 shows a detailed block diagram of the straight line calculation unit LINE.
- the linear computation unit LINE performs five DDA computations (S-DDA, D-DDA, R-DDA, G-DDA, and G-DDA) that perform DDA operation (operation that mainly performs subtraction: Digital Differential Analyzer) in one cycle.
- B-DDA DDA computations
- a 13-bit register group 132 for selecting the register group 132.
- the line operation unit LINE is a module that executes a line drawing algorithm.
- Side calculation unit Executes a straight line calculation according to the side calculation result stored by the EGDE.
- the line operation unit LINE stores the parameters of the start point and the end point of the line copy passed from the side operation unit EDGE by the built-in register group 132, and performs the line operation based on these parameters.
- FIG. 5 shows a detailed block diagram of the pixel operation unit DOT.
- the pixel operation unit DOT consists of a source 'memory'address' counter S-Counter, a destination 'memory' address counter D-Counter, and three units corresponding to red, green and blue. 5 bit counters R-Counter, G-Counter, B-Counter and three dedicated read / write buses It consists of R-AU, G-AU, B-AU and so on.
- Source memory address counter S-Counter and destination memory 'address' counter D-Counter counts addresses when carry occurs as a result of arithmetic. Perform a top-up.
- the three 5-bit counters R-Counter, G-Counter, and B-Counter count up the color data when a carry occurs as a result of the operation.
- the three 5-bit arithmetic units R-AU, G-AU, and B-AU are red, green, and blue, respectively, and red generated by the 5-bit counters R-Counter, G-Counter, and B-Counter. , Green and Blue are added.
- the pixel operation unit DOT is a module that executes a pixel copy algorithm.
- the address calculation for the drawing memory and the pixel calculation of the data are performed according to the result of the straight line calculation.
- the pixel operation is an operation for obtaining color data (R, G, B) of a source coordinate P, a destination coordinate Q, and a destination coordinate Q of a pixel on a line copy. Determined by incrementing the value.
- the display control unit DISP reads display data from the drawing memories FB0 and FBI and sends the read display data to the display processor DP.
- the display control unit DISP includes a refresh circuit that refreshes the command memory VRAM and the drawing memories FB0 and FB1.
- the refresh circuit simultaneously refreshes the command memory VRAM and the drawing memories FB0 and FBI, and the refresh cycle is performed based on the command memory VRAM. .
- the refresh circuit is provided with a refresh There is a register for the cycle.
- the refresh cycle is determined by writing to this register by the CPU according to the DRAM specifications.
- the command memory VRAM Since the number of refresh cycles and the number of clocks with the drawing memory FB0 and FBI are known in advance, they can be fixed.
- the display control unit DISP inputs the clock matched to the command / memory VRAM to the command's memory VRAM and the drawing memories FB0 and FB1, thereby mounting multiple DRAMs.
- the refresh cycle of the image processing equipment to be used is unified. Since the display control unit DISP knows the retrace period of the display device CRT, the DRAM is refreshed using the retrace period.
- the drawing memories FB0 and FB1 that use a 2-Mbit DRAM are refreshed twice. It will be reshuffled.
- Fig. 6 shows the connection relationship between the image processor GP, the command memory VRAM, and the drawing memories FB0 and FBI.
- a 4-Mbit DRAM of VRAM is configured using two 2-Mbit DRAM modules in an 8-bank configuration.
- the 2-Mbit DRAMs of the drawing memory FB0 and the drawing memory FB1 are each configured using two 1-Mbit DRAM modules of a 4-bank configuration.
- a DRAM module is also referred to as a memory module.
- the command memory VRAM and drawing memories FB0 and FBI have a memory array composed of 256 word lines and 124 bit line pairs.
- the memory module can be configured in 256-bit units by increasing or decreasing the number of banks. This is a memory module suitable for a semiconductor integrated circuit in which logic and memory are mixed as in the present embodiment.
