UP 7685_1106_System_Description_Apr69 7685 1106 System Description Apr69
UP-7685_1106_System_Description_Apr69 UP-7685_1106_System_Description_Apr69
User Manual: UP-7685_1106_System_Description_Apr69
Open the PDF directly: View PDF .
Page Count: 122
Download | |
Open PDF In Browser | View PDF |
UNIVAC liDS ERRATA SHEET TO SYSTEM DESCRIPTION, UP·7685 Programming Information Exchange Bu Iletin 1, UP·7700.1 .Page 6, Figure 2-1 -- Minimum system requirement is 2 UNISERVO* VI C or 2 UNISERVO* VIII C Magnetic Tape Units • • Page 30, paragraph 4.7.9. -- At end of paragraph, add: "*The j and a fields together serve to specify any of the 128 control registers." .Page 51, Table of UNISERVO* VIII C CHARACTERISTICS - For TAPE SPEED, change "42.5" to read "120" • • Page 93, Appendix B -- For topmost word format, change "MULTI PLE" to read "MULTI PLY" • • Page 105, Appendix D -- In introductory sentence, change "STAT-PACK" to read "MATH-PACK". Please mark your copies of UP-7685 according to this Errata Sheet. Bulk distribution being made is the same as distribution for UP-7685. For details refer to original bulk distribution sheet. For additional copies, this sheet and original manual order via a Sales Help Requisition through your local Univac Representative from Holyoke, Massachusetts. All Holyoke stock will contain this Errata Sheet. MANAGER, Documentation and Library Services *Trademark of S erry Rand Cor orationp-__----------1ST UP-7700.1 to Lists lOU, 217, 630, 692 and distribution on Special Bulk Distribution Sheet. OAT April 25, 1969 01 VISION OF SPERRY RAND CORPORA TION S.P L . S, P.O. BOX 8 1 00 PHILADELPHIA, PA. 19101 U P-38C50 REV. 2 PRINTED IN U.S.A. © 1969 _ SPERRY RAND CORPORATION CONTENTS CONTENTS i 1. INTRODUCTION 2. SYSTEM DESCRIPTION 2 2.1. GENERAL 2 2.2. SYSTEM COMPON ENTS 2.2.1. Central Processor 2.2.2. Main Storage 2.2.3. Auxiliary Storage 2.2.4. Operator's Control Console 3 3 3 3 4 2.3. CONFIGURATIONS 5 3. MAIN STORAGE 8 3.1. GENERAL 8 3.2. STORAGE MODULE 9 3.3. PACKAGING 9 3.4. STORAGE CAPACITY 9 3.5. STORAGE PROTECTION 3.5.1. Storage Protection Modes 3.5.1.1. Privileged Mode 3.5.1.2. User Program Mode 10 10 10 10 3.6. RELATIVE ADDRESSING 11 4. CENTRAL PROCESSOR UNIT 12 4.1. GENERAL 12 4.2. PRINCIPAL SECTIONS 12 4.3. INSTRUCTION WORD FORMAT 4.3.1. Function Code 4.3.2. Partial-Word or Immediate Operand Designator 4.3.3. Control Register Designato r 4.3.4. Index Register Designator 4.3.5. Index Modification Designator 4.3.6. Indirect Address Designator 4.3.7. Address Field 13 13 13 13 13 13 13 14 4.4. CO NTRO L REG ISTERS 4.4.1. Index Registers 4.4.2. Arithmetic Accumulators 4.4.3. Access Control Registers 4.4.4. R Registers 4.4.4.1. RO - Real Time Clock 4.4.4.2. R1 - Repeat Counter 4.4.4.3. R2 - Mask Register 14 14 15 15 15 15 17 17 4.5. ARITHMETIC SECTION 4.5.1. The Adder 4.5.2. Arithmetic Accumulators 4.5.3. Parti ai-Word Transfers 4.5.4. Split-Word Arithmetic 4.5.5. Sh i fti ng 4.5.6. Double-Precision Fixed-Point Arithmetic 4.5.7. Floating-Point Arithmetic 17 18 18 18 19 19 20 20 4.6. EXECUTIVE SYSTEM CONTROL FEATURES 4.6.1. Processor State Register 4.6.2. Interrupts 4.6.3. Guard Mode 22 22 24 24 4.7. INSTRUCTION REPERTOIRE 4.7.1. Data Transfer Instructions 4.7.2. Fixed-Poi nt Arithmeti c 4.7.3. Floating-Point Arithmetic 4.7.4. Index Regi ster Instructions 4.7.5. Logical Instructions 4.7.6. Shift Instructions 4.7.7. Repeated-Search Instructions 4.7.8. Unconditional Jump Instructions 4.7.9. Conditional Jump Instructions 4.7.10. Test (or Skip) Instructions 4.7.11. Executive System Control Instructions 4.7.12. InpuVOutput Instructions 4.7.13. Other Instructions 26 26 27 27 28 28 29 29 30 30 30 31 31 31 5. PROCESSOR INPUT/OUTPUT CONTROL SECTION ii 32 5.1. GENERAL DESCRIPTION 32 5.2. PERIPHERAL CONTROL 32 5.3. INTERNALLY SPECIFIED INDEX MODE 32 5.4. EXTERNALLY SPECIFIED INDEX MODE 33 5.5. BUFFER MODE DATA TRANSFERS 35 6. 5.6. INPUT/OUTPUT INFORMATION WORDS 5.6.1. DataWords 5.6.2. Function Words 5.6.3. Identifier Words 5.6.4. Status Wo rds 35 35 36 36 36 5.7. PRIORITY CONTROL 36 5.B. INPUT/OUTP UT INSTRUCTIONS 5.B.1. Monitored Instructions 37 37 PERIPHERAL SUBSYSTEMS 38 6.1. AVAILABLE PEHIPHERAL EQUIPMENT 38 6.2. THE FLYING HEAD DRUMS 6.2.1. UNIVAC FH-432/FH-1782 Magnetic Drum Subsystem 6.2.2. UNIVAC FH-432 Magnetic Drum 6.2.3. UNIVAC FH-1782 Magnetic Drum 6.2.4. UNIVAC FH-880 Magnetic Drum 39 40 41 6.3. FASTRAND MASS STORAGE SUBSYSTEM 44 6.4. UNISERVO MAGNETIC TAPE SUBSYSTEMS 6.4.1. UNISERVO VI C Magnetic Tape Subsystem 6.4.2. UNISERVO VIII C Magnetic Tape Subsystem 6.4.3. UNISERVO VIII C/VI C Subsystem 6.4.4. Fully Simultaneous UNISERVO VIII C Subsystem 47 48 50 6.5. UNIVAC HIGH SPEED PRINTER SUBSYSTEM 52 6.6. PUNCHED CARD SUBSYSTEM 6.6.1. UNIVAC Card Reader 6.6.2. UNIVAC Card Punch 54 54 6.7. UNIVAC 9300 SUBSYSTEM 55 7. DATA COMMUNICATIONS 42 43 51 52 55 57 7.1. GENERAL 57 7.2. UNIVAC 1106 COMMUNICATIONS SUBSYSTEM 57 7.3. UNIVAC SYNCHRONOUS DATA TERMINALS 7.3.1. Word Terminal Synchronous 7.3.2. UN IV AC Commun i cation Terminal Synchronous 58 58 59 7.4. UNIVAC DATA COMMUNICATION TERMINAL (OCT) 2000 60 iii iv 7.5. UNISCOPE 300 VISUAL COMMUNICATION TERMINAL SUBSYSTEM 7.5.1. The Keyboard 7.5.2. CRT Display 7.5.3. Subsystem Configurations 7.5.4. Reliability 7.5.5. Special Features 63 63 64 64 64 65 7.6. UNIVAC 9200/9300 REMOTE SUBSYSTEMS 7.6.1. The UN IV AC 9200 System 7.6.2. The UNIVAC 9300 System 7.6.3. The UNIVAC Data Communication Subsystem, DCS-l 66 67 67 69 8. PROGRAMMED SYSTEMS SUPPORT 70 8.1. AVAILABLE SOFTWARE 70 8.2. TH E EXECUTIVE SYSTEM 8.2.1. Multiple Modes of Operation 8.2.1.1. Batch Pro cessi ng 8.2.1.2. Demand Processing (Time-Sharing) 8.2.1.3. Real Time 8.2.1.4. Multiprogramming 8.2.2. Techniques for Utilization of Mass Storage 8.2.3. The Primary Functional Areas of the Executive System 8.2.3.1. Executive Control Language 8.2.3.2. The Supervisor 8.2.3.3. Tim e SI i ci ng 8.2.3.4. Storage Compacting 8.2.3.5. Faci I ities Assignment 8.2.3.6. The File-Control System 8.2.3.7. Operator Communi cations 8.2.3.8. The Diagnostic System 8.2.3.9. Input/Output Device Handlers 8.2.3.10. Auxiliary Processors 8.2.3.11. Utilities 8.2.3.12. System Setup 70 72 72 72 73 73 73 74 74 74 74 75 75 75 8.3. THE ASSEMBLER 8.3.1. Symbolic Coding Format 8.3.2. Assembler Directives 8.3.3. Additional Features 75 75 76 76 8.4. FORTRAN V 8.4.1. Language Extensions and Enhancements 8.4.2. Compi I er 0 rgani zation 76 76 80 8.5. CONVERSATIONAL FORTRAN V 8.5.1. System Features 8.5.2. System Concepts 8.5.3. Conversational Processor and the Executive System 8.5.4. Conversational FORTRAN Language 81 81 81 82 82 71 71 71 71 71 72 8.6. LIFT, FORTRAN II TO FORTRAN V TRANSLATOR 82 8.7. COBOL 8.7.1. The P rocesso r 8.7.2. Special Features 83 84 85 8.8. ALGOL 85 8.9. SORT/MERGE 85 8.10. MATHEMATICAL FUNCTION PROGRAMS 86 8.11. APPLICATION PROGRAMS 8.11.1. Linear Programming System 8.11.2. APT III 8.11.3. PERT 87 87 88 88 8.12. MATH-PACK 89 8.13. STAT-PACK 89 APPENDICES A. NOTATIONAL CONVENTIONS 91 B. SUMMARY OF WORD FORMATS 93 C. INSTRUCTION REPERTOIRE BY FUNCTION CODE 95 D. MATH-PACK ROUTINES 105 E. STAT-PACK ROUTINES 109 FIGURES 1-1. The Central Processor and Operator's Console 2-1. Minimum Multi-Module Storage Configuration 6 2-2. 262K Multi-Module Storage Configuration 7 4-1. Control Register Address Assignments 16 4-2. J-Determined Partial Word Operation 19 4-3. PJocessor State Register Format 23 7-1. CTMC Subsystem 57 TABLES 4-1. Fixed-Address Assignments 25 7-1. OCT 2000 Field Installable Options 61 v 1. INTRODUCTION A medium-scale extension to the proven UNIVAC 1100 series of computers is provided by the UNIVAC 1106 System. Its modular design provides growth potential within the UNIVAC 1106 .as well as possible upgrading to the larger UNIVAC 1108 System. The UNIVAC 1106 permits the user to select an executive system most applicable to his needs. This selection ranges from EXEC II, which provides a powerful batch processing capability, to the UNIVAC 1106 Executive System with its real-time, multiprogramming , demand , and batch processing ability. In addition to a versatile executive system, the UNIVAC 1106 utilizes the complete family of UNIVAC 1108 unit proces sor software, application packages , and peripheral subsystems. Figure 7 -7. The Central Processor and Operator' s Console 1 2. SYSTEM DESCRIPTION 2.1. GENERAL The UNIVAC 1106 System is a general purpose, high performance unit processor system, making good use of the latest advances in computer design, systems organization, and programming technology. Its modular structure permits the selection of systems components to fulfill most efficiently the speed and capacity requirements for applications of varying complexity. As the workload increases, this modularity also enables the addition of input/ ou tput subsystems and main storage. Among the principal features of the 1106 System are: 2 • Large, modular, parity-checked, high speed main storage • Program address relocation • Storage protection • Partial-word addressability in 6, 9, 12, and 1S-bit portions as well as full-word (36 bits) and doub1e- word (72 bits) addressing. • High speed, random access, auxiliary storage • Privileged mode for the Executive system • Guard mode for user programs. 