|
The_Groove posted:my old computers are an Indy and a SGI Visual Workstation which was terrible except for the 1600x1024 monitor. it ran Windows NT 4, had "Cobalt" graphics (still made by SGI at least) and had some unified memory thing where the CPU and Cobalt could share the system RAM. it's still heavy as poo poo like any other SGI system. i forget if it was that or the O2 that came with a VRML demo of itself, where you could click around and see the internals they figured if they put engineering into Windows NT hardware the way they did their proprietary unix stuff, they could sell Windows NT at the prices the old machines commanded. result: a $15,000 windows NT workstation that performs about as well as a $5,000 system with a wildcat card in it this sold exactly as well as you think it did
|
# ¿ Jan 19, 2016 01:57 |
|
|
# ¿ May 10, 2024 13:31 |
|
Pham Nuwen posted:adm-3a terminal connected to my pc and playing some sort of mud iirc: how did you hook a current loop terminal up to a modern pc
|
# ¿ Jan 19, 2016 01:59 |
|
Sniep posted:from the yes they are both from 1999 go figure
|
# ¿ Jan 19, 2016 02:30 |
|
Doc Block posted:"Naw, it won't matter if we basically give nVidia and Microsoft our 3D patents, fire most of our high-end 3D graphics hardware team, then sell what's left of them to nVidia!" didn't this happen after the first bankruptcy Doc Block posted:edit: it'd be interesting to see how different things would've been if Rick Belluzzo had never worked at SGI. it would have turned out exactly the same, minus the flirtation with NT sgi's business model was in serious poo poo from 1997 on, and they refused to change their culture
|
# ¿ Jan 21, 2016 03:37 |
|
Doc Block posted:Kinda. It was mostly the fault of "Chainsaw" Rick Belluzzo, who gutted HP's propriety Unix and CPU development, then went to SGI and did the same thing, both times in favor of Windows NT boxes running on Intel CPUs (Itanium, LOL). He then went on to work at Microsoft, which wasn't fishy at all, no sir. hp-ux and hppa were both also-rans in dying markets. neither one ever dominated. neither one was poised for any kind of growth. itanium was HP's hail-mary plan to outsource cpu development, and it worked pretty drat well. they went from spending billions a year to tossing intel a few million for marketing costs. they got exactly the chips they asked for, using their existing CPU design strategy. (the early itaniums were even meant to be binary compatible with hppa!) they are still getting new chips produced today, in 2016, but they let go of their last cpu guys in 2004. i don't think they could have wound it down any better.
|
# ¿ Jan 21, 2016 03:49 |
|
itanium was very successful for hp. it was a dreadful failure for intel and everyone else who touched it. most particularly the first itanium, "merced." merced was four years late, and partners finally put a gun to intel's head and demanded a date. to hit the date, intel didn't have time to actually give any chips to partners. instead, all, ALL itanium 1 hardware was intel engineering samples. the only variations by partners were paint and badges. the ultimate in lousy badge engineering. there were also some os differences. HP and Hitachi ran a half-baked port of HP-UX written for an emulator. IBM's ran a half-baked pseudo-AIX based on SCO UnixWare. Dell's ran who knows what, since Dell didn't own an OS. Windows XP? who cares. HP i2000 Workstation (the only click for big image, because HP is the only vendor without shame) Dell Precision 730 IBM Intellistation Z Pro Model 8964 (ibm loved wordy names, back when they made stuff) Fujitsu CELSIUS i800 fuji didn't even paint the drat thing, left it intel white i couldn't find any photos of the Bull or Hitachi units, which surprises me. IBM had a photo and they only sold like 50 systems, total, across all itanium models.