- a memory module is a so-called synchronous DRAM in which addresses and control signals are input in synchronization with a clock signal, and data is also input and output in synchronization with the clock signal. Therefore, the memory module operates according to the so-called command specified by the control signal and the address signal (the row address and the column address are not multiplexed as in the general-purpose standard DRAM).
- a 16-bit data bus DBUS 16 an 11-bit address bus (A0-A10), and an 8-bit mouth address (between the image processor GP and the command memory VRAM) R0-R7), 8-bit column address control (C0-C7), row address control CR, column address control CC0, CC1, 16-bit byte enable BE, RE Signals such as one drive RW, active control AC, and clock CK are connected.
- a 32-bit data bus DBUS32 and an 11-bit address bus are provided between the image processor GP and the drawing memories FBO and FB1.
- A0-A10 4-bit bank address (R0-R3), row address control CR, column address control CC0, CC1, 16-bit byte enable BE, re-write RW, active control Signals such as troll AC and clock CK are connected.
- FIG. 7 shows the basic timing of reading and writing of the memory module.
- Command / memory represents the basic timing of a series of operations from reading out source data from VRAM, converting the image with the image processor GP, and writing this to the drawing memories FB0 and FBI. is there.
- the command 'memory VRAM address ADDRVRAM, the drawing memory FB0, and the address of the FB I ADDRFB are generated by the image processor GP, and the command' memory VRAM and the drawing memory FB0, FB I are respectively generated. Entered.
- the control signals required for the memory module are also generated by the image processor GP and input to the command memory VRAM and the drawing memories FB0 and FBI.
- the active control AC, the lower dress control CR, the lower address AX and the force are taken into the memory module at the falling edge of the clock CK, and the bank is activated (T0) c
- the column control CC, the read drive RW, and the power supply address AYi are loaded into the memory module at the falling edge of the clock CK (T2). Data is read out two clocks later ( ⁇ 4).
- the source address (READ 1) is read four clocks after the low address AX is taken into the command memory VRAM, and then the low address is taken into the drawing memory FB in the same manner as in the case of c .
- the pixel data (READ2) is read.
- the command '' read from memory VRAM The read source data (READ l) and the pixel data (READ2) read from the drawing memories FB0 and FBI are latched by the bus control unit BC2 (SET0), and the combined data (SET1) is written by the pixel operation unit DOT. ) Is generated.
- the image processor GP outputs addresses and control signals to write the composite data (SET1) to the drawing memories FB0 and FBI. Then, the column content CC, the read write RW, and the column address AYi are taken into the memory module at the falling edge of the clock CK (T7). Two clocks later, data (WRITE 1) is written (T9). This writes the composite data (SET 1) to the drawing memory FB.
- the read latency of the memory module (the time from when a read command is input until data can be read) is two clocks, and the write latency (when a write command is input).
- the time from when the data is written until the data is written) is one clock. Therefore, in the case of writing, the image processor GP inserts a NOP for one cycle to match the writing and reading cycles. This allows read and write processing in the state machine to be handled in the same way, with read-write, write-read, read, read, write- There is no need to consider the combination of accesses called light on the state machine. In addition, this makes it possible to reduce the number of logic gates of the image processor.
- row address AX switching is detected in the first stage (B0: X-Y), and row address (B0: AX0) is issued (T0).
- NOP is executed to secure the precharge time (T1, T2).
- the column address ($ 0: $ 3) is issued in the fourth stage ($ 3).
- the first stage detects row address # switching ($ 2: $-$) and issues a row address ($ 2: $ 1) (Tl).
- NOP is performed to secure the precharge time (T2, T3).
- the fourth column issues column addresses ($ 2: $ 4) ($ 4).
- the first stage detects the switching of row address ⁇ ( ⁇ 3: ⁇ - ⁇ ) and issues the row address ( ⁇ 3: ⁇ 3) ( ⁇ 2).
- NOP is executed to secure the clearing time (T3, # 4).
- a column address ( ⁇ 3:) 5) is issued ( ⁇ 5).
