7 ms·

There's a great show on Netflix that's a fictional retelling of the engineers reverse engineering the BIOS for their first IBM clones. Highly stylized, but (I'v

There's a great show on Netflix that's a fictional retelling of the engineers reverse engineering the BIOS for their first IBM clones. Highly stylized, but (I've only seen the first season) a fun watch. "Halt and Catch Fire" - http://www.imdb.com/title/tt2543312/ http://www.imdb.com/title/tt2543312/

DrScump 10y ago
There was an instruction set parody circulated around the mainframe world in the 1980s (at least) entitled "Name That Instruction Set!" which claimed to be new opcodes for the alleged upcoming Osborne II. It included, among others: HCF - Halt and Catch Fire IIB - Ignore Interrupt and Branch PIL - Perform Infinite Loop etc. I can't find an exact copy, but this appears to be somewhat of a superset: http://www.physics.ohio-state.edu/~bcd/humor/instruction.set.html http://www.physics.ohio-state.edu/~bcd/humor/instruction.set... Another: http://www.softpanorama.org/Lang/Asmorama/humor.shtml http://www.softpanorama.org/Lang/Asmorama/humor.shtml
- shermanyo 10y ago
  Ah cool! I never got the reference. I like: CML Compute Meaning of Life What's the bet its opcode is 0x2A ? :P
  - King-Aaron 10y ago
    I quite like: EJD d,v Eject disk in drive d with initial velocity v This would be pretty handy at times.
    - chiph 10y ago
      I used to work with a system that had replaceable disk packs -- think 18" diameter removable hard drives transported in a plastic dome. They had read heads that were driven by a linear motor, which was an absolutely huge electromagnet. Extended use would sometimes "walk" the equipment out of position (much like an out-of-balance washing machine), so it was common to put rubber doorstops under it to prevent this. One of the drive tests performed during maintenance was a random seek. Normally, the head position was controlled by a servo with a feedback loop, but this one time the servo failed and the magnet drove the head assembly out the back of the equipment and into the wall .. right at crotch height. There was a supplemental maintenance instruction issued shortly afterward to limit the use of that procedure, and to not stand behind the equipment when it was being run.
      - digi_owl 10y ago
        Present day computing seems less exciting, unless you have something like a PSU shortcircuit.
        mikeash 10y ago
        I don't know, failure modes these days could include exciting bugs like "drive to the nearest large concentration of pedestrians and ram it at top speed." You just need to find the right field.
        King-Aaron 10y ago
        I don't know, avionics seem to have a bit of a "wow" factor when they go wrong...
  - DrScump 10y ago
    No, that should be the return value.
- ajdlinux 10y ago
  As an IBM PowerPC hacker I'm disappointed that https://twitter.com/ppcinstructions https://twitter.com/ppcinstructions seems to have stopped publishing...
  - filereaper 10y ago