2.2. SYSTEM COMPONENTS The UNIVAC 1106 System is organized to allow a number of tasks to be performed simultaneously under the direction of a common Executive control system. The system is composed of four types of components: • Central Processor • Main Storage • Auxiliary Storage Subsystems • Peripheral Subsystems 2.2.1. Central Processor The processor can perform all functions required for the execution of instructions including arithmetic, input/ output, and Executive control. Included in the processor is a set of 166-nanosecond integrated circuit control registers providing multiple accumulators, index registers, input/output access control registers, and specialuse registers. 2.2.2. Main Storage The main storage read/restore cycle time is 1.5 microseconds. The mInImUm multi-module storage of the UNIVAC 1106 System is 65,536 36-bit words and is expandable up to a maximum of 262,144 words. 2.2.3. Auxiliary Storage The auxiliary magnetic drum storage subsystems are an integral part of each UNIVAC 1106 System. Up to eight FH-432 or FH-1782 magnetic drums, or any combination of the two types, may be attached to one or two control units. Both types of drum can transfer data at 1,440,000 characters per second (240,000 words per second). The average access time of the FH-432 is 4.3 milliseconds; that of the FH-1782 is 17 milliseconds. 3 2.2.4. Operator's Control Console The UNIVAC 1106 Operator's Control Console subsystem is a free-standing input/ output device for directing and monitoring the operation of the CPU. The console is always connected to input/ output channel 15. The basic Control Console includes the following components: • Display Console The Display Console contains a keyboard and a Cathode Ray Tube (CRT) display. The keyboard is a standard four-bank keyboard which can generate 63 Fieldata codes. A row of eight interrupt keys is located above the keyboard. The Cathode Ray Tube can display 16 lines of 64 characters each. One UNIVAC PAGEWRITER printer is required for the minimum configuration if this console is selected . • Day Clock The day clock on the console displays the time of day in hours, minutes, and hundreths of minutes. It furnishes the time of day to the CPU every 600 milliseconds and sends a day clock interrupt signal to the CPU every 6 seconds. The day clock may be manually disabled from the operator's console. The day clock may be selected to be active either externally or by program. • Operator's Control and Indicator Panel The Operator's Control and Indicator Panel includes fault, disable, and mode indicators for its CPU and associated main storage modules; displays and controls associated with selecting and releasing jumps and stops, with the Program Address Counter and Memory Select Register, and with the time display of the Day Clock. It also includes system controls associated with the CPU and subsystems which are logically connected to the CPU. 4 • Additional Features - Auxiliary Console An auxiliary right- or left-wing console can be included to accomodate control/display panels for Communication Terminal Module Controller subsystems. - UNIVAC PAGE WRITER Printer The UNIVAC PAGEWRITER printer provides a hard copy of all messages for a permanent record of all completed transactions between the operator and the Executive system. The PAGEWRITER prints lines of up to 80 characters per second. 2.3. CONFIGURATIONS Figure 2-1 depicts the minimum multi-module storage configuration for the UNIVAC 1106 System. The CPU consists of the Command-Arithmetic Section and an Input/Output Section with four I/O channels. Main storage provides 65K 36-bit words. The minimum peripheral subsystems needed to support the software offered are: 1 FH-432/1782 Magnetic Drum Subsystem with 6 FH-432 drums or 1 FH-1782 drum, 1 UNISERVO Magnetic Tape Sub7" system with either 4 UNISERVO VI C or 4 UNISERVO VIII C Magnetic Tape Units, an Operator's Console, and 1 online UNIVAC 9300 System. Figure 2-2 illustrates the 262K multi-module storage configuration for the UNIVAC 1106 System. 5 POWER SUPPLY 65K STORAGE MODULE CENTRAL PROCESSOR UNIT I/O CHANNELS 12-15 r PERIPHERAL SUBSYSTEMS I I I l I CONSOLE I CONSOLE AND MINIMUM PERIPHERAL SUBSYSTEMS FOR 1106 OPERATING SYSTEM: 1 Operator's Console with: Display Console and PAGEWRITER Printer 1 FH-432/1782 Magnetic Drum Subsystem with: 1 FH-1782 Drum or equivalent 1 UNISERVO Magnetic Tape Subsystem with: 4 UNISERVO VI C Magnetic Tape Units; or 4 UNISERVO VIII C Magnetic Tape Units 1 Online UNIVAC 9300 System with: 8K Main Storage 132 Print Position Printer Multiplexer I/O Channel 600 cpm 80-Column Card Reader Inter-Computer Control Unit Figure 2-7. Minimum Multi-Module Storage Configuration 6 POWER SUPPLY POWER SUPPLY POWER SUPPLY POWER SUPPLY 65K STORAGE MODULE 65K STORAGE MODU LE 65K STORAGE MODULE 65K STORAGE MODULE 1 I I I CENTRAL PROCESSOR UNIT I I I I I I I I I I 1 I CONSOLE PERIPHERAL SUBSYSTEMS USE TIME METER - - - - - OPTIONAL NOTE: EXEC 1\ supports 131K main storage maximum. Figure 2 -2. 262 K Multi-Module Storage Configuration 7 3. MAIN STORAGE 3.1. GENERAL The main storage of the UNIVAC 1106 System is a high performance, fast access repository for instructions and data. Its design fully supports the concepts of multiprogramming, modularity, and reliability around which the entire UNIVAC 1106 System is constructed. Among its featured characteristics are: • 1.5 microsecond read/restore cycle time • 65,536, 98,304, 131,072, 196,608, or 262,144 36-bit words of multi-module storage • Parity checking on all storage references • Hardware storage protection - lockout boundaries establishable in 512-word increments • Relative addressing and dynamic program relocatability through program base registers • Online serviceability - multi- module storage pairs may be removed for servicing withou stopping the entire system. Main storage performs the following functions: • Accepts an address from the processor. • Stores or retrieves a word at that address • Issues an acknowledgement signifying that a storage reference has been completed • Checks parity on all data and delivers an interrupt sig(lal to the processor should a parity error occur. The modular storage concept has significant advantages for the immediate as well as the future needs of the system. Addition of banks of storage is simplified. It becomes a simple matter to add storage elements, or to replace them with improved equipment bank by bank, as technology advances. 8 3.2. STORAGE MODULE Each module of multi-module storage includes 32,768 words of ferrite core array. Each word is 36 bits long, and carries two additional parity bits in nonaddressable levels, one bit for each half word. The main components of the module are a IS-bit address register, a 36-bit read/restore register, parity checking circuits, and request/acknowledge circuits. The IS-bit address register of each storage module provides addressing for 32,768 words. Since an 18-.bit address is generated within the processor at each storage reference, three bits are available for selection of one of the four possible storage modules. Parity is checked on reading or calculated on writing for each storage access. If a parity error is detected, the storage bank sends a parity error interrupt signal to the processor and rewrites the word in its incorrect form to ensure subsequent data errors when the word is again referenced. Preservation of the error in this way facilitates fault location, since the Executive can determine whether the failure is transient or is associated with a marginal or complete failure of the module. 3.3. PACKAGING For multi-module storage, one or two 32,768-word storage modules (module pair) within a single cabinet constitute a bank. An adjacent cabinet contains DC power supplies for operation of the bank. 3.4. STORAGE CAPACITY Available multi- module storage capacity ranges from 65,535 words to the system maximum of 262,144 words in accordance with the following: 65,536 words (two modules) - Minimum for system 98,304 words (three modules) 131,072 words (four modules) 196,608 words (six modules) 262,144 words (eight modules) - Maximum for system 9 3.5. STORAGE PROTECTION To prevent inadvertent program reference to out-of-range storage addresses, the 1106 processor includes a hardware storage protection feature. The controlling element in this feature is the Storage Limits Register, the contents of which are as follows: DATA AREA INSTRUCTION AREA UPPER BOUNDARY LOWER BOUNDARY 27 35 26 UPPER BOUNDARY 18 17 LOWER BOUNDARY 9 8 a The Storage Limits Register (SLR) can be loaded by the Executive system to establish allowable operating areas for the program currently in execution. These areas are termed the program instruction (I) and data (D) areas. Before control is given to a particular program, the Executive loads the SLR with the appropriate I and D boundaries. Before each main storage reference, the processor performs a limits check on the address, comparing against the limits of either the I or D field of the SLR. An outof-limits address generates a guard mode interrupt, thereby allowing the Executive to regain control and take appropriate action. 