Comparison of instruction set architectures

An instruction set architecture (ISA) is an abstract model of a computer, also referred to as computer architecture. A realization of an ISA is called an implementation. An ISA permits multiple implementations that may vary in performance, physical size, and monetary cost (among other things); because the ISA serves as the interface between software and hardware. Software that has been written for an ISA can run on different implementations of the same ISA. This has enabled binary compatibility between different generations of computers to be easily achieved, and the development of computer families. Both of these developments have helped to lower the cost of computers and to increase their applicability. For these reasons, the ISA is one of the most important abstractions in computing today.

An ISA defines everything a machine language programmer needs to know in order to program a computer. What an ISA defines differs between ISAs; in general, ISAs define the supported data types, what state there is (such as the main memory and registers) and their semantics (such as the memory consistency and addressing modes), the instruction set (the set of machine instructions that comprises a computer's machine language), and the input/output model.

Data representation

edit

In the early decades of computing, there were computers that used binary, decimal[1] and even ternary.[2][3] Contemporary computers are almost exclusively binary.

Characters are encoded as strings of bits or digits, using a wide variety of character sets; even within a single manufacturer there were character set differences.

Integers are encoded with a variety of representations, including Sign-magnitude, Ones' complement, Two's complement, Offset binary, Nines' complement and Ten's complement.

Similarly, floating point numbers are encoded with a variety of representations for the sign, exponent and mantissa. In contemporary machines IBM hexadecimal floating-point and IEEE 754 floating point have largely supplanted older formats.

Addresses are typically unsigned integers generated from a combination of fields in an instruction, data from registers and data from storage; the details vary depending on the architecture.

Bits

edit

Computer architectures are often described as n-bit architectures. In the first 34 of the 20th century, n is often 12, 18, 24, 30, 36, 48 or 60. In the last 13 of the 20th century, n is often 8, 16, or 32, and in the 21st century, n is often 16, 32 or 64, but other sizes have been used (including 6, 39, 128). This is actually a simplification as computer architecture often has a few more or less "natural" data sizes in the instruction set, but the hardware implementation of these may be very different. Many instruction set architectures have instructions that, on some implementations of that instruction set architecture, operate on half and/or twice the size of the processor's major internal datapaths. Examples of this are the Z80, MC68000, and the IBM System/360. On these types of implementations, a twice as wide operation typically also takes around twice as many clock cycles (which is not the case on high performance implementations). On the 68000, for instance, this means 8 instead of 4 clock ticks, and this particular chip may be described as a 32-bit architecture with a 16-bit implementation. The IBM System/360 instruction set architecture is 32-bit, but several models of the System/360 series, such as the IBM System/360 Model 30, have smaller internal data paths, while others, such as the 360/195, have larger internal data paths. The external databus width is not used to determine the width of the architecture; the NS32008, NS32016 and NS32032 were basically the same 32-bit chip with different external data buses; the NS32764 had a 64-bit bus, and used 32-bit register. Early 32-bit microprocessors often had a 24-bit address, as did the System/360 processors.

Digits

edit

In the first 34 of the 20th century, word oriented decimal computers typically had 10 digit[4][5][6] words with a separate sign, using all ten digits in integers and using two digits for exponents[7][5] in floating point numbers.

Endianness

edit

An architecture may use "big" or "little" endianness, or both, or be configurable to use either. Little-endian processors order bytes in memory with the least significant byte of a multi-byte value in the lowest-numbered memory location. Big-endian architectures instead arrange bytes with the most significant byte at the lowest-numbered address. The x86 architecture as well as several 8-bit architectures are little-endian. Most RISC architectures (SPARC, Power, PowerPC, MIPS) were originally big-endian (ARM was little-endian), but many (including ARM) are now configurable as either.

Endianness only applies to processors that allow individual addressing of units of data (such as bytes) that are smaller than some of the data formats.

Instruction formats

edit

Opcodes

edit

In some architectures, an instruction has a single opcode. In others, some instructions have an opcode and one or more modifiers. E.g., on the IBM System/370, byte 0 is the opcode but when byte 0 is a B216 then byte 1 selects a specific instruction, e.g., B20516 is store clock (STCK).