    Haha, I've always had to look up rlwinm and its ilk no matter how many times I've looked it up previously in the ISA. Its crazy how terse PPC's instructions are, the PPC ISA book felt like reading PHD paper where as ARM's ISA manuals felt like a gentle novel. Here's a wonderful excerpt from the latest 2.07 ISA: "The following notation is used in the definitions below. When MMCR0_PMCC = 11, “PMCs” refers to PMCs 1 - 4 and “PMCj” or PMCjCE refers to “PMCj or PMCjCE, respectively, where j = 2 - 4; otherwise, “PMCs” refers to PMCs 1 - 6 and “PMCj” or PMCjCE refers to “PMCj or PMCjCE, respectively, where j = 2 - 6." Get all of that? Good, now go write code to make use of it.
- grkvlt 10y ago
  Of course, for IBM PowerPC there was the (real!) assembly instruction EIEIO which meant: Enforce In-Order Execution of I/O. [1] [1] https://en.wikipedia.org/wiki/Enforce_In-order_Execution_of_I/O https://en.wikipedia.org/wiki/Enforce_In-order_Execution_of_...
  - Shivetya 10y ago
    I remember around the time of Power 4 IBM gave us a presentation which basically stated that with regards to the speculative execution of instructions it was fast enough to just try all paths as it fast enough to not show an impact.
  - ajdlinux 10y ago
    "I would suggest re-naming "rmbdd()". I _assume_ that "dd" stands for "data dependent", but quite frankly, "rmbdd" looks like the standard IBM "we lost every vowel ever invented" kind of assembly lanaguage to me. I'm sure that having programmed PPC assembly language, you find it very natural (IBM motto: "We found five vowels hiding in a corner, and we used them _all_ for the 'eieio' instruction so that we wouldn't have to use them anywhere else")." -- Linus Torvalds, LKML (http://ozlabs.org/~rusty/linux.conf.au-2004-emails.txt http://ozlabs.org/~rusty/linux.conf.au-2004-emails.txt)
- bryanlarsen 10y ago
  Early microprocessors like the 6800 and the 6502 had a ton of 'undocumented' instructions. They weren't deliberately designed but they often performed useful behaviour by accident. So later engineers documented and used these undefined instructions, to the delight of emulator designers every where. But some of these instructions did not perform useful behaviour, many of them freezing the machine. Some were even worse, freezing the machine into a state where it was very rapidly toggling the bus. If you had the wrong thing attached to the bus, it could very literally be Halt and Catch Fire. Instruction $DD on the 6800 was labelled HCF in some instruction set listings for that reason.
- bryanlarsen 10y ago
  The 6809 had an instruction "SEX" to sign extend (aka convert) a signed 8 bit byte into a signed 16 bit word.
eddieroger 10y ago
That's exactly what this article is about - the full title is: "THE INCREDIBLE TRUE STORY BEHIND AMC’S HALT AND CATCH FIRE – HOW COMPAQ CLONED IBM AND CREATED AN EMPIRE". The show is fantastic. Season 2 was good as well, but I still liked S01 a little better.
- CWuestefeld 10y ago
  I watched the first few episodes, but I couldn't really get into it. The biggest thing was that I couldn't believe that our hacker hero (I don't recall her name) could be that person. It may be sexist, but the statistics were that such a person was almost certainly male. But even beyond that, the personality just didn't click for me.
timpark 10y ago
Further to what eddieroger said, I found S2 slow to start... Joe is a bit "lost" for several episodes, but it gets better. S3 starts in August according to IMDB.

