WO1996038780A2 - Method for performing signed division - Google Patents
Method for performing signed division Download PDFInfo
- Publication number
- WO1996038780A2 WO1996038780A2 PCT/US1996/007614 US9607614W WO9638780A2 WO 1996038780 A2 WO1996038780 A2 WO 1996038780A2 US 9607614 W US9607614 W US 9607614W WO 9638780 A2 WO9638780 A2 WO 9638780A2
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- equal
- dividend
- temporary
- register
- setting
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/38—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation
- G06F7/48—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices
- G06F7/52—Multiplying; Dividing
- G06F7/535—Dividing only
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2207/00—Indexing scheme relating to methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F2207/535—Indexing scheme relating to groups G06F7/535 - G06F7/5375
- G06F2207/5353—Restoring division
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/38—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation
- G06F7/48—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices
- G06F7/499—Denomination or exception handling, e.g. rounding or overflow
- G06F7/49905—Exception handling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/38—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation
- G06F7/48—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices
- G06F7/499—Denomination or exception handling, e.g. rounding or overflow
- G06F7/49936—Normalisation mentioned as feature only
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled
- G06F7/38—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation
- G06F7/48—Methods or arrangements for performing computations using exclusively denominational number representation, e.g. using binary, ternary, decimal representation using non-contact-making devices, e.g. tube, solid state device; using unspecified devices
- G06F7/499—Denomination or exception handling, e.g. rounding or overflow
- G06F7/49994—Sign extension
Definitions
- the present invention relates generally to microprocessor architecture, and more particularly to the architecture of a microprocessor execution unit which performs arithmetic and logic operations concurrently with address computations.
- microprocessors have a "pipelined architecture" whereby the processor is divided into stages. This permits the processor to perform several tasks at once thereby allowing the processor to work on different parts of the instructions simultaneously as they are advanced through the pipe with each clock cycle. Under ideal conditions, one instruction can leave the pipeline and another instruction enter the pipeline every clock cycle.
- One such microprocessor is the Intel486TM microprocessor.
- Compatible x86-type microprocessors include those made by Advanced Micro Devices and Cyrix.
- the x86 microprocessor has a complex instruction set architecture which includes over 400 instructions. However, some of these instructions are rarely used by either the operating system or the compiler. Thus, it would be desirable to optimize the architecture for commonly used instructions.
- the memory of an x86-type microprocessor is organized as a sequence of 8- bit bytes and each byte is assigned a unique physical address.
- application programs do not directly address the physical address, but instead use a virtual addressing scheme whereby the physical address is calculated based on a memory management model that includes segmentation and paging.
- the physical memory is divided into independent memory spaces called segments. Each segment has a segment descriptor which contains its base address and a size limit for that segment.
- An application program issues a logical address which the addressing hardware translates into a linear address by adding an offset to the base address, so long as the offset does not exceed the size limit.
- T e offset is the sum of three components, namely, a displacement value, a base register and an index register.
- Paging is also supported by x86-type architecture, whereby linear address space, which may be part of physical memory or disk storage, is divided into 4k blocks of memory called pages. If paging is employed, addressing hardware translates the linear address into a physical address. If not, the linear address is the same as the physical address.
- x86-type addressing hardware must handle the worst case scenario, namely, wherein the effective address is the sum of the base and all offset components. However, in many applications, only one offset component is present. Therefore, it would also be desirable to optimize the addressing hardware to handle the usual rather than the worst case addressing computation.
- An architecture for a microprocessor execution unit is disclosed.
- the architecture is generally described as having an arithmetic unit and an addressing unit.
- the arithmetic unit performs arithmetic and logical operations on a pair of operands in response to control signals.
- the addressing unit operates in conjunction with the arithmetic unit to calculate linear addresses as well as offsets and limits.
- the arithmetic unit includes a first portion for performing addition, subtraction and logical operations, a second portion for performing multiply, divide and single-bit shift operations, and a third portion for performing multi-bit shift and rotate operations.
- the addressing unit is coupled to the output of the arithmetic unit.
- the base component is input to the addresing unit and the index and displacement components are input to the arithmetic unit.
- the results are summed in a single cycle to yield a linear address.
