Intel microcode

On the Intel 80486 and AMD Am486 there are approximately 5000 lines of microcode assembly, totalling approximately 240 Kbits stored in the microcode ROM.{{cite report|url=https://ir.amd.com/sec-filings/content/0000898430-94-000804/EX-99_1.txt|first=Patricia V.|last=Trumbull|date=1994-10-07|access-date=2021-05-10|series=United States District Court for the Northern District of California|location=San Jose|via=Advanced Micro Devices|title=Intel Corporation v. Advanced Micro Devices|number=C-93-20301 PVT|type=Findings of fact and conclusions of law following "ICE" module of trial|quote=Twelve pins are affiliated with the "ICE" circuitry. … AMD 486DXL and DXLV connect three pins associated with "ICE" in order to implement its "SMM" feature. … 250 lines or 12,032 bits of the "ICE" microcode in the 486. "ICE" constitutes about five percent of the total 486 microcode. … two lines … (used to set the "ICE" mode "flip flop") … blue coded lines of microcode are associated with production testing and not used for "ICE" related purposes. … Seventy-five red coded lines were used by Intel to perform "SMM" in its 486SL, a data sheet function of this version of the chip. About 32 yellow coded lines perform routine operations which are not unique to "ICE." About two lines remain dedicated solely to "ICE."}}

{{Main|Micro-operation}}

Starting with the Pentium Pro, in most Intel x86 processors, instructions are converted by the instruction fetch and decode unit to sequences of processor-specific micro-operations that are directly executed by the processor. For the instructions that are implemented in microcode, the microcode consists of micro-operations fetched from on-chip memory.{{cite web|url=http://www.intel.com/procs/ppro/info/p6white/index.htm|title=A Tour of the Pentium Pro Processor Microarchitecture|website=Intel|archive-url=https://web.archive.org/web/19961220080210/http://www.intel.com/procs/ppro/info/p6white/index.htm|archive-date=1996-12-20|url-status=dead}}

On the Pentium Pro, each micro-operation is 72-bits wide,{{cite journal|url=https://people.eecs.berkeley.edu/~kubitron/courses/cs152-S04/lectures/lec25-power.pdf|title=Dynamic Scheduling in P6 (Pentium Pro, II, III)|journal=Low Power Design, Advanced Intel Processors|series=CS152 Computer Architecture and Engineering|issue=Lecture 25|date=3 May 2004|first=John|last=Kubiatowicz|quote=Complex 80x86 instructions are executed by a conventional microprogram ({{abbr|8K|8192}} x 72 bits) that issues long sequences of micro-operations}}{{rp|43}} or 118-bits wide.{{cite news|url=https://pdfs.semanticscholar.org/fe2b/b73d7046a6ed87ce9b18d62f194d67fa2100.pdf|archive-url=https://web.archive.org/web/20181008134943/https://pdfs.semanticscholar.org/fe2b/b73d7046a6ed87ce9b18d62f194d67fa2100.pdf|url-status=dead|archive-date=8 October 2018|first=Linley|last=Gwennap|date=16 February 1995|work=Microprocessor Report|publisher=MicroDesign Resources|volume=9|number=2|pages=1–7|title=Intel's P6 Uses Decoupled Superscalar Design|s2cid=14414612|quote=P6 uops have a fixed length of 118 bits, using a regular structure to encode an operation, two sources, and a destination. The source and destination fields are each wide enough to contain a 32-bit operand.}}{{rp|2}}{{rp|14}} This includes an opcode, two source fields, and one destination field,{{cite web|url=http://datasheets.chipdb.org/Intel/x86/P6/p6updt.pdf|title=A 0.6 μm BiCMOS Processor With Dynamic Execution|first1=Robert P.|last1=Colwell|first2=Randy L.|last2=Steck|author3=Intel Corporation|date=1995-04-12|access-date=2020-05-27|page=7|quote=Micro-ops are the atomic unit of work in the P6 processor and are {{sic|comprised |hide=y|of}} an opcode, two source and one destination operand. These micro-ops are fixed length and are more general than the Pentium(R) processor's microcode since they need to be scheduled.}}{{rp|7}} with the ability to hold a 32-bit immediate value.{{rp|14}} The Pentium Pro is able to detect parity errors in its internal microcode {{abbr|ROM|Read-Only Memory}} and report these via the Machine Check Architecture.{{cite report|url-status=unfit|url=x|archive-url=http://folk.uio.no/inf242/doc/242692_1.pdf|archive-date=6 September 2001|title=16.6.1. Simple Error Codes|page=401|date=3 January 1996|access-date=1 October 2018|work=Machine Check Architecture|volume=3: Operating System Writer's Guide|series=Pentium® Pro Family Developer's Manual|quote=unique codes indicate global error information … Microcode ROM Parity Error|issue=December 1995}}

Micro-operations have a consistent format with up to three source inputs, and two destination outputs. The processor performs register renaming to map these inputs to and from the real register file (RRF) before and after their execution. Out-of-order execution is used, so the micro-operations and instructions they represent may not appear in the same order.