DrScump 10y ago

There was an instruction set parody circulated around the mainframe world in the 1980s (at least) entitled "Name That Instruction Set!" which claimed to be new opcodes for the alleged upcoming Osborne II. It included, among others: HCF - Halt and Catch Fire IIB - Ignore Interrupt and Branch PIL - Perform Infinite Loop etc. I can't find an exact copy, but this appears to be somewhat of a superset: http://www.physics.ohio-state.edu/~bcd/humor/instruction.set.html http://www.physics.ohio-state.edu/~bcd/humor/instruction.set... Another: http://www.softpanorama.org/Lang/Asmorama/humor.shtml http://www.softpanorama.org/Lang/Asmorama/humor.shtml

shermanyo 10y ago
Ah cool! I never got the reference. I like: CML Compute Meaning of Life What's the bet its opcode is 0x2A ? :P
- King-Aaron 10y ago
  I quite like: EJD d,v Eject disk in drive d with initial velocity v This would be pretty handy at times.
  - chiph 10y ago
    I used to work with a system that had replaceable disk packs -- think 18" diameter removable hard drives transported in a plastic dome. They had read heads that were driven by a linear motor, which was an absolutely huge electromagnet. Extended use would sometimes "walk" the equipment out of position (much like an out-of-balance washing machine), so it was common to put rubber doorstops under it to prevent this. One of the drive tests performed during maintenance was a random seek. Normally, the head position was controlled by a servo with a feedback loop, but this one time the servo failed and the magnet drove the head assembly out the back of the equipment and into the wall .. right at crotch height. There was a supplemental maintenance instruction issued shortly afterward to limit the use of that procedure, and to not stand behind the equipment when it was being run.
    - digi_owl 10y ago
      Present day computing seems less exciting, unless you have something like a PSU shortcircuit.
      - mikeash 10y ago
        I don't know, failure modes these days could include exciting bugs like "drive to the nearest large concentration of pedestrians and ram it at top speed." You just need to find the right field.
      - King-Aaron 10y ago
        I don't know, avionics seem to have a bit of a "wow" factor when they go wrong...
- DrScump 10y ago
  No, that should be the return value.
ajdlinux 10y ago
As an IBM PowerPC hacker I'm disappointed that https://twitter.com/ppcinstructions https://twitter.com/ppcinstructions seems to have stopped publishing...
- filereaper 10y ago
  Haha, I've always had to look up rlwinm and its ilk no matter how many times I've looked it up previously in the ISA. Its crazy how terse PPC's instructions are, the PPC ISA book felt like reading PHD paper where as ARM's ISA manuals felt like a gentle novel. Here's a wonderful excerpt from the latest 2.07 ISA: "The following notation is used in the definitions below. When MMCR0_PMCC = 11, “PMCs” refers to PMCs 1 - 4 and “PMCj” or PMCjCE refers to “PMCj or PMCjCE, respectively, where j = 2 - 4; otherwise, “PMCs” refers to PMCs 1 - 6 and “PMCj” or PMCjCE refers to “PMCj or PMCjCE, respectively, where j = 2 - 6." Get all of that? Good, now go write code to make use of it.
grkvlt 10y ago
Of course, for IBM PowerPC there was the (real!) assembly instruction EIEIO which meant: Enforce In-Order Execution of I/O. [1] [1] https://en.wikipedia.org/wiki/Enforce_In-order_Execution_of_I/O https://en.wikipedia.org/wiki/Enforce_In-order_Execution_of_...
- Shivetya 10y ago
  I remember around the time of Power 4 IBM gave us a presentation which basically stated that with regards to the speculative execution of instructions it was fast enough to just try all paths as it fast enough to not show an impact.
- ajdlinux 10y ago
  "I would suggest re-naming "rmbdd()". I _assume_ that "dd" stands for "data dependent", but quite frankly, "rmbdd" looks like the standard IBM "we lost every vowel ever invented" kind of assembly lanaguage to me. I'm sure that having programmed PPC assembly language, you find it very natural (IBM motto: "We found five vowels hiding in a corner, and we used them _all_ for the 'eieio' instruction so that we wouldn't have to use them anywhere else")." -- Linus Torvalds, LKML (http://ozlabs.org/~rusty/linux.conf.au-2004-emails.txt http://ozlabs.org/~rusty/linux.conf.au-2004-emails.txt)
bryanlarsen 10y ago
Early microprocessors like the 6800 and the 6502 had a ton of 'undocumented' instructions. They weren't deliberately designed but they often performed useful behaviour by accident. So later engineers documented and used these undefined instructions, to the delight of emulator designers every where. But some of these instructions did not perform useful behaviour, many of them freezing the machine. Some were even worse, freezing the machine into a state where it was very rapidly toggling the bus. If you had the wrong thing attached to the bus, it could very literally be Halt and Catch Fire. Instruction $DD on the 6800 was labelled HCF in some instruction set listings for that reason.
bryanlarsen 10y ago
The 6809 had an instruction "SEX" to sign extend (aka convert) a signed 8 bit byte into a signed 16 bit word.

eddieroger 10y ago

That's exactly what this article is about - the full title is: "THE INCREDIBLE TRUE STORY BEHIND AMC’S HALT AND CATCH FIRE – HOW COMPAQ CLONED IBM AND CREATED AN EMPIRE". The show is fantastic. Season 2 was good as well, but I still liked S01 a little better.

CWuestefeld 10y ago
I watched the first few episodes, but I couldn't really get into it. The biggest thing was that I couldn't believe that our hacker hero (I don't recall her name) could be that person. It may be sexist, but the statistics were that such a person was almost certainly male. But even beyond that, the personality just didn't click for me.

timpark 10y ago

Further to what eddieroger said, I found S2 slow to start... Joe is a bit "lost" for several episodes, but it gets better. S3 starts in August according to IMDB.