Wow! This brings back memories. I worked at the IBM East Fishkill plant back in the 1980's. I worked in the manufacturing technology areas where the ceramic modules were created. At its peak, the plant employed about 15,000 people, and the weekly production could be driven away in the back of two station wagons. An unbelievable amount of engineering went into the design of these modules and the manufacturing processes to make them. The ceramic green sheet material that formed each layer of the module was created by a long sheet roll process on a machine that was approximately 63 feet long. The roll was then taken to a die machine that precisely cut the dimensions of each sheet. The sheets were precisely punched for layer-to-layer vias, then fed into screening machines that spread molybdenum paste through masks to create the pattern on each layer. The next step was a stacker that aligned and stacked each of the 63 layers so that vias from layer-to-layer would align. The stack was then pressed and readied to go through the sintering process. The sintering or firing, of the ceramic substrates was done in huge sintering furnaces that had a hydrogen forming gas. Each furnace was loaded by robotic arms that moved assemblies that weighed hundreds of pounds. When the process was running, the excess hydrogen gas was burned off in a stack. It looked like something from the Wizard of Oz. Occasionally, there would be an air leak into one of these furnaces where the oxygen/hydrogen ratio would be just right to cause an explosion of the entire furnace, shutting it down for an internal rebuild. The ceramic module would then go to the thin films processing section, where it was CMP polished for flatness and the multiple thin film layers were deposited using technology adapted from wafer manufacturing. In the newest generation of TCM, there were six layers of thin film redistribution wiring on top of the ceramic, before the chips were attached. This process had to provide for corrections in lower layer yield defects, and had to achieve 100% yield over the entire substrate. There were multiple customization laser repair processes that added/subtracted single trace defects between the chip sites. Wow! I could go on, but I'm running out of steam... maybe someone else can pick up and describe the additional complexity in the manufacturing processes of this unique assembly.
Wow! Thanks for sharing this btw. I also saw a comment on this video of someone who worked at the same IBM plant mabey you could add this to that thread too
Kinda funny/fascinating reading your post when you talk about "Each furnace was loaded by robotic arms ..." and in another comment to this video is a guy who developed those robotic arms and talks about that. And all that was 30 years ago.
Hey Edward, I'm actually a mechanical engineering student in the Dutchess County area and I've been looking into IBM in Poughkeepsie. What you explained is probably one of the most detailed and eye opening accounts of what happens during the manufacturing process, and has definitely increased my interest about this industry even more, thank you for that.
About the processing power: The 390 is not a scientific computer but a business machine. It was not designed to run calculations but to run data through. "Want to run a batch job over all 50 million accounts of you bank's customers? Sure, will be done in 5 minutes."-type stuff. Or to connect a couple thousand end user terminals to one system and have guaranteed response times below 1/10 second. The missing chips are due to increased load, as far as I know. Those modules were highly redundant, so the advertised capacity probably was only half of what was possible with the extra chips as backups and to compensate for production errors. I wouldn't be surprised if a third of those chips didn't work in early production. But there were enough of them for the complete module to still be functional and have enough backups for plenty of them to die later. Later when the yield got better they could leave out some chips and get the same result.
As it was replaced with today System Z with dual everything (all operations are done on tow separate machines and compared then) It newer was intended to be fast system but extremely reliable
@@DavePoo What is? the S/390 ran OS/390 (previously called MVS) or VM, wasn't unix or even a flavour of. Mind you you could run nearly everything under VM,
I thought about taking it apart for years. So glad I did not. When I discovered Dave I immediately hoped he would enjoy it for mailbag. I learned so much more from watching Dave disassemble this that I would have doing it myself.
So, you're the fellow who sent him this? I think that you can count on Dave as being your friend for life now. And anytime you two visit, the pints are on him! Thank you for giving up what must be a prized possession for Dave to work on. Truly beautiful and amazing technology and a historical dive back in time.
Wow yea the memories are flying now!! I worked at I.B.M. East Fishkill N.Y. for 31 years and for about 2 years during the late 70s to about 1981 I worked in one the many lines that made these TCMs when the TCM had 33 layers!! now about the Molybdenum , The MOLY was deposited by an electroless plating process that used 3 or 4 steps on the pads that supported the pins, the final process plated the pin pads Gold. The internal wiring was made of Moly paste that was "screened onto" each individual ceramic layer to "wire" the chips together!! The real trick was to keep all the via holes aligned for each layer to have continuity with the layer above and below itself when the TCM was being "fired" to cure the ceramic layer's that formed the TCM the via hole would misalign and cause loss of continuity in areas of the TCM due to inconsistency of the shrinkage rate of the layers of ceramic!!!!! This was a HUGE PROBLEM for a while but the engineering departments figured it all out and solved this daunting problem and done the fix!! In later years the layers went from 33 to 64 and the via problems remained solved due to changes in the ceramic mix!! And yes I seem to remember the name Knickerbocker as one of the top engineers in the MLC (Multi Layer Ceramic) program but never actually met the fellow! John A Bellas KC2UVN
I was going to make a similar comment. I worked for IBM in the 80's/90's too on 8100's and 3745's, in the factory as a test technician. I also worked on some CNC milling machines to +/-2 microns. The 3745's had TCMs, though were air cooled. Sys390 9121 (ES9000) was my third mainframe with 256MB of real RAM! We went from 24bit to 31bit addressing. WOW! 16MB to 2GB address spaces for MVS/ESA 4.3.3 (Which initially didn't run natively, we have to run it under VM/ESA). ESCON channels too. It was one of the very first ES9000's in the UK.
They are called TCM. This brings memories :). I use to know so much about them from time I have worked in Dismantling IBM facility in the Germany. My job was to take special parts from preowned machines and package them for the clients that could not afford new or refurbished parts. TCM's where most expressive parts ever. Some of them costing in millions of DM (Deutsch Mark, German money before Euro for the new kids :P ) . I have build those out of the main frames. There is whole procedure for taking them out. You need special key that unlocks the cooler block. There is a special pen like tool if pins bend. There is a special conductive plastic case with special foam inside for caring them. There was a huge call it a main board with 9 sockets. and it was massive. 3x3. Some frames have had two of those one on the top one on the bottom. There was a special tool that you screw in those to use it as a handle when taking those big boards out. IT was two man job tro carry it out of the frame. I was young and stupid and was doing ti alone until my back gave up. board with TCM's in it was 90KG. It cost me months to fix my back and it involved putting electrodes and some current true my back. First you take a hex bit and unscrew all water blocks from all of the TCM's. Then you would Pull out whole tubing and cooling blocks. then you would unscrew huge power rails from the so called main board. Then you crew in handles and unscrew the mounting screws. You take the thing out and place it on the table. Needles to say you are grounded with a wrist bracelet and you wear special anti-static shoes and on grounded table and special painted floor that can not generate static. Teh you unscrew TCM and then you use special key tool to unlock the TCM from it's socket. Turning the key was moving the TCM sideways unlocking the pins. Then you inspect it and make sure pins are all good and correct them if needed. You stick your bar code label on the side so people can trace your work.Then you place TCM in the ESD bag and then you place it in the special container. Then you lock it and seal it with special wire and lead seal with special tool. To open the container you have to cut the seal or wire off. I remember looking at part list and pricing. And some of those costed like 4,5 Million DM. I think Cheapest where like 1 million DM. Some of the TCM's where processor units having in them 11x11 CPU cores. Each core have had a little cylinder with a spring making a connection to a silicone core for direct cooling and the top block would be like heat spreader. Water block would crew on it. Imagine those pig pumps. Those have had huge electric motors few KW each. Tubing was a maze. Individual tubes where like size of a little finger and they would all collect to bigger hoses. Some TCM's where memory only and would have up to you guested it 11x11 memory chips. So every TCM have had it's function. Whole system was made of 7 main frame units/closets. One of them would have two gigantic pumps and special reservoir that decriminalizes liquid used in the cooling system. BTW it was cooled with some alcohol mixture. Now imagine 6 out of 7 frames would be all full stock with 18 TCM's and one frame would be Cooling. Yes no permanent storage!!!! Whole system needed storage solution that was again contained in more frames/huge closets. Imagine costs when Bank or some Government institution orders a system. And yes many have ordered multiply systems in the cluster working as one back then gigantic system. And yes client side units where thin clients connected with Token ring adapters or later BNC 10baseT networks and having black and green CRT monitors up to big 14" :). They have had big mechanical keyboards with heavy metal base. I was taking them out of the 3900 series. There where some systems for the storage and they have had floppy disks drive. 8" fasynchronous
Jasper Janssen back in the time the use to cost between 1 and 4,5 million Deutsch Marks. Even without inflation figured out that is 500K to 2,25 million bucks. One DM was 50 Euro cents and 1 Euro was same as 1 US $. IBM used to ask original prices for computer parts even 5 years after while those would cost nothing in the other shops. For example in time of first Pentium CPU main board from model 80 was still few thousand DM but it was 386 equipped. No one would even take gratis 386 CPU at that time 😀
NINEWALKING yeah, I’m a little surprised that even for IBM mainframes you can still buy refurb parts 30 years later. Although if there’s *any* company and *any* product, that would obviously be it.