During development of the Pentium Pro, several microcode fixes were included between the A2 and B0 steppings.{{cite news|url=http://web.cecs.pdx.edu/~berkina/R10_papworth_ieeemicro_1996.pdf|page=14|work=IEEE Micro|title=Tuning the Pentium Pro Microarchitecture|first=David B.|last=Papworth|author2=Intel Corporation|issn=0272-1732|date=April 1996|access-date=8 October 2018|quote=B0 stepping incorporated several microcode bugs and speed path fixes for problems discovered on the A-step silicon|archive-date=8 October 2018|archive-url=https://web.archive.org/web/20181008095801/http://web.cecs.pdx.edu/~berkina/R10_papworth_ieeemicro_1996.pdf|url-status=dead}} For the Pentium II (based on the P6 Pentium Pro), additional micro-operations were added to support the MMX instruction set. In several cases, "microcode assists" were added to handle rare corner-cases in a reliable way.

The Pentium 4 can have 126 micro-operations in flight at the same time.{{rp|10}} Micro-operations are decoded and stored in an Execution Trace Cache with 12,000 entries, to avoid repeated decoding of the same x86 instructions.{{rp|5}} Groups of six micro-operations are packed into a trace line.{{rp|5}} Micro-operations can borrow extra immediate data space within the same cache-line.{{cite web|url=https://www.agner.org/optimize/microarchitecture.pdf|title=The microarchitecture of Intel, AMD and VIA CPUs|type=An optimization guide for assembly programmers and compiler makers|first=Agner|last=Fog|publisher=Technical University of Denmark|date=2020-05-25|page=49|quote=… If a μop has an immediate 32-bit operand outside the ±2¹⁵ interval so that it cannot be represented as a 16-bit signed integer, then it will use two trace cache entries unless it can borrow storage space from a nearby μop. … A μop in need of extra storage space can borrow 16 bits of extra storage space from a nearby μop that doesn't need its own data space.}}{{rp|49}} Complex instructions, such as exception handling, result in jumping to the microcode ROM.{{rp|6}} During development of the Pentium 4, microcode accounted for 14% of processor bugs versus 30% of processor bugs during development of the Pentium Pro.{{cite journal|url=https://www.intel.com/content/dam/www/public/us/en/documents/research/2001-vol05-iss-1-intel-technology-journal.pdf|quote=Bug Discussion|title=Validating The Intel® Pentium® 4 Processor|first1=Bob|last1=Bentley|first2=Rand|last2=Gray|pages=29–26|journal=Intel Technology Journal|issue=Q1|year=2001|editor1-first=Lin|editor1-last=Chao}}{{rp|35}}

The Intel Core microarchitecture introduced in 2006 added "macro-operations fusion" for some common pairs of instructions including comparison followed by a jump. The instruction decoders in the Core convert x86 instructions into microcode in three different ways:

For Intel's hyper-threading implementation of simultaneous multithreading, the microcode ROM, trace cache, and instruction decoders are shared, but the micro-operation queue is not shared.

In the mid-1990s, a facility for supplying new microcode was initially referred to as the Pentium Pro BIOS Update Feature.{{cite report|url=http://datasheets.chipdb.org/Intel/x86/Pentium%20Pro/PPPBIOS.PDF|title=8: Pentium Pro Processor BIOS Update Feature|version=2.0|date=12 January 1996|publisher=Intel |access-date=3 November 2020|page=45|quote=authentication procedure relies upon the decryption provided by the processor to verify an update from a potentially hostile sources.}} It was intended that user-mode applications should make a BIOS interrupt call to supply a new "BIOS Update Data Block", which the BIOS would partially validate and save to nonvolatile BIOS memory; this could be supplied to the installed processors on next boot.

Intel distributed a program called BUP_UTIL.EXE, renamed CHECKUP3.EXE that could be run under DOS. Collections of multiple microcode updates were concatenated together and numerically numbered with the extension .PDB, such as PEP6.PDB.{{cite book|url=http://computarium.lcd.lu/library/PDF/MUELLER_Upgrading_and_Repairing_PCs_1998.pdf|title=Upgrading and Repairing PCs|edition=Tenth Anniversary|first1=Scott|last1=Mueller|first2=Craig|last2=Zacker|date=September 1998|isbn=0-7897-1636-4|publisher=Que Publishing|editor1-first=Jim|editor1-last=Minatel|editor2-first=Jill|editor2-last=Byus|editor3-first=Rick|editor3-last=Kughen|page=79|quote=Processor Steppings (Revisions) and Microcode Update Revisions Supported by the Update Database File PEP6.PDB … Using the processor update utility (CHECKUP3.EXE), … can easily verify … the correct microcode update|access-date=1 October 2018}}{{rp|79}}

The processor boots up using a set of microcode held inside the processor and stored in an internal ROM. A microcode update populates a separate SRAM and set of "match registers" that act as breakpoints within the microcode ROM, to allow jumping to the updated list of micro-operations in the SRAM. A match is performed between the Microcode Instruction Pointer (UIP) all of the match registers, with any match resulting in a jump to the corresponding destination microcode address.{{rp|3}} In the original P6 architecture there is space in the SRAM for 60 micro-operations, and multiple match/destination register pairs.{{rp|3}} It takes one processor instruction cycle to jump from ROM microcode to patched microcode held in SRAM. Match registers consist of a microcode match address, and a microcode destination address.