|
# ¿ Jan 21, 2016 04:16 |
|
eschaton posted:also even though SunFreeware shut down in favor of a paid service, there are still some SunFreeware mirrors around so I don't have to hunt down a copy of Solaris Studio 9 just to build stuff, though it might be interesting to try openCSW still has archive snapshots of their solaris 8 and solaris 9 package trees. i know this because i have been trying to get solaris 8 working on an old sun also almost any version of sun studio from the past ten years will work on solaris 9, you don't need specifically 9.x
|
# ¿ Jan 21, 2016 04:17 |
|
eschaton posted:why not 9 for your sun? and what model sun? i have more software that runs on solaris 8. also the glory that is OPEN LOOK it's an ultra 80, so solaris 8 is ~*~ period appropriate ~*~ eschaton posted:I thought it was like 8-9 that run on Solaris 9 sun studio 11 still supported solaris 8 until very recently. solaris 9 is probably still supported under current versions edit: the last version to support sol 9 was sun studio 12.0 http://www.oracle.com/technetwork/server-storage/solarisstudio/training/index-jsp-141991.html
|
# ¿ Jan 21, 2016 04:31 |
|
eschaton posted:what about patches? I have the Solaris 9 recommended patches from 2006 which are all currently failing with error 5, 99% of the time, error 5 means you didn't go to single user mode before applying the patch i find old-timey solaris refreshing because it's so drat simple. but simple is a knife that cuts both ways. simple is also crude. lol @ shutting down the system to apply patches. eschaton posted:and one of the things I saw online referenced Patch Check Advanced which seems worth checking out… oracle doesn't give out patches for free anymore, so download tools are not very useful unless you have patch sets from a prior employer or something, you're SOL Notorious b.s.d. fucked around with this message at 06:22 on Jan 21, 2016 |
# ¿ Jan 21, 2016 06:19 |
|
these are separate questions, so i will address them separatelyFarmer Crack-rear end posted:what was the advantage itanium had over other architectures? i don't mean x86 but the other architectures that companies had already been using for their 'big iron' intel was then, as now, the 800 lb gorilla in the semiconductor industry. they could make bigger chips, cheaper, on more advanced process nodes. hppa/pa-risc was already very competitive. everyone knew that hp was working with intel on a pa-risc competitor. and pentium pro was eating them all alive. hobbyists barely remember ppro because it was expensive and regular people didn't buy them. but it was a real eye opener for the industry: a $3,000 chip that could humiliate the processors in $50,000 workstations. intel had declared an interest in conquering RISC, and nobody wanted to bet against intel. Farmer Crack-rear end posted:i guess my question really is why did so many companies buy into itanium and then keep buying itanium even after the initial flop hp was the only company to stick with itanium after its flop nature became apparent.
|
# ¿ Jan 21, 2016 17:57 |
|
Pham Nuwen posted:from what I remember of my computer architecture class and other sources, itanium's VLIW architecture was supposed to be super hot poo poo just as soon as they got optimized compilers written for it. the buzzword was EPIC: Explicitly Parallel Instruction Computing. the idea was that you could move pipelining / scheduling / branch prediction into the compiler, and issue an instruction stream that would maximize the use of functional units this didn't work at all. not even a little bit. itanium 2, the first chip you could actually buy, was superscalar, defeating the whole point of EPIC. Pham Nuwen posted:hp ported VMS to itanium though which is kind of cool, even though VMS is the finest metaphor for government bureaucracy ever made (counterpoint: VMS actually functions ok once you figure it out) hp sold openvms to a spinoff, VMS Software http://www.vmssoftware.com/ mission #1: port to x86
|
# ¿ Jan 21, 2016 18:02 |
|
SYSV Fanfic posted:What computer can I get today that will be fetishized in 20 years despite being overpriced garbage. I don't have the mid six figgies to dump on a $2,000 trash can. it'll be the trash can. rare poo poo no one can have is a sure bet for fetishization the sun desktop i've been trying to get working would have been pretty expensive back in 2000. using the figures from sun's own price list: A27-ULD4-9V-1024AQ WS U80/4X450,IFB,1GB,18G 30,795.00 H X7005A OPT MEMORY 512MB (2*256MB) 2,600.00 H X5240A Opt int Disk 36.4GB/10k USCSI 2,650.00 H totals out to $54,245. in 2000 dollars. that's like $75k in today's money. for a desktop.
|
# ¿ Jan 22, 2016 23:03 |
|
eschaton posted:oh yeah, I'm confused by OpenCSW because I wasn't able to easily find a "this is what we have for Solaris 8" listing, and they talk about 8 being deprecated, it really seems to be for people who are running the latest Solaris opencsw used to do releases like a linux distribution. the last version for sol 8 was "dublin." i think sol 9 got at least one more. http://ftp.heanet.ie/pub/opencsw/dublin/sparc/5.8/ edit: looks like they have a precompiled gcc 4.3, so that's cool
|
# ¿ Jan 26, 2016 03:10 |
|
eschaton posted:just helped a coworker resurrect a Fujitsu S-4/Leia2 you found the only japanese openstep user i didn't even know there was a japanese localization
|
# ¿ Jan 27, 2016 04:46 |
|
eschaton posted:dude I work at NeXT well now i understand the bizarre defenses for xnu's crimes
|
# ¿ Jan 27, 2016 05:01 |
|