3.5.1. Storage Protection Modes The Executive system can establish two different modes of storage protection by means of control fields in the Processor State Register (PSR) described in Section 4. Normally, the Executive itself operates in open mode; that is, the Storage Limits Register may be loaded but the PSR is set to disregard this, and the Executive can reference any location in main storage. 3.5.1.1. Privileged Mode Another mode can be establis4ed in the PSR for privileged programs. This privileged mode protects against out-of-bounds writes. Privileged programs (such as real-time programs or Executive-controlled subroutines) may enter non-alterable (re-entrant) su brou tines, which are part of the Executive. Though these privileged programs are assumed to be thoroughly checked out, the system is still fully protected against unexpected occurrences since write protection is in effect. 3.5.1.2. User Program Mode In the user program mode, read, write, and jump storage protection is in effect. Therefore, user programs are limited to those areas assigned by the Executive. If the user program reads, writes, or jumps to an out-of-limits address, an interrupt returns control to the Executive for remedial action. Read/jump protection allows the Executive to stop the program at the point of error, terminate it, and provide diagnos tic informa tion to the programmer thereby minimizing wasted time and smoothing the checkout process. 10 3.6. RELA TIVE ADDRESSING Relative addressing is a feature of great significance in multiprogramming, timesharing, and real-time operations, for it allows storage assignments for one program (the one going into execution) to be changed dynamically by the Executive to provide continuous storage for operation of another program, and it permits programs to dynamically request additional main storage according to processing needs. An additional advantage is that systems programs stored in auxiliary storage may be brought in for operation in any available area without complicated relocation algorithms. Relative addressing is provided for through base registers contained within the CPU. Two separate registers control the basing of the program instruction and data bank, and a third register controls the selection of the appropria te base register. 11 4. CENTR.AL PROCESSOR UNIT 4.1. GENERAL 1:he UNIVAC 1106 Central Processor Unit (CPU) is the principal component of the UNIVAC 1106 System and, generally, the one by which the entire sys tern is identified. It can operate under Executive or user program modes of control; it performs both arithmetic and logical operations; and it accommodates and supervises up to 16 input/output channels. 4.2. PRINCIPAL SECTIONS The processor is logically divided into six interacting sections each of which is identified and briefly described below. 12 • Control Registers - The CPU has 128 program-addressable control registers used for arithmetic operations, indexing, and input/output buffer control. • Arithmetic Section - This section contains the adder regis ters, and control circuits necessary for performing fixed and floating point arithmetic, partial-word selection, shifting, logical operations, and tests. • Control Section - This section provides the basic control and logic for instruction decoding and execution. It includes the Program Address Counter used for the sequential accession of instructions; the Program Control Register in which instructions are staticized for execution; and the Processor State Register (PSR), which determines various processor operating modes. The Control Section also services interrupts. • Input/Output Section - This section controls and multiplexes data flow between main storage and 16 input/output channels. It includes an interrupt priority network and paths to peripheral subsystems for both function signals and data. • Indexing Section - This section contains parallel index adders and threshold test circuitry. It is used generally for processor control functions, operand address development, program relocation, and input/output transfer control. • Storage Class Control Section - The Storage Class Control section receives the final operand address from the index adder and establishes address and data paths to one of eight possible storage modules. Storage Class Control also determines whether a final address refers to a control register. 4.3. INSTRUCTION WORD FORMAT The format of the 1106 instruction word is illustrated below followed by an explanation of each field. Some fields have more than one meaning depending on the class of ins truction. f 4.3.1. x u Function Code These six bits specify the operation to be performed. For function codes above 70 , 8 the f and j fields are combined to produce a 10-bit function code. An illegal function code generates an interrupt. 4.3.2. Partial-Word or Immediate-Operand Designator For function codes less than 70 8 , the j designator specifies partial-word or immediateoperand selection. (See Figure 4-2 for specific partial-word selections.) 4.3.3. Control Regis ter Designator The four-bit a field designates which control register, within a group selected by the function code, is involved in the operation. For some operations, the a field refers to an arithmetic register; for others, it refers to either an index register or some other control register. In input/output instructions, it specifies the channel and its associated input or output access control register. For function code 708 the a and j designators together address one of the 128 control registers. 4.3.4. Index Register Designator The x field specifies one of the 15 index registers to be used in address modification. When register 00 is designated, indexing is suppressed. 4.3.5. Index Modification Designator The h field controls modification of the index value (Xm) by the increment field (Xi) after indexing (see 4.4.1). If h = 1, the right half of the index register is modified by the contents of its left half; if h = 0, modification is suppressed. 4.3.6. Indirect Address Designator The i designator controls the use of indirect addressing during instruction execution. If i = 0, the instruction functions normally. If i = 1, the 22 least significant bit positions of the instruction (x, h, i and u fields) are replaced in the instruction register with the contents of the 22 least significant bit positions of the word at the effective address. Indirect addressing con tinues as long as i = 1 with full indexing capa bility a t each level. 13 4.3.7. Address Field The u field normally specifies the operand address. However for certain instructions it holds constants. For example, the shift instructions use the. seven least significant bit positions to hold the shift count. In all instructions, the value in the u field may be modified by the contents of an index register. 4.4. CONTROL REGISTERS The 128 program-addressable control registers are grouped to provide multiple index registers, accumulators, input/output access control registers, and special registers (see Figure 4-1). The control registers consist of 36-bit integrated-circuit registers, with a basic cycle time of 166 nanoseconds. Two parity bits are included with each control register. Effective use of multiple accumulators and index registers for the development and use of cons tan ts, index values, and operands substantially improves performance. UNIVAC 1106 compilers, for example, perform significantly better through multiple register usage, and can produce highly efficient. code. In the following descriptions only programmable registers are discussed. The Executive, through modes established by the Processor State Register, has exclusive use of tge duplicate set o-f control registers as well as the Access Control Registers indicated by the shaded areas in Figure 4-1. 4.4.1. Index Registers Control register locations DO-IS are Index Registers and have the following format: X·I 35 MODIFIER INCREMENT OR DECREMENT Xm 18 17 INDEX MODIFIER o The Xm portion of the index register is an 18-bit modifier to be added to the base operand address of the ins truction. The Xi portion of the index word updates the Xm portion, after base operand address modification. Index register modification is specified by a 1 bit in the h field of the instruction, while indexing itself is specified by a nonzero value in the x field. Both functions take place within the basic instruction execution cycle. When cascaded indirect addressing is used in a programmed operation, full indexing capabilities are provided at each level. Indirect addressing replaces the x, h, i and u portions of the instruction register, beginning with a new indexing cycle for each cascaded sequence. This process continues until the i field is zero. 14 4.4.2. Arithmetic Accumulators Control register locations 12-27 are arithmetic accumulators, for programmed storage of arithmetic operands and results. The computation is performed in nonaddressable transient registers wi thin the arithmetic section. Depending upon the instruction, use of the accumulators results in a variety of word formats. Double precision instructions and a number of logical instructions reference two contiguous accumulators, i.e., A and A + 1. In arithmetic operations, A + 1 always holds the least significant part of an operand or result. Some ins tructions, such as single precision floating point operations, ca11 on a oneword operand from memory but produce a two-word result in the specified A and A + 1. 