Gene Amdahl 1922 - 2015
November 14, 2015 7:21 AM Subscribe
As a young computer scientist at International Business Machines Corporation in the early 1960s, he played a crucial role in the development of the System/360 series, the most successful line of mainframe computers in IBM’s history. Its architecture influenced computer design for years to come. Computer pioneer Gene Amdahl has died.
My favorite line in the obit is from Frederick P. Brooks Jr.: “By sheer intellectual force, plus some argument and banging on the table, he maintained architectural consistency across six engineering teams.”
posted by benito.strauss at 7:37 AM on November 14, 2015 [15 favorites]
posted by benito.strauss at 7:37 AM on November 14, 2015 [15 favorites]
.
posted by eriko at 7:39 AM on November 14, 2015 [1 favorite]
posted by eriko at 7:39 AM on November 14, 2015 [1 favorite]
.
My introduction to assembler programming was a simulated System/360 machine running on a Vax/VMS. While I never want to do that again it certainly brings back some fond memories.
posted by mmascolino at 7:50 AM on November 14, 2015 [2 favorites]
My introduction to assembler programming was a simulated System/360 machine running on a Vax/VMS. While I never want to do that again it certainly brings back some fond memories.
posted by mmascolino at 7:50 AM on November 14, 2015 [2 favorites]
He was also the intellectual grandfather of the PC compatible, since when he created Amdahl Corporation, he introduced a line of computers that were architecturally compatible the the IBM 360.
posted by CheeseDigestsAll at 7:59 AM on November 14, 2015 [3 favorites]
posted by CheeseDigestsAll at 7:59 AM on November 14, 2015 [3 favorites]
.
I learned programming at Penn State on a System/370 learning Fortran, PL/C and 370 assembly language and PDP-11 assembly language via a simulator. This was back in the early eighties and the VM system could give each user his/her own virtual machine to run machine code on without risk of disturbing anyone else's system. Amazingly forward thinking stuff.
posted by octothorpe at 8:05 AM on November 14, 2015 [2 favorites]
I learned programming at Penn State on a System/370 learning Fortran, PL/C and 370 assembly language and PDP-11 assembly language via a simulator. This was back in the early eighties and the VM system could give each user his/her own virtual machine to run machine code on without risk of disturbing anyone else's system. Amazingly forward thinking stuff.
posted by octothorpe at 8:05 AM on November 14, 2015 [2 favorites]
.
posted by Smart Dalek at 8:05 AM on November 14, 2015
posted by Smart Dalek at 8:05 AM on November 14, 2015
Given that the System/360 architecture still runs today in the form of OS/390 and zOS, you really cannot downplay Amdahl's contribution to the industry. Even today, when 5 9s isn't good enough, you go to big iron and run a system based on what he built. Hell, there are still actual System/360s running out there!
Given how fast the industry changed, this is basically the equivalent of us using technology developed in Ancient Egypt on a day to day basis.
Finally, we talk about the amazing Apollo Guidance Computer as being the tool that got us to the moon, but it really didn't. It was instructed to fly a path, and it did -- an amazing feat for the time -- but if you booted it bare, it couldn't get you anywhere. The machines that did the real work of determining that path and correcting the AGC was the IBM System/360 91 and a pair of IBM System/360 95s. The 91 did the early work, but when NASA realized they might need to calculate this stuff in near real time, that's when the pair of 95s were brought in. To handle the real time telemetry, five IBM System/360 75s were installed at Houston -- they decoded the telemetry and displayed it on monitors. Cameras were pointed at the monitors and a mask with labels was placed on the monitor. In the MOCR, the had monitors on the console that they could select which monitor they were watching, so that any controller could see any part of the telemetry stream. You can see the monitors here.
All screamingly ancient, but all worked and worked well, thanks to the System/360.
posted by eriko at 8:06 AM on November 14, 2015 [24 favorites]
Given how fast the industry changed, this is basically the equivalent of us using technology developed in Ancient Egypt on a day to day basis.