The processor must be in protection ring zero ("{{nowrap|Ring 0}}") in order to initiate a microcode update.{{rp|1}} Each CPU in a symmetric multiprocessing arrangement needs to be updated individually.{{rp|1}}

An update is initiated by placing its address in eax register, setting ecx = 0x79, and executing a wrmsr (Write model-specific register).{{rp|435}}

Intel distributes microcode updates as a 2,048 (2 kilobyte) binary blob. The update contains information about which processors it is designed for, so that this can be checked against the result of the CPUID instruction. The structure is a 48-byte header, followed by 2,000 bytes intended to be read directly by the processor to be updated:

1. A microcode program that is executed by the processor during the microcode update process. This microcode is able to reconfigure and enable or disable components using a special register, and it must update the breakpoint match registers.
2. Up to sixty patched micro-operations to be populated into the SRAM.
3. Padding consisting of random values, to obfuscate understanding of the format of the microcode update.

Each block is encoded differently, and the majority of the 2,000 bytes are not used as configuration program and SRAM micro-operation contents themselves are much smaller. Final determination and validation of whether an update can be applied to a processor is performed during decryption via the processor. Each microcode update is specific to a particular CPU revision, and is designed to be rejected by CPUs with a different stepping level. Microcode updates are encrypted to prevent tampering and to enable validation.{{cite magazine|url=http://www.techweb.com/se/directlink.cgi?EET19970630S0007|url-status=dead|archive-url=https://web.archive.org/web/19991113012445/http://www.techweb.com/se/directlink.cgi?EET19970630S0007|archive-date=1999-11-13|title=Intel preps plan to bust bugs in Pentium MPUs|first=Alexander|last=Wolfe|magazine=EE Times|via=Techweb|access-date=3 October 2018|date=30 June 1997|issue=960|quote=obscure moniker "BIOS Update Feature." … "Each BIOS Update is tailored for a particular stepping of [a] processor," … data block is mapped directly-… after decryption-to the microcode itself.}}

With the Pentium there are two layers of encryption and the precise details explicitly {{Em|not}} documented by Intel, instead being only known to fewer than ten employees.{{cite magazine|url=http://www.eetimes.com/news/97/963news/hole.html|archive-url=https://web.archive.org/web/20030309102752/http://www.eetimes.com/news/97/963news/hole.html|archive-date=2003-03-09|magazine=EE Times|date=30 June 1997|first=Alexander|last=Wolfe|title=Hole seen in Intel's bug-busting feature|location=Santa Clara|quote=Ajay Malhortra, a technical marketing manager based here at Intel's microprocessor group. "Not only is the data block containing the microcode patch encrypted, but once the processor examines the header of the BIOS update, there are two levels of encryption in the processor that must occur before it will successfully load the update." … closely guarded secret. "There is no documentation," said Frank Binns, an architect in Intel's microprocessor group. "It's not as if you can get an Intel 'Red Book' with this stuff written down. It's actually in the heads of less than 10 people in the whole of Intel."}}

Microcode updates for Intel Atom, Nehalem and Sandy Bridge additionally contain an extra 520-byte header containing a 2048-bit RSA modulus with an exponent of 17 decimal.{{rp|7,8}}

Special debugging-specific microcode can be loaded to enable Extended Execution Trace, which then outputs extra information via the Breakpoint Monitor Pins. On the Pentium 4, loading special microcode can give access to Microcode Extended Execution Trace mode. When using the JTAG Test Access Port (TAP), a pair of Breakpoint Control registers allow breaking on microcode addresses.

During the mid-1980s NEC and Intel had a long-running US federal court case about microcode copyright.{{cite journal|url=https://repository.jmls.edu/cgi/viewcontent.cgi?article=1423&context=jitpl|

journal=Computer/Law Journal|volume=10|issue=4|date=Winter 1990|first=David S.|last=Elkins|title=NEC v. Intel: A Guide to Using "Clean Room" Procedures as Evidence|page=453|quote=NEC's use of its clean room procedures as trial evidence … Judge Gray defined microcode … within the Copyright Act's definition of a "computer program," … Intel's microcode is copyrightable. … Intel's microcode did not contain the required copyright notice. … copyrights had been forfeited. … Intel was left with no basis for its claim of copying}} NEC had been acting as a second source for Intel 8086 CPUs with its NEC μPD8086, and held long-term patent and copyright cross-licensing agreements with Intel. In August 1982 Intel sued NEC for copyright infringement over the microcode implementation.{{cite journal |last=Hinckley |first=Robert C. |date=January 1987 |title=NEC v. Intel: Will Hardware Be Drawn into the Black Hole of Copyright Editors' |url=https://digitalcommons.law.scu.edu/cgi/viewcontent.cgi?article=1031&context=chtlj |journal=Santa Clara High Technology Law Journal |volume=3 |issue=1 |quote=Appendix: Microcode formats; 8086/8088 Format; V20/V30 format}}{{cite magazine|url=https://books.google.com/books?id=JRgDwCkMX_cC&pg=PP84|title=Intel witness recants story|first=Kathy Chin|last=Leong|magazine=Computerworld|pages=83, 84|location=San Jose|date=28 March 1988|access-date=2 October 2018|volume=22|number=13|issn=0010-4841}} NEC prevailed by demonstrating via cleanroom software engineering that the similarities in the implementation of microcode on its V20 and V30 processors was the result of the restrictions demanded by the architecture, rather than via copying.