- the bank column address AY can be issued continuously. As a result, in normal use conditions, the performance is increased by the absence of the wait due to the miss cycle.
- the detection of the switching of the row address AX can be realized by comparing the row address AX of the previous cycle with the row address AX of the current cycle in the bus control units BC2, BC3, and BC4. .
- the number of bits that can be made active by issuing a row address once is 124 bits.
- a read command or a write command can be issued immediately.
- a read command or a right command cannot be issued immediately to secure the time for the clearing.
- the image processor GP uses three clocks in the case of a missit. Krka has activated two banks of two memory modules. In other words, multiple banks are activated at the same time, reducing overhead when switching banks.
- the command' memory VRAM uses a 1M memory module and the image memory For FB0 and FBI, use a 512K memory module. In this case, it becomes possible to make four hundred and ninety-six bits active four times in one row address access.
- the memory module of the present embodiment can continuously execute read or write processing by outputting only the column address AY when the row address AX is in a hit state. .
- the lower address AX misses, it issues an address after precharge, so it is necessary to wait several cycles for the command to be issued. Therefore, if the source data is continuously read without being missed and a miss occurs when writing data to the destination, the data overflows and is lost. Therefore, in the present embodiment, it is detected in advance that a miss occurs at the time of writing, and even if the reading on the source data side is not a miss, the miss operation is caused to occur. , Waiting for data. Conversely, if the source data side is missed and the destination side is written, the miss process is executed.
- FIG. 11 shows a specific configuration of the memory module in the present embodiment.
- the memory module consists of three types of modules: a knock module BANK, an amplifier module AMP, and a power supply module PS.
- the NK module BANK includes BANK-0 BANK-n, and includes a plurality of sub-memory cell arrays SUBARY (SUBARY-00 SUBARY-i7), a puncture control circuit BNKCNT-1, and a bank control circuit BNKCNT-2.
- the sub-memory cell array SUBARY includes a plurality of pairs of bit lines B / B, a plurality of read lines W, a plurality of memory cells (indicated by circles in the figure), and a bit before reading the memory cells.
- a bit line precharge circuit PS for setting the potential of the bit line in advance to a predetermined level, a sense amplifier SA for amplifying a signal from a memory cell, and one of a plurality of pairs of bit lines B / B are selected. It consists of a Y selection circuit and global bit lines GBL / GBL that connect the selected bit lines B / B to the amplifier module AMP.
- the sub memory cell array SUBARY is a unit for dividing the I / O lines in the bank module BANK.
- the link control circuit BNKCNT-1 includes an X decoder XD for selecting the word line W and a Y decoder YD for selecting the bit line BZB.
- the selected bit line B / B transmits / receives data to / from the amplifier module AMP via the global bit line GBL / GBL arranged in parallel with the bit line B / B.
- the bank control circuit BNKCNT-2 has a sense amplifier control signal Includes a group of sensors that detect when the bell has been reached.
- the amplifier module AMP includes a main control circuit MAINCNT that supplies control signals and address signals to the bank module BANK in synchronization with the clock signal, and data to the bank module group (BANK-0 to BANK-n). It consists of a byte control circuit BYTCNT that controls the reading and writing of data.
- the data input / output lines DQ (DQ00, .., DQ07, ... DQ07..., DQi7) from outside the memory module are input to the memory cell through here.
- the byte control signal BEi is a signal that opens and closes the data input / output line DQ in byte units.
- the power supply module PS is supplied to the link module BANK. It is necessary for the VCH generation circuit VCHG, which generates the required lead line voltage VCH> supply voltage VCC) required for the lead line drive circuit WD, and the bit line precharge.
- Line precharge voltage generating circuit HVCG that generates a high voltage HVC (power supply voltage VCC Z 2), substrate voltage in the array (back bias voltage) VBB power supply voltage VSS in the array that generates the VSS (ground potential) Generation circuit This module generates various voltages such as VBB G.