Jasper Janssen those systems where build to work for 100 years and engendered to be repaired. They where so expensive that once you have them you need to use them as long as possible. Remember there where no server clusters approaching this power and almost no one was making any other solution. IBM have had practical monopoly on the main frame systems. Only later on Fujitsu Siemens started with som products but that was already competing wit ES9000 systems.
That's hilarious! IBM actually had an exchange program where you could bring in a defective TCM and they would replace it. The rate depended on the type but was mostly $22,500/ea. They would have suspected fowl-play when we brought it back dripping with 10W-40 from our failed oil change.
IDK man, now its tons and tons of spaghet running on the modern procs! We got dumber and Moore's law is over now. HW like this is so nice, and the software was typically as intelligent. Now we run everything on JS where 2+2=5... 🤦♂️
I checked. They have one that supports this platform as well as i7 and Threadripper AMD. Can also run a Pentium II off a riser board for some reason. $69.50 USD
I never worked on mainframes that used these back in my IBM New Zealand days. But they were so expensive that the IBM Maintenance Parts Centre had an aircraft on standby to fly replacement parts across the country, to maintain service level contracts. It was much more cost effective to have a small central stock and a plane to freight the parts around the country.
Stop the presses! Are those bodge wires left of screen at 17:39!? Also if you can’t get an X-ray or µCT of it a physical cross section might be easier to organize.
Looks like an ultrasonic weld; at first it seemed like they were configuration or to bridge unpopulated chips but no, there doesn't seem to be any pattern Genuine ibm certified bodge wires
Ken Shirriff - Let's take a chip die apart layer by layer with chemicals, then do a detailed circuit analysis of the photolithography and metal deposition layers. Dave Jones - Let's scratch the whole thing up with wood fibre paper towels, and rubber gloves, and then smash the dies with a huge pair of pliers...
5 microns flatness is only ~10 wavelengths of visible light. That's meh for an optical flat made out of glass, but absolutely incredible for what is effectively a multilayer ceramic circuit board on steroids.
Marc Berte I hadn’t watched the whole video yet when I read this and I thought it was a joke. “Surely it doesn’t have pistons, he’s just joking right?”
To take one from a v e's Channel. Release the shmoo! Remember back when they had all that goop inside of devices not necessarily what I'm called planning compound but I think that's what it was but it was more like just a jelly Flash paste felt almost like I don't know man he's or something let me a good way to say it but I guess it was some form of rain for Herbal potting compound I don't know but it was often used in things like some power supply bricks so I assume in my defense when do if thermal properties are pooling and or just to keep people from messing with devices because it was actually somewhat liquid and cover much or like and not necessarily A Johnson but more liquid than that I don't know what would be the best way to describe at the anything I can think of at this time at first because I saw it in some battery powered equipment I slept that the batteries and exploded but they were in fact and then I remembered later singing and other equipment that was the first couple times I saw it though so yeah surprised me back then Maine this was like 80's and 90's naming somewhere in there. I do remember one of those things was like it sort of looks like a dustbuster is kind of deal but was rhetorical flashlight possibly from Radio Shack? Reminded me of of a dustbuster except a bit flattered and with a square lens on the front. And when I realized what was I did wind up for parent the batteries were shot wound up getting it those replaces cell rechargeable batteries from RadioShack for the purpose actually I was able to put in larger cells then it had originally either the night size up or the next I think it was Sub C and went up to either a c-cell or a D cell battery cannot remember long time ago but Walmart 1st hacks to increase battery life in something well at least Mike that bit whenever built-in rechargeable battery of that click nicad not nickel metal hydride back then that was all that was available in that type of battery as far as I would know from personal experience. Other than a Bee Line acid but not something like you'd use in something like that. People assaulted rechargeable tools were new who sings Even free at prior to that however there were actually something that you would call cordless tools except they were not truly cordless but you did not require Hunter and 10 volt power to run them just recharge them but they did have a cord going to a battery pack it was a rat acid battery pack I think it was more like on a belt I'm not talkin about the system for Black & Decker either that you was one of those referral batteries with the cigarette lighter receptacle which by the way is that battery pack was actually designed for another product at one time and I were some of those like in at leatherette case as well back in the old good old days. Also rechargeable radios and stuff go back further than you think as well. But for the virtual cordless tool system I do believe is that the battery was maybe on a belt pack of some sort but for the original cordless tools system I do believe that the battery was maybe on a belt pack of some sort I don't remember how many volts are anything laws but someone I know used to have this system years ago and they told me about it you could say he was one of the greybeards in whatever he worked on. I used to have a old to tube AM radio that got excellent reception I actually brought the power supply for it. I literally breadboard this thing by the way the word breadboard actually comes from using a cutting board as a building base to prototype something that's what was first used to do something like that when something was designed can't remember the entire story but that's for the phrase breadboard comes from. If I remember correctly I use standoffs I think I made from sorghum sort of TV antenna I think it was an old rooftop antenna I want up cutting sections off the antenna elements to act as my standoffs is there were even wood screws as well as machine screws holding it all together and yes fasten hook Clips as well for connections need both antenna and ground and later on I built the symbol amplifier and I use the old PA speaker as well I used a pair of antique headphones as well. I cautioned my folks if they were using it not to touch the amplifier when it was on it did not have a chassis it was just mounted on a board everything was exposed and yes I'm a line voltage as well it later on I made a chassis out scrap pegboard material when we put up Peg boards.. later on I use that stuff for terminal strips as well for some projects It worked out great for building things that need to be connected to other things as well. For that purpose. I also want spilt the radio for someone that fit into a junction box cover don't remember the size of it. Also wants built a landline phone into the junction box cover as well using Salvage phone parts. one of my friends gave me a power strip that his grandfather made. It had the old Wire like that would be used for two men knob a light socket as a fuse holder twist time for rotary switch not the push button switch of the time these were the little round ones that were surface mounted. There were some receptacles in addition to Edison screw sockets those were in parallel the idea was it could be either used four things with an actual plug or something that had to have a screw show plug on it hey Carrie r u v a light socket. I actually wound up putting a damper in one room for a night light because it was hard to see the switch at night in that room the other one had one of those screw base adapters to take a standard plug later on I ran acrossed neon indicator bulb with a standard screw base on there and use that for the night light. also I had a couple from other people that use the round Bakelite type receptacles like you would use in barn or other wiring with surface fan wiring using the nonmetallic sheathed cable. I had both grounded and ungrounded versions of the once which even had a switch as well. obviously that was only used for certain things like where you hand a plug-in say stereo components or something where you wouldn't pull much power because a chord on some of these was only about 18 gauge possibly 16 gauge In one or more cases. what are friends had one that his grandparents had used probably their parents hard to say that have been built that took d screw base Plug & went to more than one light socket so essentially a power strip before that was a thing it was mounted on a piece of wood that been salvaged from something else and also there was switches on there as well. also has anyone ever seen a hard-wired service about night light evidently these we're sometimes not on stairways some of these were even had internal switch although they weren't available also with plug pins but they were the exact same socket assembly they accept screws instead of the for the assembly attached. This was like the brown Bakelite? also what's the deal with those old round light sockets is mounted on usually a black metal plate with a ceramic socket often times has a screw-on part on top of the socket I just see those quite often in places sometimes it's just one of them in the entire place I don't know if it's coincidence or not. A lot of them are others are just ceramic the standard sockets either pull chain or not. although sometimes I've seen those like in a garage or something but I stared ones are in a house and sometimes both in the same area. also there is one of these at my parents place in the garage however there is no evidence of a switch ever being there and the wiring just seems to go nowhere there is a few holes where it might have gone but no evidence of it ever being there old metallic flexible conduit of some sort. possibly no evidence of an even leaving the garage which is just weird. Although there's one on thing in the house downstairs there is one circuit has ran in Black conduit which is only one circuit like that. It's got the old School round box again black. and this goes through the side of the house but then heads up through a conduit body and up the side of the house to the upstairs We still don't know what it feeds but it obviously if it was possibly done later? Is everything else is ran inside Possibly from the 30s or 40s? this dude feed into in the junction boxes in the basement that was used for the lighting I suspect that it's possible that may have fed the garage at one time but had but had been redone but had been redone and repurposed for the upstairs? also that conduit down in the basement tests live with a no-contact tester so little skeptical about grounding? things have been checked over when the wiring was doing but this has me scratching my head what was the purpose of that conduit why would just one circuit that way. also most of the wiring in my house is the old BlackNonmetallic sheathed cable I'm not sure which brand But it's all the same sort of stuff I don't know if this is rubber covered or not but could be I'm just not familiar with itBefore my time no evidence of knob and tube whatsoever. however original telephone wiring was to however original telephone wiring however original telephone wiring wasI'm single pair Twisted wire Black in color Supported with little Tak like wire holders possibly wood or plastic or similar material + 2 Connector elongated terminal blocks 4 Connection pointsAnd two or three places in the basement Along the Floor joist. And one Where the wiring came into the houseOh and one more thing there was some iron wire And you're at one time that was insulated Not aluminum but iron yeah evidently that was a thing.