The Intel 386 can perform a built-in self-test of the microcode and programmable logic arrays, with the value of the self-test placed in the EAX register.{{cite web|url=x|url-status=unfit|archive-url=http://pdf.datasheetcatalog.com/datasheet/Intel/mXtuvqv.pdf|archive-date=3 September 2004|title=Intel386 DX Microprocessor 32-BIT CHMOS Microprocessor with Integrated Memory Management|date=December 1995|issue=231630–011|quote=self-test checks the function of all of the Control ROM … EAX register will contain a signature of 00000000h indicating the Intel386 DX passed its self-test of microcode and major PLA contents}} During the BIST, the microprogram counter is re-used to walk through all of the ROMs, with the results being collated via a network of multiple-input signature registers (MISRs) and linear-feedback shift registers.{{cite journal|url=https://nptel.ac.in/courses/Webcourse-contents/IIT%20Kharagpur/Embedded%20systems/Pdf/Lesson-40.pdf|title=5.1 Exhaustive Test in the Intel 80386|series=Testing of Embedded System|journal=Built-In-Self-Test (BIST) for Embedded Systems|page=21|publisher=IIT Kharagpur|date=7 October 2006|access-date=6 October 2018|quote=For ROMs, the patterns are generated by the microprogram counter which is part of the normal logic.}} On start up of the Intel 486, a hardware-controlled BIST runs for 2²⁰ clock cycles to check various arrays including the microcode ROM, after which control is transferred to the microcode for further self-testing of registers and computation units.{{cite conference|url=https://www.computer.org/csdl/proceedings/iccd/1989/1971/00/00063355.pdf|title=Computer Aided Design and Built In Self Test on the i486™ CPU|first1=Patrick|last1=Gelsinger|author-link1=Pat Gelsinger|first2=Sundar|last2=lyengar|first3=Joseph|last3=Krauskopf|first4=James|last4=Nadir|author5=Intel|publisher=IEEE|year=1999|pages=200–201|conference=1989 IEEE International Conference on Computer Design: VLSI in Computers and Processors}} The Intel 486 microcode ROM has 250,000 transistors.

AMD had a long-term contract to reuse Intel's 286, 386 and 486 microcode. In October 2004, a court ruled that the agreement did not cover AMD distributing Intel's 486 in-circuit emulation (ICE) microcode.

Direct Access Testing (DAT) is included in Intel CPUs as part of the design for testing (DFT) and Design for Debug (DFD) initiatives allow full coverage testing of individual CPUs prior to sale.{{cite web|last1=Wu|first1=David M.|last2=Lin|first2=Mike|last3=Reddy|first3=Madhukar|last4=Jaber|first4=Talal|last5=Sabbavarapu|first5=Anil|last6=Thatcher|first6=Larry|author7=Intel Corporation|year=2004|title=An An optimized DFT and test pattern generation strategy for an Intel high performance microprocessor|url=https://eecs.ceas.uc.edu/~jonewb/TESTING/Papers/Intel.pdf|journal=|pages=38, 43, 44|quote=Direct Access Testing (DAT) for array access and diagnosis and Programmable Weak Write Test Mode (PWWTM) for memory cell stability test to reduce the test time. … Array {{abbr|DFT|Design for Test}} test strategy is to use PBIST (Programmable Built-In Self Test) to test the second level cache and use DAT to test the remaining arrays … PBIST is available through the JTAG TAP controller. … DAT mode in PX as shown in Figure 4 … PX has more arrays (>110) … array test coverage of PX is 99.3% ‒ the highest in Pentium 4 family}}

In May 2020, a script reading directly from the Control Register Bus (CRBUS){{cite web |last1=Team |first1=uCode Research |title=chip-red-pill/crbus_scripts |website=GitHub |url=https://github.com/chip-red-pill/crbus_scripts |access-date=26 May 2020 |date=25 May 2020}} (after exploiting "Red Unlock" in JTAG USB-A to USB-A 3.0 with Debugging Capabilities, without D+, D− and Vcc{{Citation|author=Positive Research|title=ptresearch/IntelTXE-PoC|date=2020-07-21|url=https://github.com/ptresearch/IntelTXE-PoC|access-date=2020-07-25}}) was used to read from the Local Direct Access Test (LDAT) port of the Intel Goldmont CPU and the loaded microcode and patch arrays were read.{{cite twitter|number=1262697756805795841|user=_markel___|first=Mark|last=Ermolov|date=2020-05-19|title=Using the Local Direct Access Test (LDAT) DFT feature of Intel Atom CPU, we dumped Microcode Sequencer ROM. Also, we extracted what we think is IROM (Immediates for uops) and even managed to modify MS Patch RAM and Match/Patch registers}} These arrays are only accessible after the CPU has been put into a specific mode, and consist of five arrays accessed through offset 0x6a0:{{cite web|url=https://pbx.sh/ldat/|title=Intel LDAT notes|date=2020-05-22|access-date=2020-05-26|first=Peter|last=Bosch|quote=PDAT CR: 0x6A0; Array Select: 0‒4}}