- the bank module BANK of this embodiment has 256 word lines and 1 line, (8 X 8 Xi pairs of bit lines intersect, 1 Y 8 is selected by the Y decoder, and (8 X i pairs of global bit lines are input / output
- i 16
- one link module BANK has a capacity of 256 K bits and a width of 128 bits.
- data can be input / output by means of a memory module module with a variable capacity in units of 256 Kbits, and the link module BANK-n is shown in Fig. 6. Corresponds to one bank of multiple banks (B0 to B7).
- FIG. 1 Schematic structure of a layout image of a semiconductor integrated circuit SIC according to the present invention The result is shown in FIG.
- the semiconductor integrated circuit SIC has a horizontally long shape, with a command 'memory VRAM on the left side, drawing memories FB0 and FBI on the right side, and an image processor PS arranged therebetween.
- Memory VRAM has two 2-Mbit memory modules arranged in a mirror-like manner so that address buses, data buses, control signals, etc. can be input and output between the two memory modules.
- the drawing memories FB0 and FBI each have two 1-Mbit memory modules arranged in a mirror-like manner, so that address buses, data buses, control signals, etc. can be input and output between the two memory modules. ing.
- the bus width between the image processor GP and the memory module is relatively narrow, such as 16 bits or 32 bits. Since the memory module has a maximum width of 128 bits, the bus width between the image processor GP and the memory module can be increased to 128 bits. In this case, as shown in FIG. 14, changing the arrangement of the memory module makes it easier to interface data input / output.
- Command memory VRAM and drawing memories FB0 and FBI have the same storage capacity and differ in the configuration of the memory module.However, the power supply module PS and amplifier module AMP are smaller than the bank module BANK. , Almost the same shape and the same area.
- FIG. 15 is a block diagram showing a test mechanism inside the semiconductor integrated circuit device SIC in this embodiment.
- the semiconductor integrated circuit device SIC has a normal bus NB connected to the image processor GP for normal operation, a normal terminal NT connected to the normal bus NB, an image processor GP, a command memory VRAM, and a drawing.
- a common test bus TB used for test operation connected to the memories FB0 and FB I, a test terminal TT connected to the common test bus TB, a normal mode, a test mode, etc.
- a mode selection terminal MST for controlling the mode is provided.
- the internal control signals TEM0 to TEM5 are the selection signals of the memory module to be tested output from the mode selection terminal MST.
- the internal buses IB0, IB1, and IB2 are internal buses for normal operation that are not connected to the outside.
- the test of the memory module including the command / memory VRAM and the drawing memories FB0 and FBI and the test of the drawing processor GP are performed in independent formats.
- the test of the memory module is performed by the memory test, and the test of the drawing processor GP is performed by the logic tester.
- the memory module in the present embodiment includes a normal port NP used during a normal operation and a test port TP used during a test operation.
- This is usually a control port such as memory control on the port NP side. This is because the port is connected via internal buses IB0, IB1, and IB2, minimizing the port load during normal operation.
- the normal port and the test port do not necessarily have to be separate, and can be combined into a single port by using a multiplex or other configuration.
- the internal control signals TEM0 to TEM5 output from the terminal MST and the mode selection signal TL select and test the memory modules of the image processor GP, command memory VRAM, drawing memory FB0, and FBI. .
- the input signals TE0 to TE3 of the mode selection terminal MST are supplied from an external test device (tester) or an external CPU. Therefore, the external input signals TE0 to TE3 generate the internal control signals TEM0 to TEM5 and the mode selection signal TL internally via the mode selection terminal MST, and are input to each module to test each module. Is performed.
- each memory module is connected to the common test path TB by a wired OR, and only the output of the memory module selected by the internal control signals TEM0 to TEM5 is output to the common test bus TB. You. As a result, the number of test wirings can be reduced, and the chip area of the semiconductor integrated circuit device SIC can be reduced.
- FIG. 16 shows a specific configuration of the normal port NP and the test port TP provided in the memory module of the command memory VRAM and the drawing memories FB0 and FBI.