- Figure 1 is a block diagram showing portions of an integrated microprocessor system
- Figure 2 is a l ⁇ nctional block diagram of the execution unit portion of the integrated microprocessor system of Figure 1
- Figure 3 is a more detailed block diagram of the execution unit of Figure 2
- Figure 4 is a schematic diagram of the result registers for a division operation
- Figure 5 is a schematic diagram of the result registers for a multiplication operation
- Figure 6 is a block diagram of a portion of the execution unit of Figure 3
- the preferred embodiment of the present invention is an integrated microprocessor system 1 having a pipelined architecture, wherein the pipeline includes, for example, a decoder stage 2, an execution unit 3 and a writeback stage 4
- the present disclosure is directed to the execution unit 3, which provides a 32-bit data path for operands and instructions stored in general registers, including register file 6 and shadow register 7, and wherein arithmetic, logical and addressing computations are performed by the execution unit for programming instructions executed by the integrated microprocessor system 1
- the microprocessor system 1 includes a control logic unit 5 which is coupled to send control signals CTRL to the execution unit 3 and to receive data signals SGNL from the execution unit
- the control logic unit 5 is also coupled to other components of the microprocessor system 1 and receives microcode and other input for making programmed control decisions
- the execution unit 3 is implemented to be substanually compatible with the Intel x86 instruction set, as set forth in the "Intel486TM MICROPROCESSOR FAMILY PROGRAMMER'S REFERENCE MANUAL," which is expressly incorporated herein by reference
- the x86 instructions will be referenced herein in their common mnemonic form, such as ADD, SUB, MUL, DIV, etc
- FIG. 2 A simplified functional diagram of the execution unit 3 is shown in Figure 2
- An arithmetic unit 50 has two inputs 51 and 52 for receiving operands OpA and OpB, respectively, from register file 6
- the arithmetic unit 50 generates an arithmetic or logical result 53 in a single cycle for many x86 instructions
- the arithmetic unit 50 includes a first portion for performing addition, subtraction and logical operations, a second portion for performing multiply, divide and single-bit shift operations, and a third portion for performing multi-bit shift and rotate operations, as will be shown and desc ⁇ bed in more detail below
- An addressing unit 55 has one input selectively coupled to the output of the arithmetic unit 50 or to OpA When an addressing instruction is received, the segment base component is provided to the addressing unit 55 on input 56, and the base, index, or displacement components, or immediate segment address operands, are provided to the arithmetic unit 50 on inputs 51 and 52 The addressing unit 55 then sums the address components to yield output 57 which is a linear address
- a limit check unit 60 is provided to make sure the offset 58, 1 e , output 53 or OpA, is not addressing a location outside of the segment as determined by the control signal LIMIT
- Sign extension unit 101 is a 3: 1 multiplexor that selects a byte and sign extends it into 32 bits, or selectes a word and sign extends it into 32 bits, or selects a dword, and then outputs the 32 bit result a, concern onto data line 201.
- the term "sign extend” means copying the sign bit into the 24 highest order bits for a byte or into the 16 highest order bits for a word.
- Operand B is received into a sign extension unit 102.
- Sign extension unit 102 includes a 5: 1 multiplexor that selects a signed byte and sign extends it into 32 bits, or a signed word and sign extends it into 32 bits, or an '.nsigned byte and sign extends it into 32 bits, or an unsigned word and sign extends it into 32 bits, or a dword.
- the output 202 is a 32 bit result b m .
- Sign extension unit 102 also includes a 2: 1 multiplexor that selects OpB or its complement.
- An adder 103 receives and operates upon data lines 202 and 203 and carry input Cl 204.
- Data line 203 is from the output of a 2:1 multiplexor 104, which selects either a, genre data line 201 or UpperQ data line 205.
- Adder 103 performs logical operations on data lines 202 and 203 to generate logic output 207, which is available to the user through output gate 111.
- the adder 103 also performs addition on data lines 202, 203 and 204 to generate sum output 206, which is available to the user through output gate 112.
- Two 32 bit registers are provided for performing multiply, divide and single-bit shift operations.
- a 3:1 multiplexor 105 selects from a, n data line 201, UpperQ data line 205, or SUM data line 207. The selected value may be shifted either left or right by one bit by left/right shifter 106 and then stored in register 107.
- a 2:1 multiplexor 108 selects from b, n data line 202 or from LowerQ data line 208. The selected value may be shifted either left or right by one bit by left right shifter 109 and then stored in register 1 10.
- the least significant bit (LSB) of left/right shifter 106 is coupled to the most significant bit (MSB) of left/right shifter 109 to permit up to 64 single bit position shifts.
