|
| |
| | Pentium - Wikipedia, the free encyclopedia |
 | | Pentium OverDrive processors were released at speeds of 63 and 83MHz as an upgrade option for older 486-class computers. | | | i586 is used in programming though, referring to all the early Pentium processors, and Pentium-like processors made by Intel's competitors. | | | However, software must be specially optimized to make use of MMX, and the increased speed the P55C showed at its launch was mainly due to the fact that the internal cache had been doubled in size to 32 KB. |
|
http://en.wikipedia.org/wiki/Intel_Pentium
(759 words)
|
|
| |
| | Forward of Pentium FDIV bug discussion |
| | Had all the large bugs been exceedingly unlikely, and had all the likely bugs all been of very low magnitude, I would not dispute so strenuously the emerging (this week) industry consensus that the FDIV bug is not serious other than politically for the computer supply side. | | | The rate at which errors are encountered, which for this bug is extremely dependent on the application, has to do with the number and distribution of occurrences of the bug in operand space. | | | This is the largest error observed to date, and 2^-14 may well be the maximum possible error for this bug. |
|
http://www.interesting-people.org/archives/interesting-people/199412/msg00014.html
(889 words)
|
|
| |
| | Bugs in the Intel Microprocessors |
| | The problem does not occur on the specific use of the divide instruction to compute the reciprocal of the input operand in single precision. | | | It was caused by an error in a lookup table that was a part of Intel's SRT algorithm that was to be faster and more accurate. | | | Related instructions that are affected by the bug are FDIVP, FDIVR, FDIVRP, FIDIV, FIDIVR, FPREM, and FPREM1. |
|
http://www.cs.earlham.edu/~dusko/cs63/fdiv.html
(532 words)
|
|
| |
| | MacKiDo/History/History_PentBug |
| | The bug was critical to many users, and converted many users computers from a cost saving device into a liability. | | | While it is not unusual for Hardware bugs in processors it is rare that they are this serious, furthermore Intel went out of their way to hide the bug. | | | But he was soon silenced by Intel with an NDA as Intel tried to bury the problem. |
|
http://www.mackido.com/History/History_PentBug.html
(1098 words)
|
|
| |
| | P54C Erratum 23 (FDIV bug). |
| | I made the mistake of stating publicly that I didn't think that the FDIV bug was that big of a problem. | | | When this bug was first reported, Intel denied that it existed. | | | But Intel's method of responded to the bug was a very big deal. |
|
http://www.x86.org/errata/feb97/bugs.htm
(547 words)
|
|
| |
| | Pentium Division Difficulties |
| | The result will be an efficient and reliable assembly language macro replacement for the Pentium FDIV instruction. | | | The basis for the workaround is Coe's characterization, now confirmed by Tang and Intel, of bit patterns that must be present in the divisors of operand pairs that lead to the error. | | | With this approach, it is not necessary to test the magnitude of the residual resulting from a division. |
|
http://www.csc.fi/math_topics/Mail/NANET94/msg00877.html
(601 words)
|
|
| |
| | Pentium FDIV bug - a picture |
| | The correct values all would round to 1.3338 and the incorrect values all would round to 1.3337, an error in the 5th significant digit. | | | A well known operation which exhibits this phenomenon is 4195835/3145727. | | | A ZIP file containing the Microsoft Excel spreadsheet along with an Encapsulated Postscript copy of this graphic can be downloaded here. |
|
http://www.ku.edu/cwis/units/IPPBR/pentium_fdiv/pentgrph.html
(207 words)
|
|
| |
| | The Role of Proof in Ensuring Software and Hardware Reliability |
| | The article goes on to describe a bug in Intel’s latest chip that remained unfixed, even as the chips were on their way to market. | | | No longer is the reliability of our software and hardware systems the concern of specialists in the field of computer science—technology’s growing involvement in daily life raises concerns among users as to its safety and reliability. | | | "The Intel Corp. will not formally introduce its Pentium II microprocessor until Wednesday, but the World Wide Web is already buzzing with reports of a bug that causes the new chip to make errors in some complex mathematical calculations" ("Flaw" Technology/Cybertimes). |
|
http://www.cs.berkeley.edu/~gporter/pubs/roleofproof.html
(2385 words)
|
|
| |
| | Citations: Inside the Pentium Fdiv bug - COE (ResearchIndex) |
| | The bug was not detected inspite of simulation patterns. | | | ....as this is where the FDIV error occurrred. | | | Although Intel argued for a longer time that in practice this mistake would not have a serious influence on the computations [42] a recall offer became unavoidable. |