4.4.3. Access Control Registers Control register loca tions 32-63 are Input and Output Access Control Re gis ters (ACR 's). They are guard mode protected and may be referenced only by the Executive. Formats of the Access Control Words are detailed in Section 5. The word-by-word transmission of data over an I/O channel is governed by the contents of the ACR's. Two ACR's, one for input and one for output, are assigned to each of the sixteen channels. Input ACR's (locations 32-47) control input data transfers while output ACR's (locations 48-63) govern the transmission of output data and Function Words. When an input/output operation is initiated, the programmed ACR word is loaded into the ACR corresponding to the channel associated with the specified peripheral unit. 4.4.4. R Registers The sixteen control register locations 64-79 are R Registers. The first three of these (RO, R1, R2) have specified functions and formats as described below. The remaining R Registers are not specifically assigned; typically they are used as loop counters, transient regis ters, or storage for intermedia te values or constants. 4.4.4.1. RO - Real Time Clock UNASSIGNED CLOCK COUNT This register is initially loaded by the program. The contents are then decremented once each 200 microseconds. A real time clock interrupt occurs when the clock count goes through zero. Thus, if the clock is initially loaded with the value 5000, an interrupt occurs in exactly one second. 15 DECIMAL OCTAL > I 27 33 34 16 ACCUMULATORS (A) ) 28 4 UN ASSIGN ED 31 37 32 V(BU'FER ADDRESS) *16 INPUT ACCESS CONTROL REGISTERS OR 57 OR INPUT IDENTIFIER OUTPUT IDENTIFtER 60 v w G 12 JII _______ ~ *16 OUTPUT ACCESS CONTROL REGISTERS 63 __ R.uJ.:.1J!41.~~L _________ J..OL __________ MASK l~I ESI IDENTIFIER REGISTERS 41 48 __________ ...§! ~~A~OU~~~ST~ 103 _________ ~ _ _ _ _ _ _ _ _ _ _ _6.§. !!f~ll..R 16 SPECIAL REGISTERS (R 67 UNASSIGNED 117 79 lu.. __ -· _____ .- __.J!.NE!!., <) • Literals up to 132 characters • The ENTER Verb • The LOCK option on the CLOSE Verb • The ADVANCING option on the WRITE Verb • The REVERSED option of the OPEN Verb • Operands used in arithmetic can be up to 18 digits long. • AND and OR connectors in compound conditions • Parentheses in compound conditions • All abbreviations of conditional statements 83 • The OBJECT-COMPUTER paragraph • The APPLY clause • RERUN • The DATA-COMPILED clause • Library provisions • Multiple results from arithmetic verbs The following features, implemented in UNIVAC 1106 COBOL, are special UNIVAC extensions to the COBOL language: COBOL sub-program communication MONITOR Dynamic date Common Storage Page Control A Report Generator, not part of the COBOL processor, is available for COBOL programs. 8.7.1. The Processor The COBOL processor is a six-phase compiler. It is com pletely modular in relocatable elements and it is handled as any program in the system for easy expandability and maintenance. Likewise, the COBOL processor produces as its output relocatable binary elements stored on the drum or mass storage, which are indistinguishable from other elements in the system. Other output from the compiler includes extensive diagnostic messages, source language listings, machine language listings, and special cross reference listings of name definitions and their references. The machine language listing consists of side-by-side Procedure Division statements and the corresponding generated symbolic machine code. The compiler diagnostics are of two categories: Warning - A minor source language error has been detected which does not affect the program being produced. This type of diagnostic is identified by the word ERROR preceding the actual message. Fatal - A major source language error has been detected which very likely will adversely affect the program being produced. The compiler will continue to process the source language but will flag the program produced so that it cannot easily be executed. This type of diagnostic is identified by the word ERROR* preceding the a ctual message. 84 8.7.2. Special Features Segmentation - COBOL programs can be segmented by use of priority numbers on procedural sections. Monitor - Provides dynamic program checkout facilities. Library - The Procedure Definition Processor is available to store Environment, Data and Procedure Division Descriptions so they can be retrieved by the COpy and INCLUDE verb. Rerun - The programmer can specify rerun after any number of records have been processed or when an end of reel is encountered. Common Storage - Since COBOL programs can be chained (an Executive function), intermediate data results can be maintained between programs using the common storage provision of UNIVAC COBOL. The elements sharing common storage may be from another 1106 processor such as FORTRAN V. Overpunched Sign Convention - Tapes and cards prepared in the overpunched sign convention can be processed. 8.8. ALGOL The ALGOL language allows the mathematician or engineer to prepare programs for the UNIVAC 1106 without the necessity of becoming familiar with the details of the internal machine operation. From this pseudo-mathematical source language, the ALGOL Compiler generates efficient coding in a relocatable binary format acceptable to the Executive for execution, the filing system for cataloging and filing, or both. The basis for the UNIVAC 1106 ALGOL is the "Revised Report on the Algorithmic Language, ALGOL 60" (Communications of the ACM, Vol. 6, January 1963, 1-17). UNIVAC 1106 ALGOL is an extended hardware representation of ALGOL 60 designed to employ the UNIVAC 1106 processor and associated peripheral equipment efficiently. Certain extensions to basic ALGOL have been made. It can handle the powerful input/ output logic; it has the ability to name strings; and it is capable of performing complex and double precision arithmetic. 8.9. SORT/MERGE The UNIVAC 1106 Sort/Merge package is fully modular, with every functional unit completely self-contained. This permits the various units to be individually adapted to their own particular tasks, ena bling them to be a ssocia ted in the most effective form, and allowing updating and augmentation. The package is not a generator of specialized Sort/Merge routines; rather, the user calls and adapts the independent modules for all his specific sorting needs by presenting his parameter values on control cards at load time. 85 In the internal sort, the replacement selection method is used to take advantage of any inherent sequence in the original data. Strings may be written upon magnetic tape or drum. The UNIVAC FH-432 and FH-1782 Drums, because of their high transfer rates and rapid access, minimize processor waiting time and thus greatly speed efficient sort operations. Any random access unit areas may be defined by the Supervisor and these are automatically used by the subsystem if advantage can be gained thereby. The input data to be sorted may be stored on magnetic tape, punched cards or magnetic drum. User own coding may be inserted on the first and final passes of the Sort and Merge operation and may also replace the standard comparison routines. Sorting generally requires the use of two magnetic tape units, although additional units can be employed to give faster times. Keys may be in multiple form and can be recorded, modified, and packed. Standard collating sequences are intrinsically provided for, but the user may define any collating sequences he requires, up to a maximum of seven, and any combination of these may be utilized in the same run. Fixed or variable length items can be handled. The Sort/Merge Package normally uses 20,000 words of main storage, 262,000 words of magnetic drum storage, and magnetic tape units of any kind as required; but the user may specify more main and drum storage, and additional magnetic tape units to increase efficiency and speed. S.10. MATHEMATICAL FUNCTION PROGRAMS The UNIVAC Software System includes an extensive collection of basic mathematical subroutines and functions. This collection includes all of the standard FORTRAN functions and has been expanded by over 50% to give the programmer a more complete coverage of the often used mathematical routines. Each of these mathematical routines has been carefully developed to offer the programmer maximum accuracy and range with a minimum routine size and executive time. These routines are available to each of the program languages, FORTRAN V, Assembler, COBOL (through the use of the ENTER verb option), and ALGOL. One group of routines; the series of exponentiation routines (NEXPi), are automatically referenced when the FORTRAN V Source Prqgram indicates exponentiation with the operator **, and inline exponentiation is not feasible. The various mathematical function programs are: 1. Library Functions 2. The FORTRAN Built-in Function 3. The Exponentiation Functions These routines are available to the process ors in different manners. For example, the routine SQRT provides a single precision square root of the argument (x). To utilize this routine the processors employ the following calling sequences: • FORTRAN V Calling Sequence: ROOT = SQRT (X), when X is a FORTRAN V real variable. 