Operands

edit

Addressing modes

edit

Architectures typically allow instructions to include some combination of operand addressing modes

Direct
The instruction specifies a complete (virtual) address
Immediate
The instruction specifies a value rather than an address
Indexed
The instruction specifies a register to use as an index. In some architecture the index is scaled by the operand length.
Indirect
The instruction specifies the location of a pointer word that describes the operand, possibly involving multiple levels of indexing and indirection.
Truncated
Base-displacement
The instruction specifies a displacement from an address in a register
autoincrement/autodecrement
A register used for indexing, or a pointer word used by indirect addressing, is incremented or decremented by 1, an operand size or an explicit delta

Number of operands

edit

The number of operands is one of the factors that may give an indication about the performance of the instruction set. A three-operand architecture (2-in, 1-out) will allow

A := B + C

to be computed in one instruction

ADD B, C, A

A two-operand architecture (1-in, 1-in-and-out) will allow

A := A + B

to be computed in one instruction

ADD B, A

but requires that

A := B + C

be done in two instructions

MOVE B, A
ADD C, A

Encoding length

edit

As can be seen in the table below some instructions sets keep to a very simple fixed encoding length, and other have variable-length. Usually it is RISC architectures that have fixed encoding length and CISC architectures that have variable length, but not always.

Instruction sets

edit

The table below compares basic information about instruction set architectures.

Notes:

  • Usually the number of registers is a power of two, e.g. 8, 16, 32. In some cases a hardwired-to-zero pseudo-register is included, as "part" of register files of architectures, mostly to simplify indexing modes. The column "Registers" only counts the integer "registers" usable by general instructions at any moment. Architectures always include special-purpose registers such as the program counter (PC). Those are not counted unless mentioned. Note that some architectures, such as SPARC, have register windows; for those architectures, the count indicates how many registers are available within a register window. Also, non-architected registers for register renaming are not counted.
  • In the "Type" column, "Register–Register" is a synonym for a common type of architecture, "load–store", meaning that no instruction can directly access memory except some special ones, i.e. load to or store from register(s), with the possible exceptions of memory locking instructions for atomic operations.
  • In the "Endianness" column, "Bi" means that the endianness is configurable.
Archi-
tecture
Bits Version Intro-
duced
Max #
operands
Type Design Registers
(excluding FP/vector)
Instruction encoding Branch evaluation Endian-
ness
Extensions Open Royalty
free
6502 8 1975 1 Register–Memory CISC 3 Variable (8- to 24-bit) Condition register Little
6800 8 1974 1 Register–Memory CISC 3 Variable (8- to 24-bit) Condition register Big
6809 8 1978 1 Register–Memory CISC 5 Variable (8- to 32-bit) Condition register Big
680x0 32 1979 2 Register–Memory CISC 8 data and 8 address Variable Condition register Big
8080 8 1974 2 Register–Memory CISC 7 Variable (8 to 24 bits) Condition register Little
8051 32 (8→32) 1977? 1 Register–Register CISC
  • 32 in 4-bit
  • 16 in 8-bit
  • 8 in 16-bit
  • 4 in 32-bit
Variable (8 to 24 bits) Compare and branch Little
x86 16, 32, 64
(16→32→64)
v4 (x86-64) 1978 2 (integer)
3 (AVX)[a]
4 (FMA4 and VPBLENDVPx)[8]
Register–Memory CISC
  • 8 (+ 4 or 6 segment reg.) (16/32-bit)
  • 16 (+ 2 segment reg. gs/cs) (64-bit)
  • 32 with AVX-512
Variable (8086 ~ 80386: variable between 1 and 6 bytes /w MMU + intel SDK, 80486: 2 to 5 bytes with prefix, pentium and onward: 2 to 4 bytes with prefix, x64: 4 bytes prefix, third party x86 emulation: 1 to 15 bytes w/o prefix & MMU . SSE/MMX: 4 bytes /w prefix AVX: 8 Bytes /w prefix) Condition code Little x87, IA-32, MMX, 3DNow!, SSE,
SSE2, PAE, x86-64, SSE3, SSSE3, SSE4,
BMI, AVX, AES, FMA, XOP, F16C
No No
Alpha 64 1992 3 Register–Register RISC 32 (including "zero") Fixed (32-bit) Condition register Bi MVI, BWX, FIX, CIX No
ARC 16/32/64 (32→64) ARCv3[9] 1996 3 Register–Register RISC 16 or 32 including SP
user can increase to 60
Variable (16- or 32-bit) Compare and branch Bi APEX User-defined instructions
ARM/A32 32 ARMv1–v9 1983 3 Register–Register RISC
  • 15
Fixed (32-bit) Condition code Bi NEON, Jazelle, VFP,
TrustZone, LPAE
No
Thumb/T32 32 ARMv4T-ARMv8 1994 3 Register–Register RISC
  • 7 with 16-bit Thumb instructions
  • 15 with 32-bit Thumb-2 instructions
Thumb: Fixed (16-bit), Thumb-2:
Variable (16- or 32-bit)
Condition code Bi NEON, Jazelle, VFP,
TrustZone, LPAE
No
Arm64/A64 64 v8.9-A/v9.4-A,[10] Armv8-R[11] 2011[12] 3 Register–Register RISC 32 (including the stack pointer/"zero" register) Fixed (32-bit), Variable (32-bit or 64-bit for FMA4 with 32-bit prefix[13]) Condition code Bi SVE and SVE2 No
AVR 8 1997 2 Register–Register RISC 32
16 on "reduced architecture"
Variable (mostly 16-bit, four instructions are 32-bit) Condition register,
skip conditioned
on an I/O or
general purpose
register bit,
compare and skip
Little
AVR32 32 Rev 2 2006 2–3 RISC 15 Variable[14] Big Java virtual machine
Blackfin 32 2000 3[15] Register–Register RISC[16] 2 accumulators