{{ordered list|start=0

| ROM: Microcode triads

| ROM: Sequence Words

| RAM: Sequence Words (updatable)

| RAM: Match/Patch pairs (updatable)

| RAM: Microcode triads (updatable)

}}

{{reflist|refs=

{{cite report|url=http://www.cs.tau.ac.il/~afek/p6tx050111.pdf|title=Micro Operations (Uops)

|work=The Pentium II/III Processor "Compiler on a Chip"|first=Ronny|last=Ronen|author2=Intel Labs|location=Haifa|publisher=Tel Aviv University|date=18 January 2005|access-date=23 January 2018|archive-url=https://web.archive.org/web/20070416221626/http://www.cs.tau.ac.il/~afek/p6tx050111.pdf|archive-date=16 April 2007|pages=26, 31, 32, 43, 44, 46|quote=Each "CISC" {{abbr|inst|instruction}} is broken into one or more uops … Canonical representation of {{abbr|src|source}}/{{abbr|dest|destination}} (3 {{abbr|src|source}}, 2 {{abbr|dest|destination}}) … e.g., pop eax becomes esp1<-esp0+4, eax1<-[esp0] … {{abbr|ID|Instruction Decoder}}: Convert instructions into {{abbr|uops|micro-operations}}. Buffers up to 6 {{abbr|uops|micro-operations}} … {{abbr|Alloc|Allocation}} & {{abbr|RAT|Register Alias Table}} … able to work on up to 3 {{abbr|uops|micro-operations}} per clock … Reservation station (RS) … Pool of all "not yet executed" {{abbr|uops|micro-operations}} (up to 20) … In order Retirement: … Retires up to 3 {{abbr|uops|micro-operations}} per clock … {{abbr|OOO|Out Of Order}} Cluster … Up to 5 resource-ready {{abbr|uops|micro-operations}} are selected, and dispatched per clock}}

{{cite news|archive-url=https://web.archive.org/web/20091221182054/https://www.ele.uva.es/~jesman/BigSeti/ftp/Cajon_Desastre/MPR/111204.pdf|archive-date=21 December 2009|url=https://www.ele.uva.es/~jesman/BigSeti/ftp/Cajon_Desastre/MPR/111204.pdf|url-status=dead|title=P6 Microcode Can Be Patched|first=Linley|last=Gwennap|date=15 September 1997|work=Microprocessor Report|access-date=23 January 2018|quote=Intel has implemented a microcode patch capability in its P6 processors, including Pentium Pro and Pentium II … allows the microcode to be altered after the processor is fabricated, repairing bugs that are found after the processor is designed. … originally intended the feature to be used only for debugging, but after dealing with the expense of the Pentium FDIV bug … Intel decided to make it usable in the field. … P6 chip contains a complete set of microcode in an internal ROM … BIOS writes a memory address into a special CPU register to trigger a download sequence … P6 processors contain a small SRAM that holds up to 60 microinstructions. The patch code is downloaded into this SRAM … also contains a set of "match" registers that cause a trap when a particular microcode address is encountered. (This is similar to the "instruction breakpoint" capability used to debug assembly code.) This trap, which takes a single cycle to process, vectors microcode execution into the patch RAM. … downloaded microcode consists of two segments. … first is an initialization routine that is run immediately … also initializes the match registers, if necessary. … second segment contains one or more patches that remain in the patch RAM during normal operation and are accessed via a match-register trap. … original microcode is stored in ROM, … match registers allow the operation of the microcode to be changed. In this way, an x86 instruction that is operating incorrectly can be repaired, assuming it is implemented in microcode. … a patch is created to replace a section of the original microcode, performing the correct operation and then jumping back. … number of match registers, … more than one. … single bug, … might require multiple patches, and some bugs are too complex to repair … mechanism could allow multiple bugs to be fixed, … features of the P6 processor can be disabled via a special register … 2,048-byte block of data. The block contains a 48-byte header—which includes a date code, the CPU ID (which includes the stepping level) of the target processor, and a checksum—and 2,000 bytes of data to be downloaded by the processor. … checksum … is not used by the CPU. … 2,000 data bytes are encrypted in a way that Intel claims will be extremely difficult to break. The bytes are divided into blocks of varying lengths, each of which is encoded differently. … typically much smaller than 2,000 bytes, the remaining data is random noise intended to confuse anyone attempting to break the encryption. … Intel has not published any information on the format of its microcode, … is deliberately designed to be difficult to understand. Only a small number of Intel employees know the P6 microcode formats.}}