- the normal port NP and the test port TP are configured such that the operation is different for each of the normal mode and the test mode.
- FIG. 16 (a) shows a case of the normal operation mode in which the semiconductor integrated circuit device SIC performs a normal operation.
- the memory module is normally operated from the port NP by the image processor GP. Is accessed.
- the test port TP side is set to a high impedance state based on the selection signal so that no information is output to the outside. That is, in the normal operation mode, the image processor GP and the memory module are operated in a directly connected state.
- the selection signal is generated by ANDing the internal control signals TEM0 to TEM5 and the mode selection signal TL.
- FIG. 16 (b) shows the case of the memory test mode.
- the memory test mode the memory module is accessed from the test port TP card.
- the normal port NP side is set to a high impedance state based on the selection signal, so that no information is output to the outside.
- the image processor GP and the memory module are disconnected, and the memory module is connected to an external test device or an external test device via the test port TP.
- the operation is performed in a state directly connected to CPU.
- the conventional general-purpose semiconductor memory test method can be used as it is for the memory module mounted on the semiconductor integrated circuit device SIC.
- FIG. 16 (c) shows the case of the logic test mode.
- the logic test mode is the test mode of the image processor GP.
- memory modules are usually
- test port TP Accessed from NP.
- external monitoring is possible through the test port TP.
- the operation is performed with the image processor GP and the memory module directly connected, and the memory module connected directly to an external test device or an external CPU via the test port TP. It is made to do.
- the image processor GP communicates with the memory module according to the test pattern of the logic tester, and can monitor the state of the memory module at that time.
- Fig. 17 shows an example of the switching circuit between the normal port NP and the test port TP.
- Transistor gate TG1 composed of n-channel MOS (nMOS) transistor Ql and p-channel MOS (pMOS) transistor Q2, and nMOS transistor Q3 and pMOS transistor Q4
- a switching circuit is configured with the configured transfer gate TG2.
- the transfer gates TG1 and TG2 are controlled by the control signals SN and ST generated from the module selection signals (TL and TEM0 to TEM3).
- a similar function can be realized by a clocked inverter or the like instead of this transfer gate.
- FIG. 18 shows the assignment of the test control pins of the mode selection terminal MST.
- test control pins receive the 4-bit encoded signal, and based on this signal, the internal control signals TEM0 to TEM5 and the mode selection signal TL are switched as shown in FIG. Generated. In addition, based on the internal control signals TEM0 to TEM5 and the mode selection signal TL, each memory module of command.memory VRAM, drawing memory FB0, and FBI is selected and tested.
- the internal control signal TEM is the result of decoding an external input signal to the test control pins (TE0 to TE3), and is output to the image processor GP, command memory VRAM, drawing memory FB0, and FBI modules. It is input and determines the target module at the time of testing. In the present embodiment, the value is “00000000” in the normal operation and in the STNBY mode.
- the mode selection signal TL sets the normal operation mode, logic test mode, and memory test mode.
- Fig. 18 mode When the selection signal TL is "1”, the normal operation mode and logic test mode are set. When the selection signal TL is "0”, the memory test mode is set. In this embodiment, in addition to the normal operation mode, the logic test mode, and the memory test mode, a standby mode can be set.
- the test module in the present embodiment has two DRAM modules (M0-M1, M2-M3, M4) in the logic test mode.
- the test is performed with -M5), and in the memory test mode, the test is performed in units of one DRAM module (M0, Ml, M2, M3, M4, M5). This is based on the difference in the test method between the logic test mode and the memory test mode.In the logic test mode, the test is performed in units of FB0 and FBI. On the other hand, in memory testing mode, testing is performed in DRAM units.
- test control pins (TE0 to TE3), and a module that matches each test method is tested.
- the test control pins (# 0 to # 3) do not necessarily have to be encoded as in the present embodiment, and each test control pin has its own A configuration in which a specific memory module is directly selected may be used.For example, if ⁇ 2 becomes “1”, one memory module in the drawing memory FB0 is selected and a test is performed. You may do it.