|
http://citeseer.lcs.mit.edu/context/156324/0
(1170 words)
|
|
| |
| | File 59014 |
| | The Pentium bug could cause it to behave slightly less optimally, but there is no possible way that this could corrupt data or generate incorrect results. | | | DETAILS Here are some places where the floating-point bug could potentially be a problem for Oracle software: o It is possible to write a Pro*C or an OCI program that binds a FLOAT variable to be inserted into the database, or defines a FLOAT variable to accept output from the database. | | | The only place where data corruption could occur would be in front-end client-side software running on Pentium systems. |
|
http://www.uaex.edu/srea/osee/59014.htm
(647 words)
|
|
| |
| | Ooops! |
| | The best characteristic and only characteristic of the bug to come from Intel is its 1 in 9 billion probability of occurring with random operands. | | | The message described the bug as a "subtle flaw" and estimated that the average "spreadsheet user" would encounter the problem only once in every 27,000 years. | | | What so enraged the Internet-based users, was not so much the bug itself; bugs *do* appear in processors and all processors go through a constant process of improvement. |
|
http://davefaq.com/Opinions/Stupid/Pentium.html
(4557 words)
|
|
| |
| | Geek.com Geek News - Pentium FDIV bug revisted |
| | They claim no one running Word or WordPerfect will ever see the bug, and there should be only one occurrence every 27,000 years for spreadsheet users, one occurrence every 2,000 years for financial programs like Quicken, Microsoft Money, etc., and one occurrence every 270 years for Paintshop, Acrobat, and various games. | | | Sure, the bug might crop up for them when they're ripping a cd or displaying a complex piece of porn on their screen. | | | That bug can be detected using the Windows calculator program. |
|
http://www.geek.com/news/geeknews/2002mar/bch20020326010903.htm
(1384 words)
|
|
| |
| | Index of /pub/FDIV/ |
| | For the remaining buggy pairs, the bug can only be detected on the Pentium by extracting the 12-bit extension word completing the mantissa to 64 bits. | | | The file "all1precerrs" lists all single-precision pairs each in the interval [2^23,2^24) generating an error, grouped under magnitude of the error. | | | Host: boole.stanford.edu Directory: /pub/FDIV This file: README Other files: all1precerrs, bugs, triples, triplex Subdirectory: individual.bugs Either fetch the whole thing as one big file, namely bugs, or get individual files as needed from the individual.bugs directory. |
|
http://boole.stanford.edu/pub/FDIV
(718 words)
|
|
| |
| | "FDIV Pentium Bug Fix in BP" by TERJE MATHISEN |
| | During the unit startup code, I test the FDIV instruction, and if if fails, initialize a 16-byte table of critical nibble values. | | | Here is a BP version of the algorithm: The FDIV_FIX unit defines a single function: function fdiv(x: extended; y: extended): extended; which will use about 25% more cycles than a single fdiv call, but always return exact results for all double precision numbers, including the special cases of Nan, Inf and denormal. | | | It also handles all extended precision numbers, giving a maximum error of a single bit in the last position (1ulp). |
|
http://www.bsdg.org/SWAG/DATATYPE/0037.PAS.html
(332 words)
|
|
| |
| | FEATURE: Pentium Update |
| | Most recently, IBM has halted the use of the Pentium processor in its x86-based computer systems until further notice. | | | With this in mind, it is understandable why the FDIV flaw went unnoticed for so long. | | | After several months of increasing hysteria, Dr. Andrew Grove, CEO of Intel, decided to post a message in which he apologized, admitting that there was a flaw in the Pentium and stating that Intel engineers were busy working on a solution to the problem. |
|
http://www.monitor.ca/monitor/issues/vol2iss6/feature2.html
(994 words)
|
|
| |
| | Intel Microprocessor Bugs |
| | A good source for finding about Intel's bugs and for anybody interested in Intel and anti-Intel information that can not be found on Inte's page. | | | This is a *very* new bug, so Intel hasn't cought up with it yet. | | | Includes important information on the bug presented by Intel. |
|
http://www.cs.earlham.edu/~dusko/cs63/bibliography.html
(395 words)
|
|
| |
| | Pentium 1 FDIV bug |
| | Pentium 60 to 100 had the bug but many systems with those CPUs don't. | | | I may be mistaken, but I believe certain versions of Windows have been patched for this bug (especially NT). | | | People could send in the old CPU and Intel would send a new fixed one at the same speed for free. |
|
http://www.dosgames.com/forum/about2535.html
(355 words)
|
|
| |
| | port-i386: Re: Pentium FDIV bug |