{{cite news|url=https://www.anandtech.com/show/1998/3|title=Intel Core versus AMD's K8 architecture|first=Johan|last=De Gelas|date=1 May 2006|access-date=23 January 2018|work=AnandTech|page=3|quote=Core architecture is equipped with four x86 decoders, 3 simple decoders and 1 complex decoder … to translate the 1 to 15 byte variable length x86 instructions into … fixed length RISC-like instructions (called micro-ops). … common x86 instructions are translated into a single micro-op … complex decoder is responsible for the instructions that produce up to 4 micro-ops. … really long and complex x86 instructions are handled by a microcode sequencer. … macro-op fusion … the x86 compare instruction ({{abbr|CMP|Compare}}) is fused with a jump ({{abbr|JNE TARG|Jump Not Equal to Target}}).}}

{{cite web|url=https://www.dcddcc.com/docs/2014_paper_microcode.pdf|title=Security Analysis of x86 Processor Microcode|first1=Daming Dominic|last1=Chen|first2=Gail-Joon|last2=Ahn|publisher=Arizona State University|date=11 December 2014|access-date=23 January 2018|pages=1, 5, 7|quote=supervisor privileges (ring zero) are required to update processor microcode … Since the 1970s, processor manufacturers have decoded the x86 … into a sequence of … (RISC) micro-operations (uops) … introduced writable patch memory to provide an update mechanism for implementing dynamic debugging capabilities and correcting processor errata, especially after the infamous Pentium FDIV bug of 1994. … P6 (Pentium Pro) microarchitecture in 1995, … K7 microarchitecture in 1999 … with symmetric multiprocessing (SMP) … should be executed synchronously on each logical processor … patch RAM in addition to the {{abbr|MROM|Microcode read-only memory}} … up to 60 microinstructions, with patching implemented by pairs of match and destination registers. … a 520 byte block containing a 2048-bit RSA modulus that appears to be constant within each processor family. This is followed by a four byte RSA exponent with the fixed value 11h}}

{{cite web|url=http://www.hardice.org/hardice/reference/intel/probe-mode/details-of-intel-probe-mode|title=Details of Intel Probe mode|work=Hardice|access-date=23 January 2018|quote=emit a packet over the {{abbr|BPM|0-7 pins; Breakpoint Monitor Pins 0‒7}} when special instructions are executed … To enable Extended Execution Trace, special microcode patches must be applied … For the Pentium 4 only, there exists a second type … called microcode Extended Execution Trace … Control Register Bus in turn allows access to internal arrays and functions on the processor, such as accessing the {{abbr|LLC|Last Level Cache}} and the microcode/{{abbr|Virtual Fuse|VFuse}} PROM. … that sits on the CPU package but is not within the CPU silicon die. This PROM also contains the microcode that the CPU loads during cold boot. … breakpoint on a 48-bit microcode address … accessed by the {{abbr|TAP|Test Access Port}} commands {{abbr|BRKPTCTLA|Breakpoint Control A}} and {{abbr|BRKPTCTLB|Breakpoint Control B}}.}}

{{cite news|url=https://www.intel.com/content/dam/www/public/us/en/documents/research/2001-vol05-iss-1-intel-technology-journal.pdf|work=Intel Technology Journal|issue=Q1|year=2001|editor-first=Lin|editor-last=Chao|title=The Microarchitecture of the Pentium 4 Processor|first1=Glenn|last1=Hinton|first2=Dave|last2=Sager|first3=Mike|last3=Upton|first4=Darrell|last4=Boggs|first5=Doug|last5=Carmean|first6=Alan|last6=Kyker|first7=Patrice|last7=Roussel|quote=IA-32 instruction bytes are then decoded into basic operations called uops (micro-operations) … advanced form of a Level 1 (L1) instruction cache called the Execution Trace Cache … between the instruction decode logic and the execution core … to store the already decoded … uops. … instructions are decoded once … then used repeatedly from there … has a capacity to hold up to 12K uops … similar hit rate to an 8K to 16K byte conventional instruction cache. … packs the uops into groups of six uops per trace line … microcode ROM … for complex IA-32 instructions, such as string move, and for fault and interrupt handling … Trace Cache jumps into the microcode ROM which then issues the uops … After the microcode ROM finishes sequencing uops … front end of the machine resumes fetching uops from the Trace Cache. … deep buffering of the Pentium 4 processor (126 uops and 48 loads in flight)}}

{{cite book|first=T.|last=Shanley|publisher=Addison-Wesley Professional|year=1998|title=Pentium Pro and Pentium II System Architecture|url=https://books.google.com/books?id=MLJClvCYh34C&pg=PA432|page=435|isbn=9780201309737}}