86 • ASSEMBLY Calling Sequence: a LMJ Bl1,SQRT a + 1 + Add res s of X a + 2 + normal return The result is left in AO. • COBOL Calling Sequence: ENTER SQRT REFERENCING X 8.11 APPLICATION PROGRAMS The UNIVAC 1106 System has an extensive library of application programs and subroutines. The major application programs such as Linear Programming (LP), Automatically Programmed Tools (APT III), and PERT are briefly described in the following paragraphs while others are simply mentioned by titles. This is by no means an exhaustive list of programs or subroutines. It is meant only to point out the many types of application programs available for UNIVAC 1106 System. 8.11.1. Linear Programming System Linear Programming eLP) has become one of the most useful and frequently used operations research techniques in manQfacturing and transportation industries. In the production and distribution of products, LP provides a solution to minimize costs or maximize profits. The LP System developed for the UNIVAC 1106 System embodies the latest advances in computer technology with the most powerful algorithm to date. The algorithm employs the "product form of the inverse" method and is improved with an advanced path selection technique. The package is coded in FORTRAN V and Assembly Language. The more prominent feature of the 1106 LP System are as follows: 2. The speed and random access properties of magnetic drums place the system at a distinct advantage over tape and disc handling procedures. 3. Both single precision and double precision computations are available. The selection may be manual or automatic. 4. The control language of this System is far superior to any other existing LP System. It is an interpretive control language. The sophisticated user may use macros in constructing his command string to implement the System. On the other hand, the average user may still execute his basic LP problem withou t detail knowledge of the con trol language. 5. The LP System is imbedded in the Executive System. This enhances the System as a powerful model builder. Matrix builders and output analyzers may be attached to the LP System to form corporate models. The entire model may be optimized and re-optimized in cycles in one computer run. Nonlinear programming may be accomplished by solving approximated linear functions in each cycle. 87 In addition to the above features, the 1106 LP System contains the flexibilities existing in other LP Systems. Vector levels can be specified and coefficients may be modified. Long LP runs can be split with restart procedures. Post-optimal parametric programming or a complete tableau can be obtained. The final output includes the objective function value, optimal basis, vector levels, and reduced costs. 8.11.2. APT III APT (Automatically Programmed Tools) is a system for the computer assisted programming of numerically controlled machine tools, flame cutters, drafting machines, and similar equipment. It is production-oriented, written to take full advantage of numerically controlled techniques in engineering and manufacturing with the least expenditure of effort, time, and money. APT enhances most of the usual advantages found in numerical control: reduced lead time, greater design freedom and flexibility, lower direct costs, greater accuracy, improved production forecasting, lower tooling costs, better engineering control of the manufacturing process, and simplified introduction of changes. The APT III program represents over one hundr~d man-years of development and testing. After extensive experience with our earlier program, APT II, the Aerospace Industries Association made a new start and wrote APT III from the beginning, during the calendar year 1961. At the completion of this package, APT III was turned over to the Illinois Institute of Technology Research Institute for further development, under the APT long-range program. The use of certain parts of APT requires membership in this long range program. Univac participated in the original writing of APT III and has been a member of the APT long-range program from the beginning. Numerical control specialists are continually working to keep the UNIVAC 1106 APT program in the forefront of the art. As implemented on the UNIVAC 1106 System APT III will continue to conform to the latest APT long range program specifications. 8.11.3. PERT The UNIVAC 1106 PERT System is a generalized applications program based upon the framework provided by the "DOD/NASA Guide to PERT/COST System Design". The DOD/NASA design is based upon the concept of costing work packages rather than individual network activities. A work package is a discrete unit of work required to complete a specific job or process. The work packages of a research and development project are directly related to activities or groups of activities on the project network. The work package is the basic unit of the PERT System for which actual project costs are collected and compared with estimates for purposes of cost control. The UNIVAC 1106 PERT System adheres rigorously to the DOD/NASA design and satisfies the purposes of PERT/COST concept. Many government agencies and contractors are currently processing PERT/TIME data on existing systems. PERT is relatively new and will be subject to the inevitable modifications that will result from its initial pilot tests by DOD. In order to lessen the impact of these changes upon existing PERT programs and to provide for efficient integration of time and cost data, a modular design was adopted. 88 The modular structure of the program permits separate processing of the time networks and of the work package costing structure while simultaneously providing for integrated time and cost reporting. The PERT/TIME Module of the UNIVAC 1106 PERT System accepts as input a deck of cards describing the PERT network. The cards are processed and used to update the PERT/TIME Master File. Network computations are performed and the time reports are generated. The PERT / COST Module accepts the cost breakdown structure, actual and estimated costs for the project work packages, and a table of labor rates and applicable overhead percentages. Cost data is accumulated up through the cost breakdown tree, integrated with the time information, and the required cost reports are generated. 8.12. MATH-PACK MATH-PACK provides the UNIVAC 1106 System with a comprehensive library of 78 fundamental mathematical subprograms coded in FORTRAN V. The purpose of this library is to present to the mathematician, the scientist, and the engineer many of the more frequently used tools of numerical analysis. These subroutines and function subprograms are designed to speed up and simplify solutions to problems encountered in many areas of scientific research. The subprograms are grouped into fourteen categories: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) Interpolation Numerical Integration Solution Of Equations Differentiation Polynomial Manipulation Matrix Manipu la tion: Real Ma trices Matrix Manipulation: Complex Matrices Matrix Manipulation: Eigenvalues and Eigenvectors Matrix Manipulation: Miscellaneous Ordinary Differential Equations Systems Of Equations Curve Fitting Pseudo-Random Number Generators Specific Functions Each of these classes contains subroutines and function subprograms that are generally useful for problems commonly encountered by mathematicians, scientists, and engineers. Appendix D lists all of the MATH-PACK subprograms. 8.13. STAT-PACK STAT-PACK provides the UNIVAC 1106 System with a comprehensive library of 91 fundamental statistical subprograms coded in FORTRAN V. The purpose of this library is to present to the statistician, the scientist, the operations research specialist, and the engineer many of the more frequently used tools of statistical analysis. These subroutines and function subprograms a re designed to speed up the preparation of solutions to s tatis tical problems encountered within many areas of scientific research. 89 The subprograms are grouped into thirteen categories: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) Descriptive Statis tics Elementary Population Statistics Distribution Fitting and Plotting Chi-Square Tests Significance Tests Confidence Intervals Analysis of Variance RegreSSion Analysis Time Series Analysis Multivariate Analysis Distribution Functions Inverse Distribution Functions Miscellaneous Each of these classes contains subroutines and function subprograms that are generally useful for problems commonly encountered by statisticians, scien tis ts, and engineers. Appendix E lists all of the STAT-PACK subprograms. 