Jim Silly-Balls posted:beos owned, I unironically paid for multiple versions of it and used it as my primary os for a couple years. when it was new, beos was better at multimedia and multitasking than Windows, and it was better at being an actual usable os than any linux at the time. beos was a tech demo, only gassee could imagine it was a viable operating system there were a ton of cool ideas running around in there, but they couldn't get the basics working: a network stack that crashed constantly, the lovely web browser, half-hearted posix compatibility, etc etc
|
# ¿ Jan 29, 2016 02:35 |
|
Doc Block posted:But they had a nifty multithreaded GUI (that was a pain to develop for IIRC) but demoed well thanks to their kernel's quasi-real-time scheduler (video keeps on smoothly playing while you drag the window around OMG!!!). the great part of this demo is that linux and windows NT could play the same trick on the same hardware, because they used a dual-processor host with scsi drives no poo poo the concurrency is 100x better than a regular PC
|
# ¿ Jan 29, 2016 03:40 |
|
error1 posted:Not really, after Haiku developed a package manager the guy who ran bebits.com shut it down, killing the primary source of beos applications haiku is still way more alive than any of the weird amiga-masturbator projects are. it runs on ordinary x86 hardware with the right graphics card, not some ppc demonstrator board from an oem in europe that only existed for six months in 2005. on the software side, beos was posix-y enough that a substantial amount of linux desktop stuff gets ported. haiku is definitely a little bit silly and dated, but it's not completely useless
|
# ¿ Jan 30, 2016 00:22 |
|
apple was also in negotiations to license solaris. a real unix with a real compiler and a non-poo poo kernel what a world that would have been
|
# ¿ Jan 30, 2016 02:32 |
|
Jim Silly-Balls posted:there was an nvidia driver that worked with my riva128, even. for 3D acceleration(!) and there was an office compliant set of apps as well. on my p2-300 dell it absolutely flew, and could play most any media you threw at it, smoother and with a higher bitrate ceiling than Windows could. on a pentium II, a boring old linux desktop could play video and counterstrike on the same pc at the same time, even
|
# ¿ Jan 30, 2016 02:36 |
|
~Coxy posted:next was also a "real company" with $50m in revenues and had WebObjects which seemed like a good idea at the time i don't think the failure of WO had much to do with WO itself apple didn't have a lot of legs as a company hawking middleware to developers. it was just not even close to on-brand. no credibility then, later, the bottom dropped out of the market. when was the last time you heard someone ask his boss for tuxedo / websphere / jboss licenses?
|
# ¿ Jan 30, 2016 16:00 |
|
Doc Block posted:IIRC wasn't Craig Federighgihi the head WebObjects guy at NeXT, but then when Apple switched it from Objective-C to Java he noped the gently caress out and quit, only to get rehired later? the webobjects java stuff started in 1996 the objectiveC crap wasn't dumped until 2001 timeline doesn't work also objC had precisely 0 advantages for a web application. they dropped objC support because it was unwanted. nobody was gonna learn a new language for no benefit.
|
# ¿ Jan 30, 2016 16:01 |
|
i wonder why the graphing calculators didn't end up in the agilent side of the business if they had, they would probably still be in production. and $500 apiece
|
# ¿ Jan 30, 2016 19:09 |
|
Doc Block posted:So about the time Apple stopped caring about WebObjects. apple still uses webobjects. they only stopped selling it in the recent past. i wouldn't say apple stopped caring, it's just that the outside world never gave a poo poo at any point
|
# ¿ Jan 30, 2016 19:26 |
|
this is kind of a chicken/egg problem which started first: apple's indifference to enterprise sales, or enterprise customers' disinterest in apple software products
|
# ¿ Jan 30, 2016 19:59 |
|
an open source webobjects would be really cool
|
# ¿ Jan 30, 2016 20:58 |
|
Panty Saluter posted:it used an r12000, which is the same family as the playstation cpu, the r3000. that amused me and upon further research i found out the r3000 debuted in 1988 it wasn't cheap enough for a james machine until six years later i guess the playstation was under development for like a decade. if it had come out on time it would have been a very expensive console. the remarkable thing about the psx is that it's really a very ordinary architecture. it looks a lot like an early 1990s unix workstation under the covers. no weird coprocessors, no funny memory, just a RISC cpu and a bus and some memory and some devices hanging off the bus
|
# ¿ Jan 31, 2016 18:53 |
|
eschaton posted:the DECstation 5000 running Ultrix was a p nice MIPS R3000 system for its time, the R2000-based DECstation 3100 ("pmax") was a bit of a dog though the decstations were reasonably nice hardware held back by literally the industry's worst unix. ultrix was horrible. it didn't even have shared libraries. if you loaded "xterm" and "xclock" simultaneously, you got two copies each of libX11 and libxcb in memory. this was a really, really horrible situation on a machine with 8 or 16 mb of RAM. dec had a port of osf/1 ready for the mips systems, but never gave it an official retail release. (yes ultrix was so bad that it made osf/1 look good)
|
# ¿ Jan 31, 2016 22:54 |
|
i have a decstation 5000/240 and a copy of osf/1 mips, but unfortunately the only scsi chassis i had that would talk to the 5000/240 went and died on me. no 1990s RISC microkernel fury for me the osf/1 mips "installer" almost works inside gxemul. it successfully boots the kernel in a miniroot, but the scsi tape emulation is not complete enough to finish an install.