{{cite journal|url=https://dspace.mit.edu/bitstream/handle/1721.1/35849/6-823Spring-2002/NR/rdonlyres/Electrical-Engineering-and-Computer-Science/6-823Computer-System-ArchitectureSpring2002/4E0FC5FE-6F01-43D7-95FE-91E32EB349CF/0/lecture20.pdf|first=Krste|last=Asanovic|author-link=Krste Asanović|issue=Spring|year=2002|journal=Microprocessor Evolution: 4004 to Pentium Pro|page=14|title=P6 uops|access-date=23 January 2018|quote=Each uop has fixed format of around 118 bits … – opcode, two sources, and destination … – sources and destination fields are 32-bits wide to hold immediate or operand}}

{{cite journal|url=https://www.smtnet.com/library/files/upload/pentium-microarchitecture.pdf|title=MMX Microarchitecture of Pentium Processors With MMX Technology and Pentium II Microprocessors|first1=Michael|last1=Kagan|first2=Simcha|last2=Gochman|first3=Doron|last3=Orenstien|first4=Derrick|last4=Lin|journal=Intel Technology Journal|issue=Q3|year=1997|pages=6, 7|quote=Pentium II processor's microarchitecture is similar to that of the Pentium Pro microprocessor … modified to convert the new MMX instructions to Pentium Pro processor-specific uops (new Single Instruction Multiple Data [SIMD] uops were added to implement the new functionality). … A microcode assist was created to correct the problem and redo the operation. An assist is a customer-invisible event that flushes out the machine and allows microcode to handle rare but difficult-to-handle problems. Since all MMX instructions zero the {{abbr|TOS|Top of Stack}}, the assist needs to write the {{abbr|TOS|Top of Stack}} to zero and restart the operation. … Illegal opcodes that are instruction holes in the MMX instruction opcode map are defined to generate a one uop assist call. This assist call instructs the ROB to flush the machine and causes an assist microcode flow to cause the processor to handle illegal opcode faults.}}

{{cite web|url=http://maggini.eng.umd.edu/pub/pre-exec-cgo2004.pdf|title=Physical Experimentation with Prefetching Helper Threads on Intels Hyper-Threaded Processors|first1=Dongkeun|last1=Kim|first2=Steve|last2=Shih-wei Liao|first3=Perry H.|last3=Wang|first4=Juan|last4=del Cuvillo|first5=Xinmin|last5=Tian|first6=Xiang|last6=Zou|first7=Hong|last7=Wang|first8=Donald|last8=Yeung|first9=Milind|last9=Girkar|first10=John P.|last10=Shen|date=11 January 2004|access-date=24 January 2018|pages=4, 5|quote=L1 Trace cache: 12K micro-ops, 8-way set associative, 6 micro-ops per line … Shared: Trace cache, … {{abbr|IA-32|Intel Architecture 32-bit}} instruction decode, Microcode ROM, {{abbr|Uop|Micro-operation}} retirement logic, … Partitioned: Uop queue}}

{{cite news|url=https://books.google.com/books?id=ZDgEAAAAMBAJ&pg=PA5|work=InfoWorld|date=17 October 1994|access-date=24 January 2018|page=5|title=Court ruling against AMD causes some concern|quote=The decision by the federal district court in San Jose, Calif., said that AMD does not have the right to use Intel's in-circuit emulation (ICE) code in the AMD microprocessors. This code is present on all AMD 486s but is only used in a low-power 486-DXL and 486-DXLV processors. … AMD has started to rework its entire line of 486s to eliminate the code.}}

{{cite magazine |title=Prozessorgeflüster |series=Trends & News / aktuell - Prozessoren |language=de |author-first1=Andreas |author-last1=Stiller |author-first2=Matthias R. |author-last2=Paul |date=1996-05-12 |volume=1996 |issue=6 |magazine=c't – magazin für computertechnik |publisher=Verlag Heinz Heise GmbH & Co KG |issn=0724-8679 |page=20 |url=https://www.heise.de/ct/artikel/Prozessorgefluester-284546.html |access-date=2017-08-28 |url-status=live |archive-url=https://web.archive.org/web/20170828172141/https://www.heise.de/ct/artikel/Prozessorgefluester-284546.html |archive-date=2017-08-28}}