Finally, we talk about the amazing Apollo Guidance Computer as being the tool that got us to the moon, but it really didn't. It was instructed to fly a path, and it did -- an amazing feat for the time -- but if you booted it bare, it couldn't get you anywhere. The machines that did the real work of determining that path and correcting the AGC was the IBM System/360 91 and a pair of IBM System/360 95s. The 91 did the early work, but when NASA realized they might need to calculate this stuff in near real time, that's when the pair of 95s were brought in. To handle the real time telemetry, five IBM System/360 75s were installed at Houston -- they decoded the telemetry and displayed it on monitors. Cameras were pointed at the monitors and a mask with labels was placed on the monitor. In the MOCR, the had monitors on the console that they could select which monitor they were watching, so that any controller could see any part of the telemetry stream. You can see the monitors here.
All screamingly ancient, but all worked and worked well, thanks to the System/360.
posted by eriko at 8:06 AM on November 14, 2015 [24 favorites]
.
posted by disclaimer at 8:36 AM on November 14, 2015
posted by disclaimer at 8:36 AM on November 14, 2015
.
I know nothing of the man or mainframes, but reading his obit, to me, his failure is as impressive as his 360 work at IBM and his 470 at Amdahl. To raise $200 million on a speculative project shows that the people in the know really thought amazing things about the guy. I also love that he moved to California to get away from the office politics in Armonk. I live near Armonk in Westchester County and everyone knows an IBMer or three. They have multiple campuses around here. Buy them a few beers and they can tell you all sorts of stories.
posted by AugustWest at 8:50 AM on November 14, 2015
I know nothing of the man or mainframes, but reading his obit, to me, his failure is as impressive as his 360 work at IBM and his 470 at Amdahl. To raise $200 million on a speculative project shows that the people in the know really thought amazing things about the guy. I also love that he moved to California to get away from the office politics in Armonk. I live near Armonk in Westchester County and everyone knows an IBMer or three. They have multiple campuses around here. Buy them a few beers and they can tell you all sorts of stories.
posted by AugustWest at 8:50 AM on November 14, 2015
_______________________________ / | / | | . | | | | | |_________________________________|posted by scruss at 8:50 AM on November 14, 2015 [25 favorites]
0x004B
posted by bonehead at 8:53 AM on November 14, 2015 [9 favorites]
posted by bonehead at 8:53 AM on November 14, 2015 [9 favorites]
.
posted by brambleboy at 9:09 AM on November 14, 2015
posted by brambleboy at 9:09 AM on November 14, 2015
.
posted by doctornemo at 9:35 AM on November 14, 2015
posted by doctornemo at 9:35 AM on November 14, 2015
Combining scruss and bonehead:
________________________________ /. | | | | . | | | | . | |_________________________________|posted by MtDewd at 9:47 AM on November 14, 2015 [2 favorites]
.
No discussion of Amdahl is complete without mentioning the famous million dollar coffee cup (and here's a photo).
posted by Itaxpica at 9:58 AM on November 14, 2015 [8 favorites]
No discussion of Amdahl is complete without mentioning the famous million dollar coffee cup (and here's a photo).
posted by Itaxpica at 9:58 AM on November 14, 2015 [8 favorites]
The million dollar coffee cup is now the top comment in the hacker news obit thread.
posted by bukvich at 10:05 AM on November 14, 2015
posted by bukvich at 10:05 AM on November 14, 2015
Requiem aeternam dona ei, Domine, et lux perpetua luceat eis. Requiescat in pace.
posted by ob1quixote at 10:23 AM on November 14, 2015 [1 favorite]
posted by ob1quixote at 10:23 AM on November 14, 2015 [1 favorite]
Interview from my neck of the woods.
.
posted by Mental Wimp at 10:34 AM on November 14, 2015 [1 favorite]
.
posted by Mental Wimp at 10:34 AM on November 14, 2015 [1 favorite]
The million dollar coffee cup is a funny story and all, but I can't imagine "Here's a way to get a steep discount on our competitor's product" is an effective sales technique, especially when you are trying to compete on price.
posted by ckape at 10:40 AM on November 14, 2015 [1 favorite]
posted by ckape at 10:40 AM on November 14, 2015 [1 favorite]
If you're not going to make the sale anyway, you may as well fuck your competitor and create some goodwill with the prospective customer. They may move somewhere else where they are either starting fresh or where switching is in fact a possibility and remember how you saved them a bunch of money.