| | Except maybe for using FP emulation and checking every fdiv instruction, resulting in a significant slowdown.. | | | NetBSD doesn't compensate for this one, as it is not possible to do so for the operating system. | | | I > assume there is no binary level workaround other than disabling the FPU? |
|
http://mail-index.netbsd.org/port-i386/1999/10/18/0008.html
(240 words)
|
|
| |
| | R. Smith==>Nicely P*ntium flaw e-mail |
| | Also he knew that Phar Lap would be interested in knowing about the bug since many of our DOS-Extender customers write just the sort of number-crunching applications that would be most affected by the FDIV bug. | | | The interesting thing is that the problem can be easily duplicated by running the Windows calculator. | | | My goal was to see how widespread the bug was. |
|
http://www.trnicely.net/pentbug/rsmith.html
(1324 words)
|
|
| |
| | P*ntium FDIV flaw FAQ |
| | Enumerated below are several questions that have frequently been posed to me, regarding the discovery, nature, and implications of the P*ntium FDIV flaw. | | | Many of these questions were submitted by Dr. Denis Delbecq of the Paris based computer periodical "Science et Vie Micro." /*************************************************************/ Q1: How can a user check a P*ntium machine for the presence of the bug? | | | I have provided a C source code, and corresponding DOS executable, for the purpose of testing for the bug; see pentbug.zip. |
|
http://www.trnicely.net/pentbug/pentbug.html
(403 words)
|
|
| |
| | The Pentium Divison Flaw - References |
| | An updated version of the Microsoft Windows calculator which fixes the rounding bug [2.01 - 2.00 = 0.0] is available from the following site. | | | OTHER PAPERS BY DAVID W. "How Computers Generate Random Numbers", 1991. | | | This bug is unrelated to the Pentium division bug. |
|
http://members.cox.net/srice1/pentbug/references.html
(140 words)
|
|
| |
| | The Risks Digest Volume 19: Issue 4 |
| | I finally got rid of my Intel Pentium computer with the FDIV bug. | | | I was surprised to find that that the calculation of: 4195835 - (4195835/3145727)*3145727 within the spreadsheet still showed the answer 512 instead of zero. | | | Last night, while checking out my new Pentium Pro computer with Microsoft Excel 97 I decided to open up the spreadsheet which demonstrates the Pentium Floating Point divide bug. |
|
http://catless.ncl.ac.uk/Risks/19.04.html
(3561 words)
|
|
| |
| | Pentium FDIV bug - Wikipedia |
| | Later, Intel claimed that it was not serious and would not affect most users; however, if you could prove that you were affected, Intel would replace your processor. | | | This result was quickly verified by other people around the Internet, and became known as the Pentium FDIV bug (FDIV is the x86 assembly language instruction for floating point division). | | | Note that this problem occured only on some models of the original Pentium processor. |
|
http://nostalgia.wikipedia.org/wiki/Pentium_FDIV_bug
(238 words)
|
|
| |
| | Related: Inside the Pentium Fdiv bug - COE (ResearchIndex) |
| | 4: Word level model checking -- Avoiding the Pentium FDIV error (context) - CLARKE, KHAIRA et al. | | | 6: Statistical analysis of floating point flaw in the Pentium processor (context) - Sharangpani, Barton - 1994 | | | CiteSeer Home Related articles for: COE, T. Inside the Pentium Fdiv bug. |
|
http://citeseer.lcs.mit.edu/nrelatedgid/156324
(307 words)
|
|
| |
| | Q&A: THE PENTIUM FDIV BUG |
| | Q: According to Intel, the Pentium conforms to the IEEE standards 754 and 854 for floating point arithmetic. | | | A: "Life is like a box of chocolates." (Source: F. Gump of Intel) Q: Why didn't Intel call the Pentium the 586? | | | Q: What algorithm did Intel use in the Pentium's floating point divider? |
|
http://grove.cs.jmu.edu/groverf/humor/fdivBug.html
(194 words)
|
|
| |
| | FDIV Replacement Program |
| | In addition to this guide, you can call the customer service numbers for information regarding the replacement program. | | | In December of 1994, Intel announced a lifetime replacement policy on the well publicized floating point unit flaw, contained in the then current version (60-100MHz) of the Pentium® processor. | | | Frequently Asked Questions (updated - 14 May 98) about the Pentium® processor replacement program |
|
http://support.intel.com/support/processors/pentium/fdiv
(211 words)
|
|
| |
| | A Discussion of and Fix for the Pentium FDIV Bug -- from Mathematica Information Center |
| | A Discussion of and Fix for the Pentium FDIV Bug -- from Mathematica Information Center | | | This patch was created by Wolfram Research, Inc. This item also contains a Mathematica notebook that presents an in-depth analysis of the Pentium bug, and discusses the patch. | | | A Discussion of and Fix for the Pentium FDIV Bug |
|
http://library.wolfram.com/infocenter/TechNotes/176
(82 words)
|
|
|