I worked at IBM in Poughkeepsie in 2000 and they were still making these large server modules. This is when they were transitioning to all copper interconnect. The ones I remember were liquid cooled and a refrigerator sized machine had rows of them. The AC unit that cooled it was larger than the computer! They had these huge shake tables that would test these systems for sensitivity to vibrations in another building.
From Roy Longbottom's benchmark web site: 1991 IBM ES/9000-900 with 6 CPUS at 111 Mhz: MIPS = 248, MFLOPS = 2,664, cost = $22,600,000 Raspberry PI 3, 64 bit, 4 cores, 1200 Mhz, ARM v8a: MIPS about 6,500, MFLOPS about 480, cost about $30
Yep, heck of a lot faster per $ these days! But as Dave said, even back in the day mainframes like this were beaten for MIPs by desktop CPUs. So why did mainframes sell, when a few PCs could beat them for speed, even in 1990? Try running an airline reservation system on that RPi 3... with 99.99% availability, mind you!
@@GooligumElectronics It was not just running fast. These were real multi user systems. And they had all the fancy periphery like "huge" mass storage drives. At this time a "PC" was DOS5 or 6, maybe Win3 or some early UNIX, LINUX. These things were just the tip of a whole eco system not just one fast running CPU.
@@GooligumElectronics Precisely. Those things were used to run a uni campus, stock exchange, or huge bank, stuff like that, not some shit Windows PC. Therefore comparing plain numbers, like those PC nerds like to do, is totally pointless.
@@AmauryJacquot That's the whole reason the mainframe still keeps going. Because at 30,000+ I/Os per second you can get transactions rates over 100,000 txns/sec and that's what's needed to run a bank, an insurance company or a supermarket. The piece that gets that running is the fibre-optic channels for the connections between the processor and the disk units. When Dave's 9121 was current that was about the time that we moved from a 4.5Mb channel to high speed fibre. Next move was from big fat disk drives (like the one Dave tore down in ruclips.net/video/CBjoWMA5d84/видео.html) to RAID arrays using inexpensive 3.5" drives. The I/Os to the mainframe all became emulated. It's all moved on in size in the thirty years since then. The machine room was the size of a football field in 1990. (when I started in 1981 it was the size of two football fields.) We can get so much more than that in an IBM z15 mainframe that will fit in the cupboard under my stairs now. There's also less heat and less power usage than back in the day when the mainframes were all water cooled.
When I was at University in the late 1980s "we" had an IBM 390. A friend doing a PhD in Chemistry got hold of a PC with a 286 processor. Running simulations on the 390 required him to book processor time a few weeks in advance. He could do the same thing on the 286 but he had to let it run for 24 hours or more. Still had it done faster than booking time on the mainframe.
As for why the pads dont come off, the solder is very soft as its 97% lead. We used molybdenum alloy metallization on ceramic components at my last job, and properly done it could easily exceed 30ksi of tensile strength before it delaminates. I did mechanical testing on small 3/4" ceramic rings brazed to kovar and they could take about 1/3 of a ton before they separated if proper braze material was used. Even more if strain relief washers were used in the joints. Since they have a fair amount of metalization anchoring the pads down they are considerably stronger than the lead solder. Looking at the cross-sections I see all the usual suspects for bonding to ceramic substrates.
I worked in a warehouse that reconfigured 3090s. Interesting thing, they ran on 400hz power. We had motor generators. They were also water cooled. Never realised the cost 😱
@@moconnell663 Aircraft spec 400Hz for the sake of smaller transformers/coils/caps to save weight. Lower frequency has less loss over long distances, but that's not a big problem on an airplane like it is on a whole power grid.
@@Draalo "Talk nerdy to me... Be gentle... I'm a n00b." Oh man, I'm sure some Hollywood guy could come up with an entire series based solely on that line.
I love how exited and enthusiastic he is about this thing. I have no idea about computer science so I've got no idea what he's talking about. But I am a fitter and turner and this is really impressive from a mechanical engineering view point
I bet my old roommate from back in the day could tell you about those substrates. He used to work as a mechanical engineer at IBM East Fishkill, NY, where they made these things. I'll check with him and see if he can say anything about it.
Hello there! Used to work there myself as a software engineer! Started in 2000. I probably could have shoveled a trash can of this back in the day. There was so much tech everywhere. worked in the 310 building.
First thing that came to my mind when I saw it open was this: www.thingiverse.com/thing:2720375 I've been wanting to make one with metal instead of printed plastic.
IBM were the best of the best in computer manufacturing magic, starting with Appolo-era computers, back in the 60's. We can recognize their design philosophy in this module. Lots of squares, all seemingly identical, but with different underlying functions.
I absolutely loved this teardown video. I was a user of the IBM 3090 ES9000 systems back in the days when this was the latest thing. I did stress and performance testing of Prodigy's systems using IBMs Teleprocessing Network Simulator (TPNS) which was a VTAM application that I used to simulate up to 256 users dialing into our Point of Presence systems or thousands of users logged into the service. I was both a hardware engineer and programmer and I really liked your enthusiasm about this technology.
2770 pins? Laughs in Treadripper with it's 4094 pins. :D Jokes aside, that thing is insane! 9 Gigs of RAM in 1990 is just incredible. That O-Ring noise sceams for RELEASE THE SCHMOO!
@@rickfeith6372 It's the more correct term for 2^40 bytes. 1 TB is 10^12 bytes (1000 GB or about 931 GiB which is what Windows/Linux shows to the user).
@@EEVblog The mineral oil dissolves solder joints apparently - there is a video of Linus Tech Tips from 2-3 years ago where they revisited a Mineral Oil build and the SMCs were literally falling off from the MoBo and other parts of the system.
At 19:33 you wrote Tin/copper, the image says 97Pb/3Sn -- that's 97% Lead, 3% Tin. :P Great video though. PD: I believe it's usually pronounced mo-lib-d'num, or something along those lines. I'm not english native, so excuse the crude pronounciation guide. :P
Yeah, back in the day, we used to "force" technology. Our company used "hybrids" to attach power devices (dice) for audio output directly to silver traced (like a PCB) ceramic substrates. Then we wire bonded them and encapsulated them in a special epoxy. This was in the 1980's. We used Beryllium oxide substrates for their thermal conductivity. They actually conducted heat better than aluminum. Today, I could design the same thing using off-the-shelf parts and it would be better. Of course these (in this video) parts were soldered down, but the oil bath improved thermal conductivity. What a mess! Our company actually used peanut oil in an aluminum tray (aluminium to you Australians lol) in a resistor heated bath to regulate the solder bath carefully. Odd to think that worked, but it did very well! I'm so glad those days are over!