8 data registers

8 pointer registers

4 index registers

4 buffer registers

Variable (16- or 32-bit) Condition code Little[17]
CDC Upper 3000 series 48 1963 3 Register–Memory CISC 48-bit A reg., 48-bit Q reg., 6 15-bit B registers, miscellaneous Variable (24- or 48-bit) Multiple types of jump and skip Big
CDC 6000
Central Processor (CP)
60 1964 3 Register–Register n/a[b] 24 (8 18-bit address reg.,
8 18-bit index reg.,
8 60-bit operand reg.)
Variable (15-, 30-, or 60-bit) Compare and branch n/a[c] Compare/Move Unit No No
CDC 6000
Peripheral Processor (PP)
12 1964 1 or 2 Register–Memory CISC 1 18-bit A register, locations 1–63 serve as index registers for some instructions Variable (12- or 24-bit) Test A register, test channel n/a[d] additional Peripheral Processing Units No No
Crusoe
(native VLIW)
32[18] 2000 1 Register–Register VLIW[18][19]
  • 1 in native push stack mode
  • 6 in x86 emulation +
    8 in x87/MMX mode +
    50 in rename status
  • 12 integer + 48 shadow +
    4 debug in native VLIW
  • mode[18][19]
Variable (64- or 128-bit in native mode, 15 bytes in x86 emulation)[19] Condition code[18] Little
Elbrus 2000
(native VLIW)
64 v6 2007 1 Register–Register[18] VLIW 8–64 64 Condition code Little Just-in-time dynamic translation: x87, IA-32, MMX, SSE,
SSE2, x86-64, SSE3, AVX
No No
DLX 32 ? 1990 3 ? RISC 32 Fixed (32-bit) ? Big ? Yes ?
eSi-RISC 16/32 2009 3 Register–Register RISC 8–72 Variable (16- or 32-bit) Compare and branch
and condition register
Bi User-defined instructions No No
iAPX 432[20] 32 1981 3 Stack machine CISC 0 Variable (6 to 321 bits) No No
Itanium
(IA-64)
64 2001 Register–Register EPIC 128 Fixed (128-bit bundles with 5-bit template tag and 3 instructions, each 41-bit long) Condition register Bi
(selectable)
Intel Virtualization Technology No No
LoongArch 32, 64 2021 4 Register–Register RISC 32 (including "zero") Fixed (32-bit) Little No No
M32R 32 1997 3 Register–Register RISC 16 Variable (16- or 32-bit) Condition register Bi
m88k 32 1988 3 Register–Register RISC Fixed (32-bit) Big
Mico32 32 ? 2006 3 Register–Register RISC 32[21] Fixed (32-bit) Compare and branch Big User-defined instructions Yes[22] Yes
MIPS 64 (32→64) 6[23][24] 1981 1–3 Register–Register RISC 4–32 (including "zero") Fixed (32-bit) Condition register Bi MDMX, MIPS-3D No No[25][26]
MMIX 64 ? 1999 3 Register–Register RISC 256 Fixed (32-bit) Condition register Big ? Yes Yes
Nios II 32 ? 2000 3 Register–Register RISC 32 Fixed (32-bit) Condition register Little Soft processor that can be instantiated on an Altera FPGA device No On Altera/Intel FPGA only
NS320xx 32 1982 5 Memory–Memory CISC 8 Variable Huffman coded, up to 23 bytes long Condition code Little BitBlt instructions
OpenRISC 32, 64 1.4[27] 2000 3 Register–Register RISC 16 or 32 Fixed Condition code Bi ? Yes Yes
PA-RISC
(HP/PA)
64 (32→64) 2.0 1986 3 Register–Register RISC 32 Fixed (32-bit) Compare and branch Big → Bi MAX No
PDP-5[28]
PDP-8[29]
12 1963 Register–Memory CISC 1 accumulator