- FIG. 19 shows the input / output of each terminal in the mouth test mode shown in FIG. 16 (c).
- test port shown in FIG. And a direct connection to an external test device or an external CPU via a PC, and as shown in Fig. 19, test can be performed for each image processor GP and each memory module accessed by the image processor GP. Has been.
- the test of the image processor GP in the present embodiment is performed by a command for a test and a test ton, which are input from the outside through the normal terminal NT. The turn is executed by the image processor GP. Therefore, the image processor GP only needs to execute the normal operation based on the test pattern using the normal terminal NT, and there is no difference from the normal operation.
- an external data processing device stores a test command and a test pattern in the command memory VRAM via the CPU interface unit CIU described above, and the image processor GP. Is performed by executing the command based on an instruction from an external data processing device.
- the image processor GP is a tester for each target memory module. Perform the turn. Therefore, first, the drawing memory FB0 is targeted, and then, the drawing memory FB1 and the command 'memory VRAM are the target memory modules of the logic test mode. In addition, which memory module is to be observed in the logic test mode is determined by the observation switching signal KS which is a decoding result of an external input signal input to the test control pins (TE0 to TE3). There are two modes: mode 1 for observing the drawing memory FB0, mode 2 for observing the drawing memory FBI, and mode 3 for observing the command memory VRAM.
- Fig. 20 is an overall block diagram mainly of the test of the semiconductor integrated circuit device SIC, and Figs. 21 to 23 show the contents of the input / output pins of the semiconductor integrated circuit device SIC. A summary is shown.
- Each memory module is connected to a common test bus TB.
- the common test bus TB has 11-bit address-nos A, 8-bit power Ramnon-Qua-Dresno, switch C, and 8-bit.
- the semiconductor integrated circuit device SIC has 34 input / output / input / output terminals required for the normal image processor GP, 7 test control terminals, 43 test dedicated terminals, and a power supply ground. It has 16 input / output / input / output terminals. As shown in FIG. 12, the terminals are arranged at 25 on one side.
- Address data VBUS, memory bit enable TEBE, and memory bank address TERC are multiplexed to reduce the number of pins.
- the address data bus VBAS is an address / data bus that performs read / write from an external data processing device to the image processor GP during normal operation, but is a test bus during test mode. Connected to the data bus DQ of the test bus TB so as to input and output the contents of the data bus DQ of the test bus TB. I have.
- the layout is optimized by following the flow of information. Routing can be simplified, and the wiring length can be shortened. As a result, the wiring area can be reduced, and the chip area can be reduced. Furthermore, since the wiring length is short, the signal delay is small, and high-speed operation is possible.
- a test terminal is provided in an image processing device having a frame chip buffer, command memory, and an image processor built into a one-chip, and a test port is provided for each memory module.
- a test bus By providing a test bus and connecting it to a test bus, the contents of each built-in memory module can be externally monitored during a test. Therefore, even if the external terminals for the memory are lost due to the mixed mounting, the conventional test method can be used as it is.
- each of the frame buffer and command memory built into the image processing device is composed of a plurality of identically configured memory modules, and the same address is assigned to each memory module.
- the depth of the memory dress can be increased.
- a plurality of identical configurations can be provided within the range that satisfies the upper limit.
- by using memory modules with the same configuration testing and refreshing of the frame buffer and command memory can be unified. .
- the control logic of the logic state machine is facilitated by equalizing the read and write operations of the frame buffer and the command memory based on the instructions of the image processor. be able to. That is, the image processor executes the non-operation instruction after the output of the write address, thereby equalizing the latencies of the read and write operations. Read and write processing within a single machine can be treated the same. Therefore, it is no longer necessary to consider the combination of read-write, write-read, read-read, and write-write access in the state machine. In addition, this makes it possible to reduce the number of logical gates of the image processor.