In my first job in back in '94, there was one guy in our shop who worked out of the office. I didn't really understand what he did, but as the rest of us were field service warriors, staying in one place looked pretty nice. One day I heard him quip, "well, I just earned my salary for the year!" Curiosity piqued, I asked what he meant. Turned out that the customer he'd been working with had been informed (by Big Blue no less) that they were going to have to replace a processor module in one of their systems, but this guy (along with others, no doubt) had figured out what the real problem was, and got the customer back in the air. No processor module needed. When he told me how much these things cost, I understood! Really great vid, by the way!! Possibly Dave's most epic geek-out to date!
Although its "mips" rating is quite low in comparison to chips today (and even back then) its seriously not the whole story... like the S36, S38, AS/400, RS/6000 etc. its all about the sum of its parts. With disk controllers (additional processors) that could sync the rotation of multiple disks (to stripe the data so that multiple spindles returned the data at the same time) and memory controllers (again additional processors) that could be asked for data and then left to do the job without main processor intervention (and interrupting the main processor when done) and with data buses that were "uncoupled" (again separate controllers; picture a 60 lane wide Ethernet for a rough and ready visualisation) the "processor" was not much more than a mega calculator and also had an instruction set that was really specific to transactional processing (think lines on an invoice, stock levels, bank transaction) and also included calculation instructions that were "fixed point & length" processing with half adjust and truncate/overflow for calculations specific to currency work and item levels. The processors were brute force machines, data in - calculate - data out, that left the rest of the work down to other processors/controllers. This meant that the spec-perf/transaction processing scores were way out in front of other processors of the day, even ones that were a lot faster... mind you, you paid for this in real money! You could also "lpar" the machines into virtual machines (even by fractional "core" metrics) many years before PC's and the machines also had "service" processors that would watch over the whole box, and if the machine failed to boot would give you the exact reasons why (disk 14 failed; co-processor under voltage; database not in consistent state on shutdown; jon smith smoking in the computer room; hello dave). I know I sound like a fan boy who has gone all stalkerish over computers, but these machines were light years ahead of everyone... even now PC's are only just (well a few years back) starting to get some of the IBM stuff from way back when (like error codes not beeps; service processors (ipmi); AMD chiplet design with uncoupled memory controller; etc.).
I agree, it's not about pure processing power... But still, for a mainframe/server system, only 20 MIPS (for the most expensive model), that's comparable to an Intel 486DX at 33MHz, which was also around in 1990... Maybe the 20 MIPS rating is meant to be per core, instead of the whole CPU assembly/package ???
@@HPPalmtopTube in 1990, we used to run 700-1500 'office' users on what wasn't the largest IBM mainframe by far, but no way you could do that on a 486 based machine. Its all about independent handling of i/o. You'd notice it wasn't actually that fast when asking cpu intensive tasks, but it remained amazingly responsive with for the time very large numbers of users. This kind of system is not for calculating primes (or such) but to facilitate hundreds or thousands of users concurrently, all of them doing relatively low intensity tasks but needing good responsivity. Say the hundreds or thousands of office workers at a bank, air traffic control etc. Its not so much 'per core' processing power, but taking lots and lots of work that doesn't have to be done by the cpu completely away from the cpu.
@@vprincessfan12 I still think its was one of the best computers ever invented. From the "object based" OS to the ease of programming using RPG (obviously both had their roots in earlier incarnations). Such a shame that IBM didn't keep promoting it after its various name changes. Its still there, still being updated, but few outside of the current users know it exists.
That is a Freedom Hercules TCM. It was actually a backup package for the System 390 because IBM was having technology issues with the glass ceramic substrates that were originally designed to go into the System 390. They did solve the issues with the glass ceramic and used these substrates also. The vias in the glass ceramic modules were faster due to copper being used.
"Let me just try a pair of pliers to get this speck of paint off... oh I just tore right through the canvas, there you go, 10 cm gash in the Mona Lisa... sorry Mr Davinci..."
@@stefan0ro this guy is fucking hilarious.. You'll love him if you like rossman. Both extremely skilled in their trade.. Eevblog is more diy friendly though 🙂
@@codycast not all videos are great, but it's a really good channel. Don't get discouraged if you hit one or two somewhat boring.. see for exampel ruclips.net/video/6dwqxsDHkKQ/видео.html or a bit older ruclips.net/video/w3c2dOw6Frg/видео.html
@@CarlJohnson-wg6wy thanks, I have many Pi's...can I make like a 32way sysplex? Hahaha. I think I may try that on a pi. Get my mind working in a happy zone. Does Hercules include a HLASM? Just dying to issue EQUs, LTORGs, DSECTs, AIFs/AGOs
When I worked at CAE in the 1990s they had an IBM 4361 System 370 that was the little brother of the System 390 that took up an entire room. We bought an IBM P/370 that was a "Personal 370" which was an OS/2 server with a plugin card that contained a S370 card to replace the 4361. In 1996/1997 was upgraded to a P/390 that was a "Personal 390" with the same setup, a plugin card in an IBM OS/2 server. IBM ran all the peripherals, disk drives, tapes, I/O, etc. virtually in OS/2. To the board it was a full System 390 that was running, we even had a 9 track reel to reel tape drive connected on a SCSI card. It was very impressive technology, this was supposed to be used for development where you could build and run your system with being on the production main frame. So the System 390 was the computer class and you bought the capabilities you needed. If I remember the software prices were based on the number of CPUs you had. With one CPU, as we had in the P/390, FORTRAN was $10,000 to $20,000, a bargain compared to the $50,000 we paid for an AN/AYK-14 assembler program. What you have there is a beast, all those CPUs together on one massive module would have cost a fortune. Mainframes were all about the I/O and parallel computing more than raw MIPS on a single thread.
@@EEVblog2I think it is used here because for the bga balls cause its more forgiving to thermal expansion/contraction (i.e. softer) www.researchgate.net/publication/267584624_High-Lead_Solder_Ball_Characteristics_in_BGA_Package (Doubt the higher melting point is relevant) Tin makes the solder stronger but stiffer, they must have wanted something reaallly compliant, it seems unusually high in lead even considering. No wonder it's so easy to remove, it's basically pure lead, which is ten times weaker than tin Edit: whiteboard chalk... I need to sleep XD Double edit: hardness =/= strength
Interesting ! I've actually worked in the french IBM assembly line back in '95 and with that kind of processor. There where some air cooled, other with freon. First time I've see one dismantled. If a chip (or something else) was not working, the whole CPU assembly won a free ticket to USA. Thanks for the back-in-time travel !!
System is just the code name for their instruction set architecture. Series is their keyword for architecture series. This was usually correlated to the OS e.g. Z-Series with Z/OS. ES was shorthand for Enterprise Systems. So System and ES are interchangeable.
Well not really so far as an Arduino Nano, but I Raspberry Pi I sure imagine beating it. It's also interesting how they work, as far as I know those mainframe CPUs (and mainframes in general) are highly optimized for absurd parallelism of operations, they're closer to graphics cards in how they operate (sort of... maybe quite far still tho... Mainframes are weird).
@@Kalvinjj It might beat it in MIPS, no way your Raspberry Pi would beat it for it's IO ability. That's what Mainframes were/still are excellent at, handling thousands of user terminals, ATM's etc all with extremely quick response times.
@@Mark.Brindle Indeed, for IO ability it still probably takes another mainframe CPU to beat it. Maybe with PCIe interfacing, but then again what kind of pre-processing would be required to mediate the data I have no idea.
@@Mark.Brindle Back then IO was not sending TB of data per second. I'd imagine a normal CPU could handle all the IO workload with ease, not because it has a huge bandwidth, but because it can serve each request in a fraction of the time. And probably still has a much higher parallel bandwidth anyway. (I have no idea what I'm talking about and would like to know).