1 multiplier quotient register

Fixed (12-bit) Condition register

Test and branch

EAE (Extended Arithmetic Element)
PDP-11 16 1970 2 Memory–Memory CISC 8 (includes program counter and stack pointer, though any register can act as stack pointer) Variable (16-, 32-, or 48-bit) Condition code Little Extended Instruction Set, Floating Instruction Set, Floating Point Processor, Commercial Instruction Set No No
POWER, PowerPC, Power ISA 32/64 (32→64) 3.1[30] 1990 3 (mostly). FMA, LD/ST-Update Register–Register RISC 32 GPR, 8 4-bit Condition Fields, Link Register, Counter Register Fixed (32-bit), Variable (32- or 64-bit with the 32-bit prefix[30]) Condition code, Branch-Counter auto-decrement Bi AltiVec, APU, VSX, Cell, Floating-point, Matrix Multiply Assist Yes Yes
RISC-V 32, 64, 128 20191213[31] 2010 3 Register–Register RISC 32 (including "zero") Variable Compare and branch Little ? Yes Yes
RX 64/32/16 2000 3 Memory–Memory CISC 4 integer + 4 address Variable Compare and branch Little No
S+core 16/32 2005 RISC Little
SPARC 64 (32→64) OSA2017[32] 1985 3 Register–Register RISC 32 (including "zero") Fixed (32-bit) Condition code Big → Bi VIS Yes Yes[33]
SuperH (SH) 32 ? 1994 2 Register–Register
Register–Memory
RISC 16 Fixed (16- or 32-bit), Variable Condition code
(single bit)
Bi ? Yes Yes
System/360
System/370
System/390
z/Architecture
64 (32→64) 1964 2 (most)
3 (FMA, distinct
operand facility)

4 (some vector inst.)
Register–Memory
Memory–Memory
Register–Register
CISC 16 general
16 control (S/370 and later)
16 access (ESA/370 and later)
Variable (16-, 32-, or 48-bit) Condition code, compare and branch auto increment, Branch-Counter auto-decrement Big No No
TMS320 C6000 series 32 1983 3 Register-Register VLIW 32 on C67x
64 on C67x+
Fixed (256-bit bundles with 8 instructions, each 32-bit long) Condition register Bi No No
Transputer 32 (4→64) 1987 1 Stack machine MISC 3 (as stack) Fixed (8-bit) Compare and branch Little
VAX 32 1977 6 Memory–Memory CISC 16 Variable Condition code, compare and branch Little No
Z80 8 1976 2 Register–Memory CISC 17 Variable (8 to 32 bits) Condition register Little
Archi-
tecture
Bits Version Intro-
duced
Max #
operands
Type Design Registers
(excluding FP/vector)
Instruction encoding Branch evaluation Endian-
ness
Extensions Open Royalty
free

See also

edit

Notes

edit
  1. ^ The LEA (all processors) and IMUL-immediate (80186 & later) instructions accept three operands; most other instructions of the base integer ISA accept no more than two operands.
  2. ^ partly RISC: load/store architecture and simple addressing modes, partly CISC: three instruction lengths and no single instruction timing
  3. ^ Since memory is an array of 60-bit words with no means to access sub-units, big endian vs. little endian makes no sense. The optional CMU unit uses big-endian semantics.
  4. ^ Since memory is an array of 12-bit words with no means to access sub-units, big endian vs. little endian makes no sense.