- the UpperQ register 107 provides an output data line 205 which is fed back to multiplexor 104 or multiplexor 105, as described above, or made available to the user through output gate 113.
- the LowerQ register 1 10 provides an output data line 208 which is fed back to multiplexor 108 as described above, or available to the user through output gate 1 14.
- a barrel shifter 120 comprising a 32 by 32 transistor array is provided for performing multi-bit shift and rotate operations.
- a pair of 32 bit 2:1 multiplexors 121, 122 couple the a, n data line 201 to the barrel shifter 120.
- a 5 bit decoder 123 provides 32 output signals, only one of which is true, to the barrel shifter 120, thus selecting one row of the barrel shifter.
- the output 209 of the barrel shifter 120 is available to the user through output gate 115.
- a multiplexor 130 selectively outputs status flags from the execution unit 3 through output gate 116, as shown in Table I:
- ZF Zero Flag: Zero result set ZF to 1 ; else ZF is cleared
- the addressing unit includes a 4: 1 multiplexor 152 that selects the a, compassion data line 201 if it is a dword, or zero extends the a, compassion data line 201 if it is a word, or the SUM output 206 if a dword, or zero extends the SUM output if a word
- zero extend means copying a zero into the 24 highest order bits for a byte or into the 16 highest order bits for a word
- An adder 154 receives the output from multiplexor 152 as well as the segment base value on data line 56 and adds the two values together, thereby generating a linear address 57
- a limit check unit 160 is also provided in execution unit 3
- the address includes a 20 bit limit value 162 which is stored in the shadow register 7
- This limit value is provided to multiplexor 164, where it is scaled to 32 bits, depending on the value of the granularity bit, then inverted through 32 bit inverter 166
- the output of inverter 166 is coupled to an adder 168, in which only the carry out function is used, and to a multiplexor 170.
- the output of multiplexor 152 is also coupled to adder 168.
- the output B of adder 168 indicates that the offset is below the scaled limit value.
- the multiplexor 170 is provided with constants HC (half ceiling) and FC (full ceiling), which provide the maximum value for addressing computations and cause selection of either 16 bit addresses (HC) or 32 bits addresses (FC).
- the output of multiplexor 170 which is the upper limit for address computations, is fed to adder 174. which is a carry save adder (CSA).
- Additional inputs to CSA 174 are from multiplexor 152 and multiplexor 176.
- the output of CSA 174 is fed to the input of adder 178 and to a single bit left shift unit 180, which effectively multiplies the value of the carry bits by 2
- the output of shift unit 180 is fed to the adder 178
- the output SegSpace of adder 178 is used for a limit calculation by a prefetch unit (not shown) and the output A of adder 178 indicates that the offset in above the scaled limit value
- execution unit 3 for arithmetic and logical instructions will now be described in more detail.
- Instructions for addition, subtraction, and logical operations are carried out in a conventional manner by utilizing the resources ot adder 103.
- a division example of 50 by 7 yields a quotient of 7 with a remainder of 1, as shown in Table II (truncated to 8 bits).
- Register 1 10 is used to provide the quotient while register 107 is used to provide the remainder, as illustrated in Figure 5
- the lower 8 bits of register 1 10 contain the quotient while the lower 8 bits of register 107 contain the remainder
- the lower 16 bits of register 1 10 contain the quotient while the lower 16 bits of register 107 contain the remainder
- all 32 bits of register 110 contain the quotient while all 32 bits of register 107 contain the remainder.
- step 302 the divisor is set equal to the absolute value of the 32 bit sign extension of OpB.
- step 304 the divisor is compared to zero. If true, then an interrupt occurs in step 306 and the routine stops. If not, thei. the data length is determined in step 308 and a temporary register initialized in step 310. If the data length is a byte, then register tempi is set equal to 80 hex. If the data length is a word, then register tempi is set equal to 8000 hex. If the data length is a dword, then register tempi is set equal to 80000000 hex.
- step 312 the dividend OpA is examined to determine if it is negative. If not, then the program jumps to step 320. Is so, then the lower dividend (register 110) is adjusted to become the two's complement value of the lower dividend in step 314.
- step 316 the adder 103 is examined to see if there is a carry out. If so, go to step 318. If not, go to step 319.
- step 318 an adjusted upper dividend is set equal to the two's complement of the upper dividend (register 107).
- step 319 the adjusted upper dividend is set equal to the one's complement of the upper dividend.