|
# ¿ Jan 31, 2016 22:57 |
|
Panty Saluter posted:oh. maybe? or i accidentally looked up the wrong computer. idk what actually used R12k? all i can think of off the top of my head is SGI and Siemens/Nixdorf
|
# ¿ Jan 31, 2016 23:14 |
|
Citizen Tayne posted:Yeah, I was gonna say, by the time that PDP-11/73 rolled around the PDP-11 was functionally obsolete and any sales were on the strength of compatibility with existing software/applications. this is not really true. the performance of an 11/73 or a pro/380 was pretty good for the money. people were using them as workstations and industrial controllers instead of micro-mainframes with tens of users.
|
# ¿ Jan 31, 2016 23:18 |
|
Citizen Tayne posted:Actually, what I said was very true. They were industrial controllers in the eighties for the same exact reason you still find 486s with RS232 output doing the same thing today. the original j-11 chipset was from the late 1970s. in 1984, cranking over at 15 MHz, it was pretty drat fast for the money. for perspective the 8 MHz 80286 was the fastest intel chip at the time. (and those sure didn't run unix very well.) Notorious b.s.d. fucked around with this message at 23:36 on Jan 31, 2016 |
# ¿ Jan 31, 2016 23:32 |
|
here is a screenshot of osf/1 failing to boot in gxemul from a hacked-up disk image. i'm not sure where i got the disk image. maybe from a physical 5000/240?
|
# ¿ Jan 31, 2016 23:51 |
|
Pham Nuwen posted:somebody gave me an octane2 and it was R12k. at that time none of the free *nixes supported it (guess netbsd does now) and i couldn't find irix. didn't have the weirdo hard drive sled either so even if i found irix discs i couldn't have installed it on a hard drive. the sleds turn up on ebay from time to time installing irix from cd really, really sucks but it can be done with enough patience. it's usually easier to set up an irix install server on freebsd / netbsd / solaris. linux does not work very reliably due to nfs / rsh quirks. at a minimum you will need two sets of discs: irix 6.5.22m or 6.5.30 "overlays" (4 discs) irix 6.5 base media (4 discs) to do anything useful you will also need: idf/idl mipspro
|
# ¿ Feb 1, 2016 01:06 |
|
eschaton posted:on the other hand I've heard playing N64 games on the SGI workstations used for development was an awesome experience, 1280×1024@60fps—possibly in stereo 3D!—and some devs added network code just for their own amusement the only sgi / n64 combinations i've ever seen or heard of were indy-based, so there wasn't gonna be anything 3d @ 1280x1024, too drat slow the n64 graphics hardware would have been considerably more capable than the 'newport' graphics in an indy. practically no hardware accleration for 3d operations
|
# ¿ Feb 2, 2016 02:32 |
|
Pham Nuwen posted:There are 3 PDP-11 systems sitting in racks at my work's reapplication warehouse... Two 11/44's and what I think is an 11/73. Also a couple RX02s, a VT240 sans monitor, and I think a disk drive. do it! i think an 11/44 will run rsx-11m if it won't do 2.11bsd. fun either way
|
# ¿ Feb 4, 2016 21:34 |
|
the scrap value is near $0 and odds are no one will care it's easier to ask forgiveness than permission etc
|
# ¿ Feb 4, 2016 21:38 |
|
Smythe posted:its called an ikea bag and the trunk of your car in this case the ikea bag might need to be a 6 person tent but yeah, this
|
# ¿ Feb 4, 2016 21:38 |
|
trilljester posted:I found a couple of old Sun Fire v440's at work in storage the other day. Anything fun I can do with them, or are they trash? they are middlin' size sun boxes from the good old days. four processors, probably a fair bit of RAM, more than fast enough to run a modern OS comfortably. but they are pretty slow and power hungry by today's standards. a fast laptop will have as much or more i/o and cpu oomph than a v440. i don't know what you, personally, might want to do with an old sun box. i'd be tempted to install debian/sparc64 and see what actually works on the sparc port. (guess: not much.) other fun things: see if anyone has backported smartos/omnios to spparc; get solaris 8 or 9 up and running for nostalgia value; see if solaris 11 will install despite dropping v440 from the HCL
|
# ¿ Feb 4, 2016 22:50 |
|
|
# ¿ May 10, 2024 13:31 |
|
well yeah if you have no interest in solaris or sparc, there's nothing innately cool about an old sun box. old computerin' has to be driven by either historical interest or nostalgia
|
# ¿ Feb 4, 2016 22:55 |