- the present invention can be applied to a personal computer or an architecture that realizes high-speed graphic processing such as an amusement device, and is intended to improve the drawing performance of a graphic LSI.
- the memory and graphics controller for the frame buffer and command are built into the one-chip, the layout should be optimized according to the flow of information, or the conventional memory test
- the logic test can be used as it is, and the frame buffer and command memory can be composed of a plurality of identically configured memory modules. It is suitable for reducing the occupied area of the device or realizing an easy-to-use image processing device.
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP1996/000731 WO1997035317A1 (fr) | 1996-03-21 | 1996-03-21 | Processeur a memoire dram integree |
EP96904330A EP0889477A1 (en) | 1996-03-21 | 1996-03-21 | Data processor with built-in dram |
KR1019980707395A KR20000064678A (ko) | 1996-03-21 | 1996-03-21 | Dram내장데이타처리장치 |
US09/142,905 US6496610B2 (en) | 1996-03-21 | 1996-03-21 | Data processing apparatus having DRAM incorporated therein |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP1996/000731 WO1997035317A1 (fr) | 1996-03-21 | 1996-03-21 | Processeur a memoire dram integree |
Publications (1)
Publication Number | Publication Date |
---|---|
WO1997035317A1 true WO1997035317A1 (fr) | 1997-09-25 |
Family
ID=14153094
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP1996/000731 WO1997035317A1 (fr) | 1996-03-21 | 1996-03-21 | Processeur a memoire dram integree |
Country Status (4)
Country | Link |
---|---|
US (1) | US6496610B2 (ja) |
EP (1) | EP0889477A1 (ja) |
KR (1) | KR20000064678A (ja) |
WO (1) | WO1997035317A1 (ja) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2002109885A (ja) * | 2000-09-28 | 2002-04-12 | Toshiba Corp | 半導体記憶装置 |
JP4786805B2 (ja) * | 2001-02-16 | 2011-10-05 | シャープ株式会社 | 半導体装置 |
US20060014151A1 (en) * | 2002-12-25 | 2006-01-19 | Jun Ogura | Optical dna sensor, dna reading apparatus, identification method of dna and manufacturing method of optical dna sensor |
TWI384485B (zh) * | 2008-09-11 | 2013-02-01 | Inventec Corp | 記憶體組態提前檢查電路 |
US9288161B2 (en) * | 2011-12-05 | 2016-03-15 | International Business Machines Corporation | Verifying the functionality of an integrated circuit |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS5891590A (ja) * | 1981-11-27 | 1983-05-31 | Fujitsu Ltd | メモリシステム |
JPH05242257A (ja) * | 1992-02-28 | 1993-09-21 | Hitachi Ltd | 高速曲線描画用ビットマップ・メモリ |
JPH06208632A (ja) * | 1992-08-25 | 1994-07-26 | Texas Instr Inc <Ti> | 図形/画像処理方法および装置 |
Family Cites Families (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE3587458T2 (de) * | 1984-04-10 | 1994-03-24 | Ascii Corp | Videoanzeigesteuersystem. |
JPS60245034A (ja) * | 1984-05-18 | 1985-12-04 | Ascii Corp | デイスプレイコントロ−ラ |
JP3579461B2 (ja) | 1993-10-15 | 2004-10-20 | 株式会社ルネサステクノロジ | データ処理システム及びデータ処理装置 |
JPH08212185A (ja) | 1995-01-31 | 1996-08-20 | Mitsubishi Electric Corp | マイクロコンピュータ |
US5701434A (en) * | 1995-03-16 | 1997-12-23 | Hitachi, Ltd. | Interleave memory controller with a common access queue |
US5941968A (en) * | 1997-04-14 | 1999-08-24 | Advanced Micro Devices, Inc. | Computer system for concurrent data transferring between graphic controller and unified system memory and between CPU and expansion bus device |
US5990913A (en) * | 1997-07-30 | 1999-11-23 | Intel Corporation | Method and apparatus for implementing a flush command for an accelerated graphics port device |
-
1996