Sales is often about relationships, and it pays to cultivate those relationships even if it isn't going to help out this quarter or even this year. The million dollar coffee cup is a great example of the long term thinking that has been lost in today's business climate.
posted by wierdo at 10:49 AM on November 14, 2015 [5 favorites]
Sales is often about relationships, and it pays to cultivate those relationships even if it isn't going to help out this quarter or even this year. The million dollar coffee cup is a great example of the long term thinking that has been lost in today's business climate.
posted by wierdo at 10:49 AM on November 14, 2015 [5 favorites]
The million dollar coffee cup is a funny story and all, but I can't imagine "Here's a way to get a steep discount on our competitor's product" is an effective sales technique, especially when you are trying to compete on price.
You're assuming the company's primary customer isn't IBM, who hadn't yet been prosecuted for ant-trust. What better way to garner a buyout offer than 'we'll stop bidding your contracts down.'
posted by pwnguin at 11:25 AM on November 14, 2015 [1 favorite]
You're assuming the company's primary customer isn't IBM, who hadn't yet been prosecuted for ant-trust. What better way to garner a buyout offer than 'we'll stop bidding your contracts down.'
posted by pwnguin at 11:25 AM on November 14, 2015 [1 favorite]
Strike the antitrust bit I guess. I forgot how long that had dragged on, and only ended around the time I was born.
posted by pwnguin at 11:27 AM on November 14, 2015
posted by pwnguin at 11:27 AM on November 14, 2015
Okay, guys, you need to get it right.
of hunger and thirst.
Next day he was buried,
Face down, Nine edge first.
The punches are significant here too.
at least now that I've got them right
posted by eriko at 11:55 AM on November 14, 2015 [3 favorites]
__________________________________________ | \ | | | | | 0| | | | 0| | 0| | | | 0| |___________________________________________|He died at the console,
of hunger and thirst.
Next day he was buried,
Face down, Nine edge first.
The punches are significant here too.
at least now that I've got them right
posted by eriko at 11:55 AM on November 14, 2015 [3 favorites]
OK, now a derail about the card...
Face down, 9-edge first (which BTW I totally agree is the way to go here) is the way cards are put into a reader. (12-edge first is for the punch side)
When you put cards in the reader, column 1 is on the left. Of course, you may have chosen to put your punches in column 80, but I miss the point of that.
So I give up. What do the punches represent? I didn't bother to go 12*80, but tried to represent a 12-3-8 for x'4B'.
Hard to tell what this is. I assume it's not literally a 10*43 card, so what do the punches mean? 1-4-5-8?
Except for column binary, you're not allowed to have more than 1 punch in rows 3-7, and I don't recognize anything in column binary here, either.
In any case, it better not be ASCII.
posted by MtDewd at 1:46 PM on November 14, 2015
Face down, 9-edge first (which BTW I totally agree is the way to go here) is the way cards are put into a reader. (12-edge first is for the punch side)
When you put cards in the reader, column 1 is on the left. Of course, you may have chosen to put your punches in column 80, but I miss the point of that.
So I give up. What do the punches represent? I didn't bother to go 12*80, but tried to represent a 12-3-8 for x'4B'.
Hard to tell what this is. I assume it's not literally a 10*43 card, so what do the punches mean? 1-4-5-8?
Except for column binary, you're not allowed to have more than 1 punch in rows 3-7, and I don't recognize anything in column binary here, either.
In any case, it better not be ASCII.
posted by MtDewd at 1:46 PM on November 14, 2015
I recall learning about Amdahl's Law in my computer architecture class. It's sad to see the demigods of computing die off one by one.
posted by shponglespore at 2:11 PM on November 14, 2015 [2 favorites]
posted by shponglespore at 2:11 PM on November 14, 2015 [2 favorites]
I have two of those red Amdahl mugs. One I keep on a shelf. One I drink from.
And yes... the coffee does taste better from it.
posted by PROD_TPSL at 3:28 PM on November 14, 2015 [6 favorites]
And yes... the coffee does taste better from it.
posted by PROD_TPSL at 3:28 PM on November 14, 2015 [6 favorites]
.
posted by Lesser Spotted Potoroo at 4:54 AM on November 15, 2015
posted by Lesser Spotted Potoroo at 4:54 AM on November 15, 2015
« Older Why the University of Tennessee has no Homecoming... | Korespondent Julien Bryan Newer »
This thread has been archived and is closed to new comments
posted by fairmettle at 7:37 AM on November 14, 2015 [1 favorite]