I started my professional career as a cobol programmer and then as a systems programmer on IBM mainframes back in 1993. We swapped out our 3090-300J for an ES/9000 9121 so really appreciate the teardown to get a look at where those electrons flowed...
Dave, this is the first of your videos that I've ever watched. I am glad that this was the one, cause thanks to your, shall we call it "enthusiasm", I am hooked. It seems like a portion of you hasn't ever grown up, and I hope it never does. Im looking forward to watching all your past stuff, & future stuff too. Have a fantastic day!
I worked on 2 different 9370 systems back when they first came out - I believe they were the little brothers to the system this came out of. Very interesting to see this. Needless to say, this was one of those systems that sat in an airconditioned room 24/7/365.
I’ve got a couple of these at home, they came out of one of several 3090 mainframes at an Australian bank. I worked for a 3rd party maintenance company. We didn’t buy spare units from IBM, we just bought old mainframes and stripped the parts from them. The 3090 was water cooled and the TCMs had a water jacket bolted on top, to replace a TCM you unbolted the water jacket then undid the retaining screws, torquing up the screws on the new unit. The main thing was to make sure that you’d powered down the correct frame (you only powered off the processor that you were working on, the rest stayed up) I heard of an engineer who pulled a live TCM, there was a big flash and bang and the machine was offline for several weeks while a new backplane was installed. Those suckers drew a whole lot of current.
Comparing MIPS on a mainframe with a 486 isn’t quite so easy. Unless someone knows the complete facilities involved it just isn’t an apples to apples comparison. Dave makes a valid point in bringing the IO into the picture. As for the porn comment I think you’ve hit a new level of bliss in a tear down.
I am probably the youngest viewer here I’m only 15. I have a love of electronics and computers and I make teardown videos taking thing apart and explaining how they work.
IBM always had amazing mechanical engineering (check out the mechanical masterpiece that was the Selectric sometime, or their high speed tape drives). It makes sense that they'd see higher performance as being mostly a thermal packaging problem, especially back in 1991.
www.popularmechanics.com/technology/design/a28816626/worlds-largest-computer-chip/ "This Is the World's Largest Computer Chip It's the size of an iPad. And it's ready to unlock the power of AI"
@Right Round They would be sanded and polished by a machine, but not lapped. Lapping is a very specific process, and the term is usually incorrectly used by computer enthusiasts.
@Lassi Kinnunen Nah, not "entertainment stuff like racing engines", but Military; that's where the big bucks are for specialist engineering in the USA these days.
I remember some more too about IBM TCM's. On the 3745 they were smaller, with just one square block of pins, I don't recall how many. We'd receive a m/c from the assembly line ready for power (These were mostly "Y" configured 415v/ac 63a m/c's) and logic testing. After power testing we'd populate the m/c with all the logic, including the TCM's. We'd have to lay a static mat to stand on, while wearing our ESD straps, which were all tested before each logic population. We'd unbolt the fans and unscrew the heatsink. Populate the logic cards, then came the TCM's. They came in a 400mmx400mmx400mm shock and ESD proof box. IIRC they each cost GBP5000, back then that was more than my annual salary. We had a special gauge that was used for checking for bent pins. It was a 3 or 4mm, machined, sheet of plastic that "floated" over the TCM until it fell in to place. Each pin had a matched hole and it was easy to see if there were bent pins (There never were and you'd never cause one or lose your job). So we'd unpack the TCM, unwarp it form the ESD bags, check for bent pins and gently place the TCM on the "socket". This was a huge thing, an over grown 486 like thing. We would then insert a wrench in to the TCM and socket, the wrench having a cam like end to it. We'd rotate the wrench and the TCM would be "wound" in to the correct seating position on the socket and locked in to place. We'd attach the heatsink and fan and power on for the logic test. The "BIOS" would go out and "discover" what hardware was in the m/c and then we'd run diagnostic and soak tests. I did have one 3745 prime power box blow up when earth went open circuit! While debugging one m/c one night, a sprung loaded component struck me in one eye. The damage is still there, optometrists always ask about it, when I go get new glasses. Fun days! Thanks for the memory.
Could you imagine doing the routing for those 63 layers? I know it would be a whole team of people doing it. But would still be a lot of traces and via for each member to be responsible for
Gets even more insane when you think about that the module was designed to keep working at its nominal capacity even after a certain number of random(!) chips failed.
@@HenryLoenwind you read the reply from John Bellas? interesting how they used a paste to form the vertical interconnects layer by layer, alignment woes, amazing to think of these guys forging ahead into the unknown to make something so specialized, like the women who wove Apollo memory modules.
i'm a geo by trade, and before COVID i spent over 10 years working in a greenstone gold deposit. as soon as i saw that moly gleam on the contacts my heart was aflutter. it's an indicator mineral in greenstone gold deposits. i was very happy to read the comments by the former IBM people in the comments. thanks all!
Wow! This brings back memories. I worked at the IBM East Fishkill plant back in the 1980's. I worked in the manufacturing technology areas where the ceramic modules were created. At its peak, the plant employed about 15,000 people, and the weekly production could be driven away in the back of two station wagons. An unbelievable amount of engineering went into the design of these modules and the manufacturing processes to make them. The ceramic green sheet material that formed each layer of the module was created by a long sheet roll process on a machine that was approximately 63 feet long. The roll was then taken to a die machine that precisely cut the dimensions of each sheet. The sheets were precisely punched for layer-to-layer vias, then fed into screening machines that spread molybdenum paste through masks to create the pattern on each layer. The next step was a stacker that aligned and stacked each of the 63 layers so that vias from layer-to-layer would align. The stack was then pressed and readied to go through the sintering process. The sintering or firing, of the ceramic substrates was done in huge sintering furnaces that had a hydrogen forming gas. Each furnace was loaded by robotic arms that moved assemblies that weighed hundreds of pounds. When the process was running, the excess hydrogen gas was burned off in a stack. It looked like something from the Wizard of Oz. Occasionally, there would be an air leak into one of these furnaces where the oxygen/hydrogen ratio would be just right to cause an explosion of the entire furnace, shutting it down for an internal rebuild. The ceramic module would then go to the thin films processing section, where it was CMP polished for flatness and the multiple thin film layers were deposited using technology adapted from wafer manufacturing. In the newest generation of TCM, there were six layers of thin film redistribution wiring on top of the ceramic, before the chips were attached. This process had to provide for corrections in lower layer yield defects, and had to achieve 100% yield over the entire substrate. There were multiple customization laser repair processes that added/subtracted single trace defects between the chip sites. Wow! I could go on, but I'm running out of steam... maybe someone else can pick up and describe the additional complexity in the manufacturing processes of this unique assembly.
Wow! Thanks for sharing this btw. I also saw a comment on this video of someone who worked at the same IBM plant mabey you could add this to that thread too
Kinda funny/fascinating reading your post when you talk about "Each furnace was loaded by robotic arms ..." and in another comment to this video is a guy who developed those robotic arms and talks about that. And all that was 30 years ago.
wow ! thanks man !
Very interesting Edward, thanks for sharing mate!
Hey Edward, I'm actually a mechanical engineering student in the Dutchess County area and I've been looking into IBM in Poughkeepsie. What you explained is probably one of the most detailed and eye opening accounts of what happens during the manufacturing process, and has definitely increased my interest about this industry even more, thank you for that.
How often do you change your processor oil? Once every 2 years, or 50 billion quadrillion instructions, whichever comes first.
Haha good one
😂
This processor never reach the 50 billion quadrillion instructions, at least on it's first 15 million years!
¡No sea mamón Stuart!
@@pedrocasique7352 ¡No sea mamón Pedro!
About the processing power: The 390 is not a scientific computer but a business machine. It was not designed to run calculations but to run data through. "Want to run a batch job over all 50 million accounts of you bank's customers? Sure, will be done in 5 minutes."-type stuff. Or to connect a couple thousand end user terminals to one system and have guaranteed response times below 1/10 second.