90 APPENDIX A. NOTATIONAL CONVENTIONS Abbreviations and symbols frequently used in the description of the instruction repertoire are given below: ( ) Contents of register or address with in parentheses. ( )' Complement of contents of register or address. \( )\ Absolute va lue or magnitude. ( )17-00 Subscripts indicate the bit positions involved. A full word is normally not subscripted. Subscripts are also used to designate octal or decimal notation. )c ( )f Floating point biased exponent. Final contents. )i Initial contents. )m Floating point fixed point part. )j j-designated portion. Function code. Partial word designator or function code extension. a Arithmetic register designator. In input/output instructions, "att designates an I/O channel. A Ar ithmetic Register. x Index register des ignator. xa Index register designator in a-field. X Index Register. Xa Index Register specified by coding xa. Xm Modifier portion of an index register. Xi Increment portion of an index register. Same as rae ra Designator specifying an R Register. It is coded in the a-designator position of an instruction word. R R Register. Ra R Register specified by coding rae 91 u The base address of the operand (or the actual operand) as coded in u-field of an instruction. U The effective address or value of the operand after application of indexing and indirect addressing. Ud Destination address. Us Source address. h h-designator of the instruction word. A value of 1 specifies incrementation of an index register. i-designator of the instruction word. A value of -1 specifies indirect addressing. PSR Processor State Register. BI BS I-Base Storage Block Number. } Program Effective Switch point. PSR Base Re lative Address ing BD D-Base Storage B lock Number. Fields SLR CSR Storage Limits Register. Channel Select Register. P Program Address Register. Bml Symbol denoting logical product, or logical AND. mil Symbol denoting logical sum, or inclusive OR. ElmI Symbol denoting logical difference, or exclusive OR. ~ 92 0 irection of data flow. APPENDIX B. SUMMARY OF WORD FORMATS FIXED-POINT MULTIPLE SINGLE INTEGER RESULT ADD HALVES WORD FORMA T Carry ________________ ~t ,~ ______________ --'t Carry _ _ _ _ _ _ _ _ ADD THIRDS WORD FORMAT Carry ________ ~t ~I __________ SINGLE-PRECISION FLOATING-POINT OPERAND S 35 34 CHARAC TERI STI C (BIASED EXPONENT) MANTISSA 27 26 SINGLE-PRECISION FLOATIN G-POIN T RESUL T S 35 34 CHARACTERISTI C (BIASED EXPONENT) MANTISSA (NORMALIZED) 2726 A CHARACTERISTI C 35 34 (BIASED EXPONENT) 2726 MANTISSA (NOT NECESSARILY NORMALIZED; CONTAINS RESIDUE, LEAST SIGNIFICANT WORD OF PRODUCT, OR· REMAINDER) A +1 DOU BL E-P RECISION FLOATIN G-POIN T OPERAN D OR RESUL T S 35 34 CHA RAC TERI STI C (BIASED EXPONENT) MANTISSA 2423 A MANTISSA A +1 STORAGE LIMITS WORD 35 I-PORTION UPPER LIMIT 2726 PROCESSOR STATE WORD I-PORTION LOWER LIMIT D-PORTION LOWER LIMIT 18 17 9 8 D-PORTION UPPER LIMIT NOT USED + BS ~I----------------------~~------------------~' • DesignatJr Section QUARTER- Re lative BD Addr~ss ing Section WORD DESIGNATOR 93 INSTRU C TION WORD x INDEX REGISTER WORD Xi 151 A CCESS CON TROL WORD W v ESI A CC ESS CONTROL WORD (HALF WORD) v E51 ACCESS CONTROL WORD (QUARTER WORD) v W SINGLE-PRECISION FIXED-POINT WORD DOU BLE-PRECISION FIXED-POINT WORD A A+l FI XED-POIN T IN TEGER MUL TI PLY RESUL T A A +1 FIXED-POINT FRACTIONAL MULTIPLY RESULT A A +1 BIASED ESI VALUES IN IACR'S BIASED OUTPUT ESI VALUE BIASED INPUT ESI VALUE 17 94 o APPENDIX C. INSTRUCTION REPERTOIRE BY FUNCTION CODE Function Code (Octo I) Mnemonic Instruction . f J 00 - 01 Desc ri ption Execution Time* in f1 sec. With Overlap - Illegal Code Causes illegal instruction interrupt O-J 5 SA Store A (A)~U 1.5 02 0-15 SN,SNA Store Negative A -(A)~U 1.5 03 0-15 SM,SMA Store Magn itude A I(A)I~U 1.5 04 0-15 S,SR Store R (Ra)~U 1.5 05 0-15 SZ Store Zero ZEROS~U 1.5 06 0-15 S,SX Store X (Xa)~U 1.5 07 - - Illegal Code Causes illegal instruction interrupt to address 2418 1.5 10 0-17 L,LA Load A (U)~A 1.5 11 0-17 LN,LNA Load Negative A -(U)~A 1.5 12 0-17 LM,LMA Load Magn itude A I(U)I~A 1.5 13 0-17 LNMA Load Negative Magnitude A -I(U)I~A 1.5 14 0-17 A,AA Add to A (A)+(U)~A 1.5 15 0-17 AN,ANA Add Negative To A (A)-(U)~A 1.5 16 0-17 AM,AMA Add Magn itude To A (A)+I(U)I~A 1.5 17 0-17 ANM,ANMA Add Negative Magnitude to A (A)-I(U)I~A 1.5 20 0-17 AU Add Upper (A)+(U)~A+' 1.5 21 0-17 ANU Add Negative Upper (A)-(U)~A+1 1.5 22 0-15 BT Block Transfer (Xx+u)~Xa+u, 23 0-17 L,LR Load R (U)~Ra 1.5 24 0-17 A,AX Add To X (Xa)+(U)~Xa 1.5 25 0-17 AN,ANX Add Negative To X (Xa)-(U)~Xa 1.5 26 0-17 LXM Load X Modifier (U)~Xa ; Xa 17-0 27 0-17 L,LX 30 0-17 MI *See notes at end of table. repeat K times unchanged - 3.5+3.0K 1.666 35-18 Load X (U)~Xa 1.5 Mu Itiply Integer (A) • (U)~A,A+ 1 3.666 95 Function Code (Octal) f Mnemonic Description Instruction J 31 0-17 MSI Multiply Single Integer (A) • (U)~A 3.666 32 0-17 MF Multiply Fractional (A) • (U)~A,A+1 3.666 33 - - Illegal CoCie Causes illegal instruction interrupt to address 2418 34 0-17 01 Divide Integer (A,A+1)7(U)~A; REMAINDER~A+1 13.950 35 0-17 DSF Divide Single Fractional (A)7(U)~A+ 1 13.950 36 0-17 OF Divide Fractional (A,A+1) 7(U)~A; REMAINDER~A+1 13.950 37 - - Illega I Code Causes illegal instruction interrupt to address 2418 40 0-17 OR Logical OR (A) 41 0-17 XOR Logical Exclusive OR (A)EIiII (U)~A+1 42 0-17 AND Logical AND (A) 43 0-17 MLU Masked Upper Load [(U) f.1mJ (R2~ OR [(A) (R2),] ~ A+1 44 0-17 TEP Test Even Parity Skip NI if (U) m!] (A) have even parity 3.00/2.166 45 0-17 TOP Test Odd Par ity Skip NI if (U) rJm] (A) have odd parity 3.00/2.166 46 0-17 LXI Load X Increment (U)~Xa 1.833 47 0-17 TLEM Test Less Than or Equa I to Mod ifier Skip NI if (U) $. (X a) ; 17-0 TNGM Test Not Greater Than Mod ifier (Xa) W (U)~A+ 1 r..1ml ; Xa 35-18 17-0 - 1.5 1.5 (U)~A+ 1 always (Xa) 1.5 mE unchanged 1.5 17-0 3.333/1.833 + (X a) ~ 35-18 17-0 50 0-17 TZ Test Zero Skip NI if (U) = ±o 3.166/1.666 51 0-17 TNZ Test Nonzero Skip NI if (U) -# ±o 3.166/1.666 52 0-17 TE Test Equal Skip NI if (U) = (A) 3.166/1.666 53 0-17 TNE Test Not Equal Skip NI if (U) f= (A) 3.166/1.666 54 0-17 TLE Test Less Than or Equal Skip NI if (U) ~ (A) 3.166/1.66 TNG Test Not Greater 55 0-17 TG Test Greater Skip NI if (U) > (A) 3.166/1.666 56 0-17 TW Test Within Range Skip NI if (A)«U)< (A+1) 3.33/1.66 *See notes at end of table. 96 Execution Time* in flsec. With Overlap Function Code (Octal) f Mnemonic Instruction Description Execution Time* in p.sec. With Overlap J 57 0-17 TNW Test Not Within Range Skip NI if (U)~ (A) or (U»(A+l) 3.33/1.66 60 0-17 TP Test Positive Skip NI if (U~5 = 0 3.0/1.5 61 0-17 TN Test Negative Skip NI if (U~5 = 1 3.0/1.5 62 0-17 SE Search Equal Skip NI if (U)=(A). else repeat 3.5+ 1.5K 63 0-17 SNE Search Not Equal Skip NI if (U):;l(A). else repeat 3.5+ 1.5K 64 0-17 SLE SNG Search Less Than or Equal Search N at Greater Skip NI if (U)~ (A). else repeat 3.5+1.5K 65 0-17 SG Search Greater Skip NI if (U» 3.5+ 1.5K 66 0-17 SW Search Within Range Skip NI if (A)«U)::;'(A+l). else repeat 3.5+ 1.5K 67 0-17 SNW Search Not Within Range Skip NI if (U)::;' (A) or (U» (A+ 1). else repeat 3.5+ 1.5K 70 t JGD Jump Greater and Decrement Jump to U if (Control Register)ja> 3.0/1.5 (A). else repeat 0; go to NI if (Confrol Register)ja ::;. 0; always (Control Register)ja-1 ~ Control Registerja 71 00 MSE Mask Search Equal Skip NI if (U) rJm] (R2) = (A) (R2). else repeat 3.5+ 1.5K Skip NI if (U) Bml (R2):;l (A) (R2). else repeat 3.5+1.5K Skip NI if (U) mID (R2) ~ (A) else repeat 3.5+1.5K mD 71 01 MSNE Mask Search Not Equal mID 71 02 MSLE Mask Search Less Than or Equal MSNG Mask Search Not Greater Dml ((R2). 71 03 MSG Mask Search Greater Skip NI if (U) BID] (R2»(A) flm] (R2). else repeat 3.5+1.5K 71 04 MSW Masked Search Within Range Skip NI if (A) amJ (R2) < (U) Dml (R2) ~ (A+l) rJml (R2). else repeat 3.5+ 1.5K 71 05 MSNW Masked Search Not Within Range Skip NI if (U) rJml (R2)::;' (A) or (U) (R2) > (A+l) (R2). else repeat 3.5+1.5K I.1IDl (R2) Masked Alphanumeric Search Less Than or Equal mm Skip NI if (U) rJml (R2) ~ (A) (R2). else repeat 3.5+1.5K 71 06 MASL mm mm *See notes at end of table. t The j and a fields together serve to specify any of the 128 control registers. 