- step 320 the adjusted upper dividend is set equal to the upper dividend.
- step 322 the adjusted lower dividend is aligned to be left justified.
- step 324 a division carry register is set equal to the most significant bit of the adjusted upper dividend, then the adjusted upper dividend is shifted left one bit position, then the LSB of the adjusted upper dividend is set equal to the MSB of the adjusted lower dividend, then the adjusted lower dividend is shifted to the left by one bit position, and finally, the LSB of the adjusted lower dividend is set equal to zero.
- a temporary result register stores the result of subtracting the value in register tempi from the adjusted divisor. Then, the temporary result register is set equal to the adjusted upper dividend less the temporary result register.
- step 328 the temporary result and the adjusted upper register are compared to zero. If true, then the parity flag hPF is set equal to one (step 329). If not, then the parity flag hPF is set equal to zero (step 330).
- step 332 the size is defined based on the data length, i.e., a byte, a word, or a dword.
- Step 334 includes several sub steps. Step 334a is the first division step and is basically a subtract then shift left. Step 334b calls for comparing the hidden parity flag hPF to 1. If true, then go to step 334c, else go to step 334d.
- Step 336 checks to see if the adjusted upper dividend and the division carry register are greater than the adjusted divisor, and that the divisor and dividend are positive values. If so, then an interrupt is generated in step 338 and the routine stops. If not, then the size is compared to 0 in step 340. If true, then go to step 350. If not, go to step 342.
- step 342 a normal division operation is performed.
- step 344 the size is decremented by one.
- Step 350 is a division end step that is similar to the normal division step, except that the difference is not shifted left one bit, but is stored directly into the upper register 107.
- the lower register 110 is updated as before.
- step 352 the temporary remainder is set equal to the upper register 107. If, in step 354, the temporary remainder is 0, and the divisor is greater than 0, and Hpf equals 1, then an interrupt is generated (step 356) and the routine stops. If not, then a temporary quotient is set equal to the lower register 110 in step 358.
- step 360 the sign of the divisor is compared to the sign of the dividend. If equal, a second temporary quotient is set equal to the first temporary quotient in step 362. If not, then the second temporary quotient is set equal to the complement of the first temporary quotient in step 364. In step 366, if the sign of the second temporary quotient is not equal to the exclusive OR of the sign of the dividend with the sign of the divisor, and the second temporary quotient is not equal to 0, then an interrupt is generated (step 368) and the routine stops If not, then go to step 370
- step 370 The dividend is examined in step 370 to see if it is negative If so, then the remainder is set equal to the temporary remainder in step 372 If not, then the remainder is set equal to the complement of the temporary remainder step 374
- the product of a multiplication operation is contained in registers 110 and 107 as illustrated in Figure 5
- a 16-bit result is contained in the upper 8 bits of register 110 and the lower 8 bits of register 107
- a 32-bit result is contained in the upper 16 bits of register 110 and the lower 16 bits of register 107
- a 64-bit result is contained all 32 bits of register 110 and all 32 bits of register 107
- the barrel shifter 120 and associated multiplexors 121 and 122 may be used to carry out multi-bit shift and rotate operations, as is more fully described in the following commonly assigned, copendmg applications “BARREL SHIFTER' by Thomas W S Thomson and H John Tarn as filed on May 26, 1995, (2) "BIT SEARCHING THROUGH 8, 16, OR 32-BIT OPERANDS USING A 32-BIT DATA PATH” by Thomas W S Thomson as filed on May 26, 1995, and (3) "METHOD FOR PERFORMING ROTATE THROUGH CARRY USING A 32-BIT BARREL SHIFTER AND COUNTER” by H John Tarn as filed on May 26, 1995
- Double precision shift operations are also fully supported by the execution unit 3, as more fully described in commonly assigned, copendmg application entitled “DOUBLE PRECISION (64-BIT) SHIFT OPERATIONS USING A 32-BIT DATA PATH” by Thomas W S Thomson and filed on May 26. 1995
- Addressing computations for x86 segmented address space are optimized in execution unit 3 for the predominant cases, l e , where the address consists only of two components, namely a scaled index and a displacement, or a base and a displacement