The missing chips are due to increased load, as far as I know. Those modules were highly redundant, so the advertised capacity probably was only half of what was possible with the extra chips as backups and to compensate for production errors. I wouldn't be surprised if a third of those chips didn't work in early production. But there were enough of them for the complete module to still be functional and have enough backups for plenty of them to die later. Later when the yield got better they could leave out some chips and get the same result.
Cool info!! Ever since I saw Jurassic Park in theaters I've been fascinated by mainframes!
As it was replaced with today System Z with dual everything (all operations are done on tow separate machines and compared then)
It newer was intended to be fast system but extremely reliable
@@masaharumorimoto4761 "It's a Unix system!"
@@DavePoo What is? the S/390 ran OS/390 (previously called MVS) or VM, wasn't unix or even a flavour of. Mind you you could run nearly everything under VM,
@@MrKnowwun ... was referring to the computer in Jurassic Park, not to the IBM.
I thought about taking it apart for years. So glad I did not. When I discovered Dave I immediately hoped he would enjoy it for mailbag. I learned so much more from watching Dave disassemble this that I would have doing it myself.
Your generosity has brought joy to multitudes of people, Jim. Thanks so much!
So, you're the fellow who sent him this? I think that you can count on Dave as being your friend for life now. And anytime you two visit, the pints are on him! Thank you for giving up what must be a prized possession for Dave to work on. Truly beautiful and amazing technology and a historical dive back in time.
Thank you so much for sharing this with him and with all of us!
You're a hero for doing that! My dumb ass...
Thank you!
I don't know why youtube recommended this video to me , but watching this guy being so happy over a processor is making me happy!
Wow yea the memories are flying now!! I worked at I.B.M. East Fishkill N.Y. for 31 years and for about 2 years during the late 70s to about 1981 I worked in one the many lines that made these TCMs when the TCM had 33 layers!! now about the Molybdenum , The MOLY was deposited by an electroless plating process that used 3 or 4 steps on the pads that supported the pins, the final process
plated the pin pads Gold. The internal wiring was made of Moly paste that was "screened onto" each individual ceramic layer to "wire" the chips together!! The real trick was to keep all the via holes aligned for each layer to have continuity with the layer above and below itself when the TCM was being "fired" to cure the ceramic layer's that formed the TCM the via hole would misalign and cause loss of continuity in areas of the TCM due to inconsistency of the shrinkage rate of the layers of ceramic!!!!! This was a HUGE PROBLEM for a while but the engineering departments figured it all out and solved this daunting problem and done the fix!! In later years the layers went from 33 to 64 and the via problems remained solved due to changes in the ceramic mix!! And yes I seem to remember the name Knickerbocker as one of the top engineers in the MLC (Multi Layer Ceramic) program but never actually met the fellow! John A Bellas KC2UVN
Very interesting.
Thank you for sharing.
Does the word Demohag ring a bell?
Pottery and electronics combined that's awesome
It's a great mainframe cpu disassembled, now our mobile phones have more powerful CPUs, in our pocket.
I was going to make a similar comment. I worked for IBM in the 80's/90's too on 8100's and 3745's, in the factory as a test technician. I also worked on some CNC milling machines to +/-2 microns. The 3745's had TCMs, though were air cooled. Sys390 9121 (ES9000) was my third mainframe with 256MB of real RAM! We went from 24bit to 31bit addressing. WOW! 16MB to 2GB address spaces for MVS/ESA 4.3.3 (Which initially didn't run natively, we have to run it under VM/ESA). ESCON channels too. It was one of the very first ES9000's in the UK.
They are called TCM. This brings memories :). I use to know so much about them from time I have worked in Dismantling IBM facility in the Germany. My job was to take special parts from preowned machines and package them for the clients that could not afford new or refurbished parts. TCM's where most expressive parts ever. Some of them costing in millions of DM (Deutsch Mark, German money before Euro for the new kids :P ) . I have build those out of the main frames. There is whole procedure for taking them out. You need special key that unlocks the cooler block. There is a special pen like tool if pins bend. There is a special conductive plastic case with special foam inside for caring them. There was a huge call it a main board with 9 sockets. and it was massive. 3x3. Some frames have had two of those one on the top one on the bottom. There was a special tool that you screw in those to use it as a handle when taking those big boards out. IT was two man job tro carry it out of the frame. I was young and stupid and was doing ti alone until my back gave up. board with TCM's in it was 90KG. It cost me months to fix my back and it involved putting electrodes and some current true my back. First you take a hex bit and unscrew all water blocks from all of the TCM's. Then you would Pull out whole tubing and cooling blocks. then you would unscrew huge power rails from the so called main board. Then you crew in handles and unscrew the mounting screws. You take the thing out and place it on the table. Needles to say you are grounded with a wrist bracelet and you wear special anti-static shoes and on grounded table and special painted floor that can not generate static. Teh you unscrew TCM and then you use special key tool to unlock the TCM from it's socket. Turning the key was moving the TCM sideways unlocking the pins. Then you inspect it and make sure pins are all good and correct them if needed. You stick your bar code label on the side so people can trace your work.Then you place TCM in the ESD bag and then you place it in the special container. Then you lock it and seal it with special wire and lead seal with special tool. To open the container you have to cut the seal or wire off. I remember looking at part list and pricing. And some of those costed like 4,5 Million DM. I think Cheapest where like 1 million DM.
Some of the TCM's where processor units having in them 11x11 CPU cores. Each core have had a little cylinder with a spring making a connection to a silicone core for direct cooling and the top block would be like heat spreader. Water block would crew on it. Imagine those pig pumps. Those have had huge electric motors few KW each. Tubing was a maze. Individual tubes where like size of a little finger and they would all collect to bigger hoses.
Some TCM's where memory only and would have up to you guested it 11x11 memory chips.
So every TCM have had it's function.
Whole system was made of 7 main frame units/closets. One of them would have two gigantic pumps and special reservoir that decriminalizes liquid used in the cooling system. BTW it was cooled with some alcohol mixture.
Now imagine 6 out of 7 frames would be all full stock with 18 TCM's and one frame would be Cooling.
Yes no permanent storage!!!! Whole system needed storage solution that was again contained in more frames/huge closets.
Imagine costs when Bank or some Government institution orders a system. And yes many have ordered multiply systems in the cluster working as one back then gigantic system.
And yes client side units where thin clients connected with Token ring adapters or later BNC 10baseT networks and having black and green CRT monitors up to big 14" :).
They have had big mechanical keyboards with heavy metal base.
I was taking them out of the 3900 series.
There where some systems for the storage and they have had floppy disks drive. 8" fasynchronous
This particular TCM refurbished from IBM cost $225,000!
EEVblog you mean today?
Jasper Janssen back in the time the use to cost between 1 and 4,5 million Deutsch Marks. Even without inflation figured out that is 500K to 2,25 million bucks. One DM was 50 Euro cents and 1 Euro was same as 1 US $. IBM used to ask original prices for computer parts even 5 years after while those would cost nothing in the other shops. For example in time of first Pentium CPU main board from model 80 was still few thousand DM but it was 386 equipped. No one would even take gratis 386 CPU at that time 😀
NINEWALKING yeah, I’m a little surprised that even for IBM mainframes you can still buy refurb parts 30 years later. Although if there’s *any* company and *any* product, that would obviously be it.
Jasper Janssen those systems where build to work for 100 years and engendered to be repaired. They where so expensive that once you have them you need to use them as long as possible. Remember there where no server clusters approaching this power and almost no one was making any other solution. IBM have had practical monopoly on the main frame systems. Only later on Fujitsu Siemens started with som products but that was already competing wit ES9000 systems.
You need to take it back to IBM and tell them it was like that when you bought it.
That's hilarious! IBM actually had an exchange program where you could bring in a defective TCM and they would replace it. The rate depended on the type but was mostly $22,500/ea. They would have suspected fowl-play when we brought it back dripping with 10W-40 from our failed oil change.
I Used to be a mainframe programmer. Really enjoyed seeing the brain that executed my code. Thanks!
wow , we came a long way just in 25 years
However it seems we've moved forward then backwards. An Intel Xeon 10 core running WS2016 seems to be such a dog!
We?