97 function Code (Octo I) f Mnemonic Instruction Description mm (R2), else repeat Execution Time* in flsec. With Overlap J Skip NI if (U) (R2) > (A) BIDl 3.5+ 1.5K 71 07 MASG Masked Alphanumeric Search Greater 71 10 DA Double Precision Fixed Point Add (A,A+l) + (U,U+l) A,A+l 3.167 71 11 DAN Double Precision Fixed Point Add Negative (A,A+l) - (U,U+ 1) --+ A,A+l 3.167 71 12 DS Double Store A (A,A+l) 3.0 71 13 DL Double Load A (U,U+l) --+ A,A+l 3.0 71 14 DLN Double Load Negative A -(U,U+ 1) --+ A,A+ 1) 3.0 71 15 DLM Double Load Magnitude A I(U,U+l)1 --+ A,A+l 3.0 71 16 DJZ Double Precision Zero Jump Jump to U if (A,A+ 1) = ±O; go to NI if (A,A+l) ~ ±O 3.167/1.667 71 17 DTE Double Precision Test Equal Skip NI if (U,U+l) = (A,A+l) 4.667/3.167 72 00 - Illegal Code Causes i lIega I instruct ion interrupt to address 2418 72 01 SLJ Store Location and Jump (P)-Base Address Modifier [BI or BD] --+ U17 - 0 ; jump to U+ 1 3.83 72 02 JPS Jump Positive and Shift Jump to U if (A)3S=0; go to NI if (A)35= 1; always shift (A) left circularly one bit position 3.0/1.5 72 03 JNS Jump Negative and Shift Jump to U if (A)35= 1; go to NI if (A)35=0; always shift (A) left circularly one bit position 3.0/1.5 72 04 AH Add Halves (A)35-18+ (U)35-18--+ A 35 - 18 ; (A)17-0+(U)17-0--+ A I7-0 1.5 72 05 ANH Add Negative Halves (A)35-18 -(U)35-18--+ A 35-18; 1.5 --+ --+ U,U+l - (A)17-o -(U)17-0--+ A 17-o 72 06 AT Add Thirds (A)35-24+ (U)35-24--+ A 35-24 (A)23-12+ (U)23-12--+ A 23-12 1.5 (A)1l-0+ (U)l1-o--+ A ll-o 72 07 ANT Add Negative Thirds (A)35-24 -(U)35-24--+ A35-24 (A)23-12 -(U)23-12--+ A 23-12 (A)l1-o - (U)11-0--+ A ll-o *See notes at end of table. 98 1.5 Function Code (Octo I) f Mnemonic Instruction Description Execution Time* in I1sec. With Overlap J 72 10 EX Execute Execute the instruction at U 1.5 72 11 ER Executive Return Interrupt to address 2428 2.33 72 12 - Illegal Code Causes illegal instruction interrupt to address 2418 - 72 13 PAIJ Prevent All I/O Interrupts and Jump Prevent all I/O interrupts and jump to U 1.5 72 14 SCN Store Channel Number If a=O: CHANNEL NUMBER ~ U3-0; If a= 1: CHANNEL NUMBER~ U3-0 and CPU NUMBER~ U5- 4 1.5 72 15 LPS Load Processor State Register (U)~ PSR 1.5 72 16 LSL Load Storage Limits Register (U) ~ SLR 1.5 72 17 Illegal Code Causes illegal instruction interrupt to address 2418 73 00 SSC Single Shift Circular Shift (A) right circularly U places 1.5 73 01 DSC Double Shift Circular Shift (A,A+ 1) right circularly U places 1.5 73 02 SSL Single Shift Logical Sh ift (A) right U places; zerofi II 1.5 73 03 DSL Double Shift Logical Shift (A,A+ 1) right U places; zerofi II 1.5 73 04 SSA Single Shift Algebraic Shift (A) right U places; signfill 1.5 73 05 DSA Double Shift Algebra ic Shift (A,A+l) right U places; signfill 1.666 73 06 LSC Load Shift and Count (U)~A; shift (A) left circularly unti I (A) 35 =i (A)34' NUMBER 2.0 - OF SHIFTS 73 07 DLSC Double Load Shift and Count ~ - A+l (U,U+ 1) ~ A,A+ 1; Shift (A,A+ 1) left circularly until (A,A+ 1)71 =i 3.830 (A,A+l)70; NUMBER OF SHIFTS ~ A+2 73 ~O LSSC Left Single Shift Circular Shift (A) left circularly U places 1.5 73 11 LDSC Left Double Shift Circular Shift (A,A+ 1) left circularly U places 1.666 *See notes at end of table. 99 Function Code (Octal) f Description Instruction Mnemonic Execution Time* in flsec. With Overlap . J 73 12 LSSL Left Single Shift Logical Shift (A) left U places; zerofill 1.5 73 13 LDSL Left Double Shift Logica I Shift (A,A+ 1) left U places; zerofill 1.666 73 14 III (a=O or 1) Initiate Interprocessor Interrupt Initiate Interprocessor Interrupt 1.5 ALRM (a= lOS) Alarm Turn on alarm 1.5 EDC (a= llS) Enable Day Clock Enable day clock 1.5 DOC (a= 12S) Disable Day Clock Disable day clock 1.5 Se lect Interrupt Location (A)2-0 ~ MSR 1.5 73 15 SIL 73 16 LCR (a=O) Load Channel Select Register (U)3-0~ CSR 1.666 LLA (a=l) Load Last Address Register (U)2-D ~ LAR 1.666 If (U)30= 1, interrupt to address 3.166/1.666 73 17 TS Test and Set 244S; if (U)30=0, go to NI; always 01 8 ~ U35-30 ; U29 -D unchanged 74 00 JZ Jump Zero Jump to U if (A)= to; go to NI if (A) i ±O 3.0/1.5 74 01 JNZ Jump Nonzero Jump to U if (A) (A) itO = to; go to Ni if 3.0/1.5 74 02 JP Jump Positive Jump to U if (A)35 (A)35 74 03 IN Jump Negative Jump to U if (A)35 (A)35 3.0/1.5 = 1; go to NI if 3.0/1.5 =0 74 04 JK J Jump Keys Jump Jump to U ifa=O or ifa=lit SELECT JUMP indicator; go to NI if neither is true 1.5 74 05 HKJ HJ Halt Keys and Jump Halt Jump Stop if a= 0 or if [a MfI I it SELECT STOP indicators] 0; on restart or continuation, jump to U 1.5 74 06 NOP No Operation Proceed to next instruction 1.5 *See notes at end of table. 100 = 0; go to NI if =1 -i Function Code (Octa I) f Mnemonic Instruction Description . Execution Time* in Ilsec. With Overlap J 74 07 AAIJ Allow All I/O Interrupts and Jump Allow all I/O interrupts and jump to U 1.5 74 10 JNB Jump No Low Bit Jump to U if (A)O = 0; go to NI if 3.0/1.5 (A)O= 1 74 11 JB Jump Low Bit Jump to U if (A)O = 1; go to NI if 3.0/1.5 (A)O=O 74 12 JMGI Jump Modifier Greater and Increment > 0; go to NI 17-0 if (Xa) < 0; always (X a) + 17-0 . 17-0 (X) ~X a 35-18 a17 3.166/1.5 Jump to U if (X a) -o 74 13 LMJ Load Modifier and Jump (P) - Base Address Modifier [BI or BO] ~ Xa ; jump to U 17-0 1.666 74 14 JO Jump Overflow Jump to U if 01 of PSR = 1; go to NlifOl=O 3.0/1.5 74 15 JNO Jump No Overflow Jump to U if 01 of PSR = 0; go to NI if 01 = 1 3.0/1.5 74 16 JC Jump Carry Jump to U if 00 of PSR= 1; go to NI if 00= 0 3.0/1.5 74 17 JNC Jump No Carry Jump to U if 00 of PSR = 0; go to NI if 00 = 1 3.0/1.5 75 00 LlC Load Input Channel For channel [a rm1 CSR]: (U) ~ IACR; set input active; clear input monitor 1.5 75 01 LlCM Load Input Channel and Monitor For channel [a mil CSR]: (U) ~ IACR; set input active; set input monitor 1.5 75 02 JIC Jump Input Channel Busy Jump to U if input active is set for channel [a ImJ CSR]; go to NI if input active is clear 1.5 75 03 OIC Disconnect Input Channel For channel [a ~ CSR]: clear input active; clear input monitor 1.5 75 04 LOC Load Output Channel For channel [a ' lmI CSR]: (U)-+, OACR; set output active; clear output monitor; clear external function (lSI only) 1.5 *See notes at end of table. 101 Function Code (Octal) f Mnemonic Description In struction j 75 05 LOCM Load Output Channe I and Monitor For channel [a (ill) CSR]; (U) ~ OACR; set output monitor; clear external function (lSI only) 1.5 75 06 JOC Jump Output Channel Busy Jump to U if output active is set for channel [a [i]3 CSR]; go to NI if output active is clear 1.5 75 07 DOC Disconnect Output Channel For channel [a CSR]: clear output active; clear output monitor; clear external function 1.5 75 10 LFC Load Function in Channel For channel [a [i]3 CSR]: (U) ~ OACR; set output active (lSI only), external function, and force external function; clear output monitor (lSI only) 1.5 75 11 LFCM Load Function in Channel and Monitor For channel [a l!m CSR]: (U) ~ OACR; set output active (lSI only), external function, force external function, and output monitor (lSI only) 1.5 75 12 JFC Jump Function in Channel Jump to U if force external function is set for channel [a [ill) CSR]; go to NI if force external function is clear 1.5 75 13 - Illegal Code If guard mode is set, causes guard mode interrupt to address 2438. If guard mode is not set, same as NOP 1.5 75 14 AACI Allow All Channel External Interrupts Allow all external interrupts 1.5 75 15 PACI Prevent All Channel External Interrupts Prevent all external interrupts 1.5 75 75 16 17 - Illegal COde} Illegal Code If guard mode is set, causes guard mode interrupt to address 2438. If guard mode is not set, same as NOP 1.5 76 00 FA Floating Add (A) + (U) ~ A; RESIDUE ~ A + 1 3.0 76 01 FAN Floating Add Negative (A) - (U) ~ A; RESIDUE ~ A + 1 3.0 76 02 FM Floating Multiply (A) • (U) 76 03 FD Floating Divide (A) -;- (U) ~ A; REMAINDER ~ A + 1 76 04 LUF Load and Unpack Floating I(U)134-2r~ A7- 0, zerofill; om ~ A, A + 1 (U)26-0 ~ A + 126- 0 , signfill *See notes at end of table. 102 Execution Time* in I1sec. With Overlap 4.0 11.5** 1.5 Function Code (Octa I) f Description In stru ction Mnemonic Execution Time* in flsec. With Overlap J 76 05 LCF Load and Convert To Floating (U~f5 4 A + 135 ; [NORMALIZED (U) 26-0 4 A + 126 - 0; if (U)35 = 0, (A)7-0 ± NORMALIZING COUNT 4 2.0 A + 134- 27 ; if (U)35 = 1, ones complement of [(A)7-0 ± NORMALIZING COUNT] 76 06 MCDU 76 07 CDU -4 A + 134- 27 Magnitude of Characteristic Difference To Upper II (A)35-27 I -I (U) 135-2714 A + 18- 0; ZEROS -4 A + 135- 9 1.5 1.5 Characteristic Difference To I(A)1 35- 27 -1(u)1 35- 27 4 A + 18- 0; Upper SIGN B'ITS 4 A + 135- 9 76 10 DFA Double Precision Floating Add (A, A + 1) + (U, U + 1) 4 A, A + 1 4.5 76 11 DFAN Double Precision Floating Add Negative (A, A+ 1) - (U, U+ 1)4 A, A+ 1 4.5 76 12 DFM Double Precision Floating Multiply (A, A+ 1)· (U, U+ 1) ---)A, A+ 1 6.667 76 13 DFD Double Precision Floating Divide (A, A + 1) + (U, U + 1) 4 A, A + 1 24.0*** 76 14 DFU Double Load and Unpack Floating I(u)134-24 4 AlO- o, zerofill; Double Load and Convert to Floating (U)35 4 A + 135 ; [NORMALIZED (U, U + 1)]59-0-4 A+ 123- 0 and 76 15 DFP 3.0 (U)23-0 -4 A + 123- 0, signfill; (lJ+1)4A+2 3.830 A + 2. If (U)35 = 0, (A)lO-O ± NORMALIZING COUNT 4 A + 134-24. If (U)35 = 1, ones complement of [(A)10-0 ± NORMALIZING COUNT] 4 A + 134- 24 76 16 FEL Floating Expand and Load If (U)35 = 0, (U)35-27 + 160°8 4 A35- 24 ; if (U)35 = I, 1.833 (U)35-27 -1600 8 .-4 A35- 24 ; (U)26-3 4 A23 - 0 ; (U)2-0 -4 A + 135- 33 ; (U)354 A A+1 32 ..() 76 17 FCL Floating Compress and Load If (U)35 = 0, (U)35-24 -16008 4 A35- 27 ; if (U)35 = 1, (U)35-24 3.167 + 16008 -4 A35 - 27 ; (U)23-0 4 A26 - 3; (U+ 1)35-33 4 A2-0 77 0-17 - Illegal Code Causes illegal instruction interrupt to address 2418 - *See notes at end of table. 