}}

{{refbegin}}

• {{cite patent|country=US|number=5404473|status=patent|inventor1-first=David B.|inventor1-last=Papworth|inventor2-first=Michael A.|inventor2-last=Fetterman|inventor3-first=Andrew F.|inventor3-last=Glew|inventor4=Lawrence O. Smith (III), Michael M. Hancock, first=Beth Schultz|assign1=Intel|title=Apparatus and method for handling string operations in a pipelined processor|pridate=1994-03-01|fdate=1994-03-01|pubdate=1995-04-04}} "the first {{abbr|Cuops|Control micro operations}} in a REP swing operation loads the {{abbr|MS|micro sequencer}} Loop Counter with the number of iterations remaining after the unrolled iterations are executed. … a small number of iterations (e.g., seven), are sent during the time it takes for the Loop Counter in the MS to be loaded. This unrolled code is executed conditionally based on the value of (E)CX … remaining three iterations are turned into NOPS."
• {{cite patent|country=US|number=5559974|inventor1-first=Darrell D.|inventor1-last=Boggs|inventor2-first=Gary L.|inventor2-last=Brown|inventor3-first=Michael M.|inventor3-last=Hancock|inventor4-first=Donald D.|inventor4-last=Parker|assign1=Intel|status=patent|title=Decoder having independently loaded micro-alias and macro-alias registers accessible simultaneously by one micro-operation|pridate=1994-03-01|fdate=1996-09-24|pubdate=1996-09-24}}
• {{cite patent|country=US|number=5566298|inventor1-first=Darrell D.|inventor1-last=Boggs|inventor2-first=Gary L.|inventor2-last=Brown|inventor3-first=Michael M.|inventor3-last=Hancock|inventor4=Donald D. Parker, Gail M. Rupnick|assign1=Intel|status=patent|title=Method for state recovery during assist and restart in a decoder having an alias mechanism|pridate=1994-03-01|fdate=1994-03-01|pubdate=1996-10-15}} "… control returns to the Micro-operation Sequence (MS) unit to issue further error correction Control micro-operations (Cuops). In order to simplify restart, the Cuops originating from the error-causing macroinstruction supplied by the translate programmable logic arrays (XLAT PLAs) are loaded into the Cuop registers, with their valid bits unasserted."
• {{cite patent|country=US|number=5600806|title=Method and apparatus for aligning an instruction boundary in variable length macroinstructions with an instruction buffer|inventor1-first=Gary L.|inventor1-last=Brown|inventor2-first=Donald D.|inventor2-last=Parker|assign1=Intel|pridate=1994-03-01|pubdate=1997-02-04|status=patent}} "ADD, XOR, SUB, AND, and OR, which are implemented with one generic Cuop. Another group of instructions representable by only one {{abbr|Cuop|Control micro-operation}} includes {{abbr|ADC|Add with Carry}} and {{abbr|SBB|Subtract with Borrow}}
• {{cite patent|country=US|number=5630083|title=Decoder for decoding multiple instructions in parallel|inventor1-first=Adrian L.|inventor1-last=Carbine|inventor2-first=Gary L.|inventor2-last=Brown|inventor3-first=Donald D.|inventor3-last=Parker|assign1=Intel|pridate=1994-03-01|fdate=1996-07-03|pubdate=2013-03-01|status=patent}}
• {{cite patent|country=US|number=6055656|inventor1-first=James A.|inventor1-last=Wilson, Jr.|inventor2-first=Anthony C.|inventor2-last=Miller|inventor3-first=Michael W.|inventor3-last=Rhodehamel|inventor4=Adrian Carbine, Derek B. I. Feltham, Sumeet Agrawal|title=Control register bus access through a standardized test access port|pubdate=2000-04-25|fdate=1995-05-02|pridate=1995-05-02|assign1=Intel|status=patent}}
• {{cite patent|country=US|number=20030196096|inventor1-last=Sutton|inventor1-first=James A.|title=Microcode patch authentication|pubdate=2003-10-16|fdate=2002-04-12|status=patent}}
• {{cite patent|country=US|number=5948097|title=Method and apparatus for changing privilege levels in a computer system without use of a call gate|inventor1-first=Andrew|inventor1-last=Glew|inventor2-first=Scott Dion|inventor2-last=Rodgers|assign1=Intel|fdate=1996-08-29|pridate=1996-08-29|pubdate=1999-09-07|status=patent}} "SYSENTER and SYSEXIT are assembly-language instructions that may be executed on an Intel architecture processor, such as the Pentium Pro processor … micro-operation is determined to be ready when its source fields have been filled with appropriate data … instruction decode unit comprises one or more translate (XLAT) programmable logic arrays (PLAs) that decode each instruction in to one or more micro-operations. … SYSENTER and SYSEXIT instructions are decoded in to micro-operations that perform the steps illustrated in FIGS. 5 and 6, respectively."
• {{cite web|url=https://opensource.apple.com/source/xnu/xnu-3789.41.3/osfmk/i386/ucode.c.auto.html|title=Microcode updater interface sysctl|work=XNU|quote=#define IA32_BIOS_UPDT_TRIG (0x79) /* microcode update trigger MSR */|access-date=24 January 2018|format=ucode.c driver}}
• {{cite conference|volume=1|doi=10.1109/TEST.2002.1041744|first1=A. T.|last1=Sivaram|first2=Daniel|last2=Fan|first3=A.|last3=Yiin|book-title=Proceedings. International Test Conference|title=Efficient embedded memory testing with APG|date=2002-10-10|pages=47–54|isbn=0-7803-7542-4|issn=1089-3539|publisher=IEEE|location=Baltimore, Maryland|s2cid=19579807}}
• {{cite web|first=Peter|last=Bosch|date=2020-10-01|access-date=2020-11-01|url=https://www.youtube.com/watch?v=4oFOpDflJMA|location=Netherlands|title=Under the hood of a CPU: Reverse Engineering the P6 microcode|website=YouTube}}

{{refend}}

• [https://github.com/chip-red-pill/uCodeDisasm uCodeDisasm] — Intel microcode disassembler in Python (from CRBUS), names of uops

Microcode, Intel

Microcode, Intel