References

edit
  1. ^ da Cruz, Frank (October 18, 2004). "The IBM Naval Ordnance Research Calculator". Columbia University Computing History. Retrieved May 8, 2024.
  2. ^ "Russian Virtual Computer Museum _ Hall of Fame _ Nikolay Petrovich Brusentsov".
  3. ^ Trogemann, Georg; Nitussov, Alexander Y.; Ernst, Wolfgang (2001). Computing in Russia: the history of computer devices and information technology revealed. Vieweg+Teubner Verlag. pp. 19, 55, 57, 91, 104–107. ISBN 978-3-528-05757-2..
  4. ^ 650 magnetic drum data processing machine (PDF). IBM. June 1955. 22-6060-2. Retrieved May 8, 2024.
  5. ^ a b IBM 7070-7074 Principles of Operation (PDF). Systems Reference Library. IBM. 1962. GA22-7003-6. Retrieved May 8, 2024.
  6. ^ UNIVAC® Solid-state 80 Computer (PDF). Sperry Rand Corporation. 1959. U1742.1r3. Retrieved May 8, 2024.
  7. ^ IBM 650 MDDPM Additional Features - Indexing Accumulators - Floating-Decimal Arithmetic - Advanced Write-Up (PDF). IBM. 1955. 22-6258-0. Retrieved May 8, 2024.
  8. ^ "AMD64 Architecture Programmer's Manual Volume 6: 128-Bit and 256-Bit XOP and FMA4 Instructions" (PDF). AMD. November 2009.
  9. ^ "Synopsys Introduces New 64-bit ARC Processor IP Delivering up to 3x Performance Increase for High-End Embedded Applications".
  10. ^ "Arm A-Profile Architecture Developments 2022 - Architectures and Processors blog - Arm Community blogs - Arm Community". community.arm.com. 29 September 2022. Retrieved 2022-12-09.
  11. ^ Frumusanu, Andrei (September 3, 2020). "ARM Announced Cortex-R82: First 64-bit Real Time Processor". AnandTech.
  12. ^ "ARM goes 64-bit with new ARMv8 chip architecture". Computerworld. 27 October 2011. Retrieved 8 May 2024.
  13. ^ Toshio Yoshida. "Hot Chips 30 conference; Fujitsu briefing" (PDF). Fujitsu. Archived from the original (PDF) on 2020-12-05.
  14. ^ "AVR32 Architecture Document" (PDF). Atmel. Retrieved 2024-05-08.
  15. ^ "Blackfin manual" (PDF). analog.com.
  16. ^ "Blackfin Processor Architecture Overview". Analog Devices. Retrieved 2024-05-08.
  17. ^ "Blackfin memory architecture". Analog Devices. Archived from the original on 2011-06-16. Retrieved 2009-12-18.
  18. ^ a b c d e "Crusoe Exposed: Transmeta TM5xxx Architecture 2". Real World Technologies.
  19. ^ a b c Alexander Klaiber (January 2000). "The Technology Behind Crusoe Processors" (PDF). Transmeta Corporation. Retrieved December 6, 2013.
  20. ^ Intel Corporation (1981). Introduction to the iAPX 432 Architecture (PDF). pp. iii.
  21. ^ "LatticeMico32 Architecture". Lattice Semiconductor. Archived from the original on 23 June 2010.
  22. ^ "LatticeMico32 Open Source Licensing". Lattice Semiconductor. Archived from the original on 20 June 2010.
  23. ^ MIPS64 Architecture for Programmers: Release 6
  24. ^ MIPS32 Architecture for Programmers: Release 6
  25. ^ MIPS Open
  26. ^ "Wave Computing Closes Its MIPS Open Initiative with Immediate Effect, Zero Warning".
  27. ^ OpenRISC Architecture Revisions
  28. ^ PDP-5 Handbook (PDF). Digital Equipment Corporation. February 1964.
  29. ^ PDP-8 Users Handbook (PDF). Digital Equipment Corporation. May 1966.
  30. ^ a b "Power ISA Version 3.1". openpowerfoundation.org. 2020-05-01. Retrieved 2021-10-20.
  31. ^ "RISC-V ISA Specifications". Retrieved 17 June 2019.
  32. ^ Oracle SPARC Processor Documentation
  33. ^ SPARC Architecture License