103 NOTES: * Execution times given are cakulated using a core memory cycle time of 1.5 microseconds and a CPU clock cycle time of 166 nanoseconds. For all comparison instructions, the first number represents the skip or jump condition, the second number is for a no skip or no jump condition. For function codes 01 through 67, add .333 microseconds to execution times for 6-bit, 9-bit, and 12-bit writes. Execution time for the Block Transfer and the search instructions depends on the number of repetitions of the instruction required. The variance is 3.0K microseconds for Block Transfer and 1.5K microseconds for searches where K equals the number of repetitions; that is, K equals the number of words in the block being transferred or the number of words searched before a match is found. 104 ** If 28 instead of 27 subtractions are performed, add .333 microseconds. *** If 61 instead of 60 subtractions are performed, add .333 microseconds. APPENDIX D. MATH-PACK ROUTINES The following is a complete listing of STAT-PACK routines grouped according to function: INTERPOLATION GN INT - Gregory-Newton Interpo lation GNEXT - Gregory-Newton Extrapolation GNPOL - Gregory-Newton Polynomial Evaluation BESINT - Bessel Interpolation ST INT - Stirling Interpolation CDINT - Gauss Central-Difference Interpolation AITINT - Aitken Interpolation YLG INT - Lagrange Interpolation SPLN 1, SPLN2 - Spl ine Interpolation> NUMERICAL INTEGRATION TRAPNI - Trapezoidal Rule SIM1NI - Simpson 1/3 Rule SIM3NI - Simpson 3/8 Rule STEPNI - Variable Step Integration GENNI - Generalized Numerical Quadrature DOUBN I - Double Integration LGAUSS - Gauss Quadrature Abscissas and Weights SIMPTS - Simpson 1/3 Rule Abscissas and Weights 105 SOLUT ION OF EQUATIONS NEWTIT - Newton-Raphson Iteration WEGIT - Wegstein Iteration AITIT - Aitken Iteration ROOTCP - Real and Complex Roots of Real or Complex Polynomial DIF FERENTIATION DE R IVl - First Derivative Approximation DERIV2 - Second Derivative Approximation NTHDER - Nth Derivative of a Polynomial POL YNOMIAL MANIPULATION GIVZRS - Polynomial Coefficients Given its Zeros CVALUE - Complex Polynomial Evaluation POLY X - Real Polynomial Multiplication CPOL YX - Complex Polynomial Multiplication MATRIX MANIPULATION: REAL MATRICES MXADD - Matrix Addition MXSU B - Matrix Subtraction MXTRN - Matrix Transposition MXSCA - Matrix Multiplication by Scalar MXML T - Matrix Multiplication MXMDIG - Matrix Multiplication by Diagonal Matrix Stored as a Vector GJR - Determinant; Inverse; Solution of Simultaneous Equations MXHOI - Inverse Accuracy Improvement 106 MATRIX MANIPULATION: COMPLEX MATRICES CMXADD - Matrix Add ition CMXSU B - Matrix Subtraction CMXTRN - Matrix Transposition CMXSCA - Matrix Mu Itipl ication by Sca lar CMXML T - Matrix Mu Itipl ication CGJR - Determinant; Inverse; Solution of Simultaneous Equations MATRIX MANIPULATION EIGENVALUES AND EIGENVECTORS TR IDMX - Trid iagona I izati on of Rea I Symmetric Matrix E I GV AL - E igenva lues of T rid iagona I Matrix by Sturm Sequences EIGVEC - Eigenvectors of Tridiagonal Matrix by Wilkinson's Method MATRIX MANIPULATION: MISCELLANEOUS DGJ R - Doub le-Prec is ion Determinant; Inverse; Solution of Simultaneous Equations PMXTRI - Polynomial Matrix Triangularization SCALE - Polynomial Matrix Scaling MX ROT - Matrix Rotation ORDINARY DIFFERENTIAL EQUATIONS EULDE - Euler's Method HAMDE - Hamm ing' s Method INVAL - Initial Values for Differential Equation Solution RKDE - Runge-Kutta Method SODE - Second-Order Equations MRKDE - Reduction of Mth-Order System to System of m First-Order Equations 107 SYSTEMS OF EQUAT IONS JACMX - Jacobi Iteration to Determine Eigenvalues and Eigenvectors of Symmetric Matrix HJACMX - Jacobi Iteration to Determine Eigenvalues and Eigenvectors of Hermitian Matrix LSIMEQ Solution to a Set of Linear Simultaneous Equations NSIMEQ Functional Iteration to Determine Solution to Set of Non-Linear Equations CURVE F ITT ING CFSRI E - Coefficients of Fourier Series on a Continuous Range FTRANS - Fourier Transform DFSRIE - Coefficients of Fourier Series on Discrete Range FITD - Fitted Value and Derivative Values for a Least-Squares Polynomial ORTHLS - Orthogonal Polynomial Least-Squares Curve-Fitting FITY - Fitted Values for a Least-Squares Polynomial COEFS - Coefficients of a Least-Squares Polynomial PSUEDO-RANDOM NUMBER GENERATORS NRAND - Interval (0,227) Generator RANDU - Uniform Distribution RANDN - Normal Distribution RANDEX - Exponential Distribution SPECIFIC FUNCTIONS BSSL - Zero- and First-Order Bessel Functions BESJ - Regular Bessel Functions of Real Argument BESY - Irregular Bessel Functions of Real Argument BESI - Regular Bessel Functions of Imaginary Argument BESK - Irregular Bessel Functions of Imaginary Argument GAMMA - Gamma Function Evaluation LEGEN - Legendre Polynomial Evaluation ARCTNQ - Arctangent of a Quotient 108 APPENDIX E. STAT-PACK ROUTINES The following is a complete listing of STAT-PACK routines grouped according to function: DESCRIPT IVE STATISTICS FREQP - Frequency Polygon HIST - Histogram MH 1ST - Mu Itivariate Histogram GROUP - Grouping of Data ELEMENTARY POPULATION STATISTICS AMEAN - Arithmetic Mean GMEAN - Geometric Mean HMEAN - Harmon ic Mean MEDIAN - Median MODE - Mode QUANT - Quantiles OGIVE - Distribution Curve IQRNG - I nterpercent i Ie Range RANGE - Range MNDEV - Mean Deviation STDEV - Standard Deviation CVAR - Coefficient of Variation ORDER - Order and Rank Statistics CMONT - Centra I Moments AMONT - Absolute Moments CUML T - Cumulants SH PCOR - Sheppard's Correction s KURSK - Skewness and Kurtosis 109 DISTRIBUTION, FITTING AND PLOTTING BINOM - Binomial Distribution POISON - Poisson Distribution HYPER - Hypergeometric Distribution PNORM - Norma I Di stribution AFSER - Arne Fisher Series CHI-SQUARE TESTS CHI21S - CHI-Square Test of Sample ProportionOne Sample CHI2JS - CHI-Square Test of Sample Proportion J Samples CHI2P - CHI-Square Test of Fit to Poisson Distribution CH 12N - CH I-Square Test of Norma I ity CHISAM - CHI-Square Test of Homogeniety CH ICNT - CH I-Square Test for Independence GENGOF - CH I-Square Test of Genera I Goodness of Fit SIGNIFICANCE TESTS SIGPRP - Test of Significance of Proportion of Successes SIGMN - Test of Significance of a Mean SIGDMN - Test of Significance of the Difference Between Two Means SIGDVR - Test of Sign ificance of the Ratio Between Two Variances 110 . CONFIDENCE INTERVALS CFDMKV - Confidence Interval for the Mean; Known Variance CFDMUV - Confidence Interva I for the Mean; Unknown Variance C F DMSU - Confidence Interva I for, the 0 ifference Between Two Means CFDVAR - Confidence Interval for Variance TOLINT - Tolerance Intervals ANALYSIS OF VARIANCE ANOVl - One-Way Cross Classification ANOV2 - Two-Way Cross Classification ANOV3 - Three-Way Cross Classification MISDAT - Missing Data VTRANS - Variable Transformations ANOVRB - Randomized Blocks ANOVLS - Latin Squares ANOVSP - Split Plot Design ANOSSP - Split-Split Plot Design ANOVN2 - Two-Way Nested Des ign ANOVN3 - Three-Way Nested Des ign ANOCO - Ana Iys i s of Covariance GLH - General Linear Hypotheses REGRESSION ANAL YSIS RESTEM - Stepwise Multiple Regression REBSOM - Back Solution Multiple Regression CORAN - Correlation Analysis 111 TIME SERIES ANALYSIS MOVAVG - Moving Averages SE ASH I - Sh i sk in's Sea sona lity Factors WEMAV - Weighted Moving Averages TRELS - Trend Analysis by Least Squares VADIME - Variate Difference Method TSFARG - Autoregressive Model GEXSMO - Generalized Exponential Smoothing AUXCOR - Auto-Correlation and Cross-Correlation Analysis POWDEN - Power Density Functions RCPROB - Residual Probabilities MUL TI-VARIATE ANAL YSIS GENVAR - Generalized Variance DISHOT - Hotelling's Distribution DSQ - Mahalanobois' Distribution SIGTMN - Significance of a Set of Means DISCRA - Discriminant Analysis FACTAN - Factor and Principal Component Analysis DISTRIBUTION FUNCTIONS RNORM - Normal Distribution CHI - CHI-Square Distribution STUD - Student's Distribution FISH - Fisher's Distribution POlS - Poisson Distribution BIN - Binominal Distribution HYGEO - Hypergeometric Distribution GAMIN - Incomplete Gamma Distribution BETINe - Incomplete BETA Distribution 112 INVERSE DISTRI BUTION FUNCTIONS TINORM - Inverse Normal Distribution STUDIN - Inverse Student's Distribution FISHIN - Inverse Fisher's Distribution CHIN - Inverse CHI-Square Distribution MISCELLANEOUS rLOTl - Plot of One Line JIM - Matrix Inversion MXTML T - Left Multiplication of a Matrix by its Transpose 113
Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.3 Linearized : No XMP Toolkit : Adobe XMP Core 4.2.1-c043 52.372728, 2009/01/18-15:56:37 Producer : Adobe Acrobat 9.13 Paper Capture Plug-in Modify Date : 2009:10:04 10:34:44-07:00 Create Date : 2009:10:04 10:34:44-07:00 Metadata Date : 2009:10:04 10:34:44-07:00 Format : application/pdf Document ID : uuid:6b1a7b51-c083-40b5-a735-a4ba3881e3e0 Instance ID : uuid:a57a2571-ed24-485a-a2cb-646c1c3bda45 Page Layout : SinglePage Page Mode : UseOutlines Page Count : 122EXIF Metadata provided by EXIF.tools