- the execution unit is capable of performing the entire address computation in a single cycle, I e , it can perform calculate the offset, the linear address and the limit in a single cycle
- FIG. 7 An address cycle is illustrated schematically in Figure 7
- a 32-bit segment base address is provided to input 56 and defines the memory segment space in which an operand resides
- a 32-bit or 16-bit segment offset value is added to the segment base to form the linear address
- the offset value is constructed from up to two general registers, namely a base register or an index register, and a literal displacement value, which is an 8-bit, 16-bit, or 32-bit value taken from the addressing instruction format
- the index register can be scaled by a factor of 2, 4, or 8 before use, thereby allowing the index register to count elements rather than bytes when indexing through an array
- the invention embodiments described herein have been implemented in an integrated circuit which includes a number of additional functions and features which are described in the following co-pending, commonly assigned patent applications, the disclosure of each of which is incorporated herein by reference: U.S. patent application Serial No. 08/ , entitled "DISPLAY CONTROLLER
Abstract
Description
Claims
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP96925249A EP0772815A2 (en) | 1995-05-26 | 1996-05-23 | Method for performing signed division |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US08/451,571 US5754460A (en) | 1995-05-26 | 1995-05-26 | Method for performing signed division |
US08/451,204 | 1995-05-26 |
Publications (2)
Publication Number | Publication Date |
---|---|
WO1996038780A2 true WO1996038780A2 (en) | 1996-12-05 |
WO1996038780A3 WO1996038780A3 (en) | 1997-01-09 |
Family
ID=23792763
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/US1996/007614 WO1996038780A2 (en) | 1995-05-26 | 1996-05-23 | Method for performing signed division |
Country Status (3)
Country | Link |
---|---|
US (1) | US5754460A (en) |
EP (1) | EP0772815A2 (en) |
WO (1) | WO1996038780A2 (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102156625A (en) * | 2011-03-31 | 2011-08-17 | 北京大学 | Method for performing division calculation by utilizing rheostatic element |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6125380A (en) * | 1998-04-13 | 2000-09-26 | Winbond Electronics Corporation | Dividing method |
DE10055659C1 (en) * | 2000-11-10 | 2002-03-28 | Infineon Technologies Ag | Calculation circuit for division of fixed point signal uses adders and logic stage with AND and OR logic |
JP4712247B2 (en) * | 2001-08-31 | 2011-06-29 | 富士通セミコンダクター株式会社 | Microprocessor development system for application programs involving integer division or integer remainder |
US7174358B2 (en) * | 2002-11-15 | 2007-02-06 | Broadcom Corporation | System, method, and apparatus for division coupled with truncation of signed binary numbers |
US7165086B2 (en) * | 2002-11-15 | 2007-01-16 | Broadcom Corporation | System, method, and apparatus for division coupled with rounding of signed binary numbers |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4224676A (en) * | 1978-06-30 | 1980-09-23 | Texas Instruments Incorporated | Arithmetic logic unit bit-slice with internal distributed iterative control |
US5097435A (en) * | 1988-12-24 | 1992-03-17 | Kabushiki Kaisha Toshiba | High speed dividing apparatus |
Family Cites Families (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4381550A (en) * | 1980-10-29 | 1983-04-26 | Sperry Corporation | High speed dividing circuit |
JPS60140428A (en) * | 1983-12-28 | 1985-07-25 | Hitachi Ltd | Divider |
JP3098242B2 (en) * | 1988-07-13 | 2000-10-16 | 日本電気株式会社 | Data processing device |
US5027309A (en) * | 1988-08-29 | 1991-06-25 | Nec Corporation | Digital division circuit using N/M-bit subtractor for N subtractions |
US5204953A (en) * | 1989-08-04 | 1993-04-20 | Intel Corporation | One clock address pipelining in segmentation unit |
US5259006A (en) * | 1990-04-18 | 1993-11-02 | Quickturn Systems, Incorporated | Method for substantially eliminating hold time violations in implementing high speed logic circuits or the like |
US5189319A (en) * | 1991-10-10 | 1993-02-23 | Intel Corporation | Power reducing buffer/latch circuit |
US5254888A (en) * | 1992-03-27 | 1993-10-19 | Picopower Technology Inc. | Switchable clock circuit for microprocessors to thereby save power |
US5493523A (en) * | 1993-12-15 | 1996-02-20 | Silicon Graphics, Inc. | Mechanism and method for integer divide involving pre-alignment of the divisor relative to the dividend |
US5404473A (en) * | 1994-03-01 | 1995-04-04 | Intel Corporation | Apparatus and method for handling string operations in a pipelined processor |