Try 30
IDK man, now its tons and tons of spaghet running on the modern procs! We got dumber and Moore's law is over now. HW like this is so nice, and the software was typically as intelligent. Now we run everything on JS where 2+2=5... 🤦♂️
@@AlexKosinski what are you even talking about
And right now, i bet ASRock is trying to come up with a microATX board for it
Did you check Ali Express?
SuperMicro probably has a Pico ITX board for it
I think the processor might be the size of a microATX board.
I checked. They have one that supports this platform as well as i7 and Threadripper AMD. Can also run a Pentium II off a riser board for some reason. $69.50 USD
@starshipeleven That's fantastic. Thanks for checking!
I never worked on mainframes that used these back in my IBM New Zealand days. But they were so expensive that the IBM Maintenance Parts Centre had an aircraft on standby to fly replacement parts across the country, to maintain service level contracts. It was much more cost effective to have a small central stock and a plane to freight the parts around the country.
Stop the presses! Are those bodge wires left of screen at 17:39!?
Also if you can’t get an X-ray or µCT of it a physical cross section might be easier to organize.
Heh, just commented this on the forum and was about to comment here...
Yeah it's insane! Also a lot more at 18:52-18:56
You are right, wow, totally missed that. Those would be crazy small!
Wow, good eye! I totally missed those.
Looks like an ultrasonic weld; at first it seemed like they were configuration or to bridge unpopulated chips but no, there doesn't seem to be any pattern
Genuine ibm certified bodge wires
Ken Shirriff - Let's take a chip die apart layer by layer with chemicals, then do a detailed circuit analysis of the photolithography and metal deposition layers.
Dave Jones - Let's scratch the whole thing up with wood fibre paper towels, and rubber gloves, and then smash the dies with a huge pair of pliers...
You know what they say, there are two kinds of people in this world....
5 microns flatness is only ~10 wavelengths of visible light. That's meh for an optical flat made out of glass, but absolutely incredible for what is effectively a multilayer ceramic circuit board on steroids.
One reason for the oil is mechanical damping, otherwise during transport the pistons beat the crap out of the chips
Marc Berte I hadn’t watched the whole video yet when I read this and I thought it was a joke. “Surely it doesn’t have pistons, he’s just joking right?”
I think the oil is a thermal interface here due to the imperfect contact of the metals, similar or same oil is used in immersion cooling for servers
To take one from a v e's Channel.
Release the shmoo!
Remember back when they had all that goop inside of devices not necessarily what I'm called planning compound but I think that's what it was but it was more like just a jelly Flash paste felt almost like I don't know man he's or something let me a good way to say it but I guess it was some form of rain for Herbal potting compound I don't know but it was often used in things like some power supply bricks so I assume in my defense when do if thermal properties are pooling and or just to keep people from messing with devices because it was actually somewhat liquid and cover much or like and not necessarily A Johnson but more liquid than that I don't know what would be the best way to describe at the anything I can think of at this time at first because I saw it in some battery powered equipment I slept that the batteries and exploded but they were in fact and then I remembered later singing and other equipment that was the first couple times I saw it though so yeah surprised me back then Maine this was like 80's and 90's naming somewhere in there.
I do remember one of those things was like it sort of looks like a dustbuster is kind of deal but was rhetorical flashlight possibly from Radio Shack? Reminded me of of a dustbuster except a bit flattered and with a square lens on the front.
And when I realized what was I did wind up for parent the batteries were shot wound up getting it those replaces cell rechargeable batteries from RadioShack for the purpose actually I was able to put in larger cells then it had originally either the night size up or the next I think it was Sub C and went up to either a c-cell or a D cell battery cannot remember long time ago but Walmart 1st hacks to increase battery life in something well at least Mike that bit whenever built-in rechargeable battery of that click nicad not nickel metal hydride back then that was all that was available in that type of battery as far as I would know from personal experience.
Other than a Bee Line acid but not something like you'd use in something like that.
People assaulted rechargeable tools were new who sings Even free at prior to that however there were actually something that you would call cordless tools except they were not truly cordless but you did not require Hunter and 10 volt power to run them just recharge them but they did have a cord going to a battery pack it was a rat acid battery pack I think it was more like on a belt I'm not talkin about the system for Black & Decker either that you was one of those referral batteries with the cigarette lighter receptacle which by the way is that battery pack was actually designed for another product at one time and I were some of those like in at leatherette case as well back in the old good old days.
Also rechargeable radios and stuff go back further than you think as well.
But for the virtual cordless tool system I do believe is that the battery was maybe on a belt pack of some sort but for the original cordless tools system I do believe that the battery was maybe on a belt pack of some sort I don't remember how many volts are anything laws but someone I know used to have this system years ago and they told me about it you could say he was one of the greybeards in whatever he worked on.
I used to have a old to tube AM radio that got excellent reception I actually brought the power supply for it.
I literally breadboard this thing by the way the word breadboard actually comes from using a cutting board as a building base to prototype something that's what was first used to do something like that when something was designed can't remember the entire story but that's for the phrase breadboard comes from.
If I remember correctly I use standoffs I think I made from sorghum sort of TV antenna I think it was an old rooftop antenna I want up cutting sections off the antenna elements to act as my standoffs is there were even wood screws as well as machine screws holding it all together and yes fasten hook Clips as well for connections need both antenna and ground and later on I built the symbol amplifier and I use the old PA speaker as well I used a pair of antique headphones as well.
I cautioned my folks if they were using it not to touch the amplifier when it was on it did not have a chassis it was just mounted on a board everything was exposed and yes I'm a line voltage as well it later on I made a chassis out scrap pegboard material when we put up Peg boards..
later on I use that stuff for terminal strips as well for some projects It worked out great for building things that need to be connected to other things as well. For that purpose.
I also want spilt the radio for someone that fit into a junction box cover don't remember the size of it.
Also wants built a landline phone into the junction box cover as well using Salvage phone parts.
one of my friends gave me a power strip that his grandfather made.
It had the old Wire like that would be used for two men knob a light socket as a fuse holder twist time for rotary switch not the push button switch of the time these were the little round ones that were surface mounted.
There were some receptacles in addition to Edison screw sockets those were in parallel the idea was it could be either used four things with an actual plug or something that had to have a screw show plug on it hey Carrie r u v a light socket.
I actually wound up putting a damper in one room for a night light because it was hard to see the switch at night in that room the other one had one of those screw base adapters to take a standard plug later on I ran acrossed neon indicator bulb with a standard screw base on there and use that for the night light.
also I had a couple from other people that use the round Bakelite type receptacles like you would use in barn or other wiring with surface fan wiring using the nonmetallic sheathed cable.
I had both grounded and ungrounded versions of the once which even had a switch as well.
obviously that was only used for certain things like where you hand a plug-in say stereo components or something where you wouldn't pull much power because a chord on some of these was only about 18 gauge possibly 16 gauge In one or more cases.
what are friends had one that his grandparents had used probably their parents hard to say that have been built that took d screw base Plug & went to more than one light socket so essentially a power strip before that was a thing it was mounted on a piece of wood that been salvaged from something else and also there was switches on there as well.
also has anyone ever seen a hard-wired service about night light evidently these we're sometimes not on stairways some of these were even had internal switch although they weren't available also with plug pins but they were the exact same socket assembly they accept screws instead of the for the assembly attached.
This was like the brown Bakelite?
also what's the deal with those old round light sockets is mounted on usually a black metal plate with a ceramic socket often times has a screw-on part on top of the socket I just see those quite often in places sometimes it's just one of them in the entire place I don't know if it's coincidence or not.
A lot of them are others are just ceramic the standard sockets either pull chain or not.
although sometimes I've seen those like in a garage or something but I stared ones are in a house and sometimes both in the same area.
also there is one of these at my parents place in the garage however there is no evidence of a switch ever being there and the wiring just seems to go nowhere there is a few holes where it might have gone but no evidence of it ever being there old metallic flexible conduit of some sort.
possibly no evidence of an even leaving the garage which is just weird.