- 1996-03-21 WO PCT/JP1996/000731 patent/WO1997035317A1/ja active IP Right Grant
- 1996-03-21 KR KR1019980707395A patent/KR20000064678A/ko active IP Right Grant
- 1996-03-21 EP EP96904330A patent/EP0889477A1/en not_active Withdrawn
- 1996-03-21 US US09/142,905 patent/US6496610B2/en not_active Expired - Fee Related
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS5891590A (ja) * | 1981-11-27 | 1983-05-31 | Fujitsu Ltd | メモリシステム |
JPH05242257A (ja) * | 1992-02-28 | 1993-09-21 | Hitachi Ltd | 高速曲線描画用ビットマップ・メモリ |
JPH06208632A (ja) * | 1992-08-25 | 1994-07-26 | Texas Instr Inc <Ti> | 図形/画像処理方法および装置 |
Also Published As
Publication number | Publication date |
---|---|
EP0889477A1 (en) | 1999-01-07 |
KR20000064678A (ko) | 2000-11-06 |
US20020027556A1 (en) | 2002-03-07 |
US6496610B2 (en) | 2002-12-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP4128234B2 (ja) | メモリ素子、処理システム、メモリ素子を制御する方法およびダイナミックランダムアクセスメモリを操作する方法 | |
US6028810A (en) | Fast accessible dynamic type semiconductor memory device | |
US6041389A (en) | Memory architecture using content addressable memory, and systems and methods using the same | |
US6381190B1 (en) | Semiconductor memory device in which use of cache can be selected | |
EP1936628A1 (en) | Memory device, memory controller and memory system | |
KR20020013785A (ko) | 반도체기억장치 | |
US8341328B2 (en) | Method and system for local memory addressing in single instruction, multiple data computer system | |
US5815456A (en) | Multibank -- multiport memories and systems and methods using the same | |
JP2004500673A5 (ja) | ||
JPH05225774A (ja) | マルチポート半導体記憶装置 | |
KR100227133B1 (ko) | 보존된 어드레싱을 이용하는 메모리 장치 및 이를 이용한 시스템및 방법 | |
EP0771008B1 (en) | A multiple-bank memory architecture and systems and methods using the same | |
US10020036B2 (en) | Address bit remapping scheme to reduce access granularity of DRAM accesses | |
WO1997035318A1 (fr) | Processeur a memoire dram integree | |
JP2593060B2 (ja) | ダイナミックランダムアクセスメモリ、ダイナミックランダムアクセスメモリのアクセス方法及びシステム | |
WO1997035316A1 (fr) | Processeur a memoire dram integree | |
WO1997035317A1 (fr) | Processeur a memoire dram integree | |
JPH08221319A (ja) | 半導体記憶装置 | |
US6222786B1 (en) | Dynamic random access memory with write-without-restore and systems and methods using the same | |
US5963468A (en) | Low latency memories and systems using the same | |
US6504548B2 (en) | Data processing apparatus having DRAM incorporated therein | |
JP2000021200A (ja) | 半導体記憶装置 | |
JP2000163990A (ja) | 半導体記憶装置 | |
JP2009151526A (ja) | 半導体記憶装置及びその半導体記憶装置を用いたシステム | |
JP2001028186A (ja) | 半導体記憶装置、それのアクセス方法、及び、データプロセッシングシステム |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 96180263.4 Country of ref document: CN |
|
AK | Designated states |
Kind code of ref document: A1 Designated state(s): CN JP KR US |
|
AL | Designated countries for regional patents |
Kind code of ref document: A1 Designated state(s): AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
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 | ||
WWE | Wipo information: entry into national phase |
Ref document number: 09142905 Country of ref document: US Ref document number: 1019980707395 Country of ref document: KR |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1996904330 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1996904330 Country of ref document: EP |
|
WWP | Wipo information: published in national office |
Ref document number: 1019980707395 Country of ref document: KR |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1996904330 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1019980707395 Country of ref document: KR |