US5574677A (en) * | 1994-11-23 | 1996-11-12 | Exponential Technology, Inc. | Adaptive non-restoring integer divide apparatus with integrated overflow detect |
-
1995
- 1995-05-26 US US08/451,571 patent/US5754460A/en not_active Expired - Fee Related
-
1996
- 1996-05-23 WO PCT/US1996/007614 patent/WO1996038780A2/en active IP Right Grant
- 1996-05-23 EP EP96925249A patent/EP0772815A2/en not_active Withdrawn
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4224676A (en) * | 1978-06-30 | 1980-09-23 | Texas Instruments Incorporated | Arithmetic logic unit bit-slice with internal distributed iterative control |
US5097435A (en) * | 1988-12-24 | 1992-03-17 | Kabushiki Kaisha Toshiba | High speed dividing apparatus |
Non-Patent Citations (2)
Title |
---|
COMPUTER DESIGN, vol. 16, no. 5, May 1977, LITTLETON, MASSACHUSETTS US, pages 124-127, XP002015482 S. SANYAL: "An ALgorithm for Nonrestoring Division" * |
K. HWANG: "Computer arithmetic: principles, architecture, and design" 1979 , J. WILEY & SONS , NEW YORK XP002015483 pages 218-221 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102156625A (en) * | 2011-03-31 | 2011-08-17 | 北京大学 | Method for performing division calculation by utilizing rheostatic element |
Also Published As
Publication number | Publication date |
---|---|
WO1996038780A3 (en) | 1997-01-09 |
EP0772815A2 (en) | 1997-05-14 |
US5754460A (en) | 1998-05-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US5272660A (en) | Method and apparatus for performing integer and floating point division using a single SRT divider in a data processor | |
US5450607A (en) | Unified floating point and integer datapath for a RISC processor | |
US5917741A (en) | Method and apparatus for performing floating-point rounding operations for multiple precisions using incrementers | |
US6269384B1 (en) | Method and apparatus for rounding and normalizing results within a multiplier | |
EP0685787A1 (en) | Multibit shifting apparatus, data processor using same, and method therefor | |
US5381360A (en) | Modulo arithmetic addressing circuit | |
US5426600A (en) | Double precision division circuit and method for digital signal processor | |
US5511017A (en) | Reduced-modulus address generation using sign-extension and correction | |
EP0685786A1 (en) | Combined multiplier/shifter and method therefor | |
US5682339A (en) | Method for performing rotate through carry using a 32 bit barrel shifter and counter | |
US5301139A (en) | Shifter circuit for multiple precision division | |
KR20000053047A (en) | Eight-bit microcontroller having a risc architecture | |
JP2500098B2 (en) | Digital computer system | |
KR100351527B1 (en) | Code breakpoint decoder | |
JPS6014338A (en) | Branch mechanism for computer system | |
US5655139A (en) | Execution unit architecture to support X86 instruction set and X86 segmented addressing | |
US5754460A (en) | Method for performing signed division | |
KR100431726B1 (en) | Method for performing signed division | |
US5237525A (en) | In a data processor an SRT divider having a negative divisor sticky detection circuit | |
US5687102A (en) | Double precision (64 bit) shift operations using a 32 bit data path | |
EP1089166A2 (en) | An integer instruction set architecture and implementation | |
US6393554B1 (en) | Method and apparatus for performing vector and scalar multiplication and calculating rounded products | |
US5649147A (en) | Circuit for designating instruction pointers for use by a processor decoder | |
US7580967B2 (en) | Processor with maximum and minimum instructions | |
US5583453A (en) | Incrementor/decrementor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AK | Designated states |
Kind code of ref document: A2 Designated state(s): DE KR |
|
AL | Designated countries for regional patents |
Kind code of ref document: A2 Designated state(s): AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
AK | Designated states |
Kind code of ref document: A3 Designated state(s): DE KR |
|
AL | Designated countries for regional patents |
Kind code of ref document: A3 Designated state(s): AT BE CH DE DK ES FI FR GB GR IE IT LU MC NL PT SE |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1019970700540 Country of ref document: KR |
|
WWE | Wipo information: entry into national phase |
Ref document number: 1996925249 Country of ref document: EP |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
WWP | Wipo information: published in national office |
Ref document number: 1996925249 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: 1019970700540 Country of ref document: KR |
|
WWW | Wipo information: withdrawn in national office |
Ref document number: 1996925249 Country of ref document: EP |
|
WWG | Wipo information: grant in national office |
Ref document number: 1019970700540 Country of ref document: KR |