Although there's one on thing in the house downstairs there is one circuit has ran in Black conduit which is only one circuit like that. It's got the old School round box again black.
and this goes through the side of the house but then heads up through a conduit body and up the side of the house to the upstairs We still don't know what it feeds but it obviously if it was possibly done later? Is everything else is ran inside Possibly from the 30s or 40s?
this dude feed into in the junction boxes in the basement that was used for the lighting I suspect that it's possible that may have fed the garage at one time but had but had been redone but had been redone and repurposed for the upstairs?
also that conduit down in the basement tests live with a no-contact tester so little skeptical about grounding?
things have been checked over when the wiring was doing but this has me scratching my head what was the purpose of that conduit why would just one circuit that way.
also most of the wiring in my house is the old BlackNonmetallic sheathed cable I'm not sure which brand But it's all the same sort of stuff I don't know if this is rubber covered or not but could be I'm just not familiar with itBefore my time no evidence of knob and tube whatsoever.
however original telephone wiring was to however original telephone wiring however original telephone wiring wasI'm single pair Twisted wire Black in color Supported with little Tak like wire holders possibly wood or plastic or similar material + 2 Connector elongated terminal blocks 4 Connection pointsAnd two or three places in the basement Along the Floor joist. And one Where the wiring came into the houseOh and one more thing there was some iron wire And you're at one time that was insulated Not aluminum but iron yeah evidently that was a thing.
@@JohnRunyon 1991 steampunk processor
Doubt that.....Thing was made to be in 1 place....Not to move around.....
I worked at IBM in Poughkeepsie in 2000 and they were still making these large server modules. This is when they were transitioning to all copper interconnect. The ones I remember were liquid cooled and a refrigerator sized machine had rows of them. The AC unit that cooled it was larger than the computer! They had these huge shake tables that would test these systems for sensitivity to vibrations in another building.
I worked at IBM in Kingston, NY in 1986. I tested, troubleshot and repaired memory boards for the IBM 3090.
Some IBM engineers where just crying when you started poking at the caps
not jut engineers...
He's thrilled by his ability to destroy something he doesn't understand and could never have created.
@@pglaskowsky Nah...
@@pglaskowsky It's almost as if exploration comes with risks and one can express joy at the learning even if it comes at a cost.
@David We're in agreement, I was assuming their argument and showing that it still doesn't matter.
MIPS, in case anyone is wondering , stands for "Meaningless Indication of Processor Speed" :-)
Probably the happiest I’ve seen Dave in 5 years of watching him.
From Roy Longbottom's benchmark web site:
1991 IBM ES/9000-900 with 6 CPUS at 111 Mhz: MIPS = 248, MFLOPS = 2,664, cost = $22,600,000
Raspberry PI 3, 64 bit, 4 cores, 1200 Mhz, ARM v8a: MIPS about 6,500, MFLOPS about 480, cost about $30
Yep, heck of a lot faster per $ these days! But as Dave said, even back in the day mainframes like this were beaten for MIPs by desktop CPUs. So why did mainframes sell, when a few PCs could beat them for speed, even in 1990?
Try running an airline reservation system on that RPi 3... with 99.99% availability, mind you!
thing is the IBM machine has much faster I/O
@@GooligumElectronics It was not just running fast. These were real multi user systems. And they had all the fancy periphery like "huge" mass storage drives. At this time a "PC" was DOS5 or 6, maybe Win3 or some early UNIX, LINUX. These things were just the tip of a whole eco system not just one fast running CPU.
@@GooligumElectronics Precisely. Those things were used to run a uni campus, stock exchange, or huge bank, stuff like that, not some shit Windows PC. Therefore comparing plain numbers, like those PC nerds like to do, is totally pointless.
@@AmauryJacquot That's the whole reason the mainframe still keeps going. Because at 30,000+ I/Os per second you can get transactions rates over 100,000 txns/sec and that's what's needed to run a bank, an insurance company or a supermarket.
The piece that gets that running is the fibre-optic channels for the connections between the processor and the disk units. When Dave's 9121 was current that was about the time that we moved from a 4.5Mb channel to high speed fibre. Next move was from big fat disk drives (like the one Dave tore down in ruclips.net/video/CBjoWMA5d84/видео.html) to RAID arrays using inexpensive 3.5" drives. The I/Os to the mainframe all became emulated.
It's all moved on in size in the thirty years since then. The machine room was the size of a football field in 1990. (when I started in 1981 it was the size of two football fields.) We can get so much more than that in an IBM z15 mainframe that will fit in the cupboard under my stairs now. There's also less heat and less power usage than back in the day when the mainframes were all water cooled.
When I was at University in the late 1980s "we" had an IBM 390. A friend doing a PhD in Chemistry got hold of a PC with a 286 processor. Running simulations on the 390 required him to book processor time a few weeks in advance. He could do the same thing on the 286 but he had to let it run for 24 hours or more. Still had it done faster than booking time on the mainframe.
I love the enthusiasm
A thing of beauty is a joy forever
Unless its a mirror
until ya smash it up
Thing of beauty? Yes. Joy forever? Obviously not when Dave get's his Hands on it ;)
But not destroyed :(
"MADE IN USA" a rare sight these days
Yea i often see
MADE IN CHINA
MADE IN KOREA
and more but USA is rare
go and vote Trump then
@@_programming_ no u
@@_programming_ vote for the man that has all.his products made outside of America. Lmao okay
@@patrickrosington1174 which product man. Trump is a builder. Real estate. WTH you are talking about.
And suddenly, processor porn turned into processor gore.
Yeah
I watched my dreams become nothing but metal mass and shards
Yeah, that escalated quickly.
Snuff movie
As for why the pads dont come off, the solder is very soft as its 97% lead. We used molybdenum alloy metallization on ceramic components at my last job, and properly done it could easily exceed 30ksi of tensile strength before it delaminates. I did mechanical testing on small 3/4" ceramic rings brazed to kovar and they could take about 1/3 of a ton before they separated if proper braze material was used. Even more if strain relief washers were used in the joints. Since they have a fair amount of metalization anchoring the pads down they are considerably stronger than the lead solder. Looking at the cross-sections I see all the usual suspects for bonding to ceramic substrates.
Documentation states it was 97% tin, 3% copper. No lead at all.
@@caodesignworks2407 19:22 states "97Pb/3Sn", or 97% lead, 3% tin solder. Tin is much harder and more brittle.
@@caodesignworks2407 Dave just mistyped the composition as Tin/Copper, which is incorrect according to the snippet he posted.
I worked in a warehouse that reconfigured 3090s. Interesting thing, they ran on 400hz power. We had motor generators. They were also water cooled.
Never realised the cost 😱
Interesting! Aircraft use 400Hz power systems. I wonder what the applications were for the equipment you worked on.
@@moconnell663 Aircraft spec 400Hz for the sake of smaller transformers/coils/caps to save weight. Lower frequency has less loss over long distances, but that's not a big problem on an airplane like it is on a whole power grid.
I'm surprised RUclips didn't stick a "Mature content" warning on this video.
electronic snuff film!
iLOLd
@amycoder Yeah, i m an engineer and without hesitation nor a doubt i can classify this as pr0n :D
@@Draalo "Talk nerdy to me... Be gentle... I'm a n00b."
Oh man, I'm sure some Hollywood guy could come up with an entire series based solely on that line.
If you ever used an IBM 9121 you are now Mature For Sure
I love how exited and enthusiastic he is about this thing. I have no idea about computer science so I've got no idea what he's talking about. But I am a fitter and turner and this is really impressive from a mechanical engineering view point
5:45 "The Fart of Death", breaking the processor's seal under non-professional conditions by unauthorized personnel.
I bet my old roommate from back in the day could tell you about those substrates. He used to work as a mechanical engineer at IBM East Fishkill, NY, where they made these things. I'll check with him and see if he can say anything about it.
Hello there! Used to work there myself as a software engineer! Started in 2000. I probably could have shoveled a trash can of this back in the day. There was so much tech everywhere. worked in the 310 building.
My dad used to design ‘em. He could probably tell you all about how this one works still...
@@normantor you wouldn't have found any finished TCMs.......BAT (Bond Assembly and Test) was done at another site.
Hugh G. Rection Close! The area was settled by the Dutch
@@JohnRunyon would love to have a chat with him if true.
Fun fact: Dave, while counting the pins, only had to start over twice!
Your nerdy excitement is infectious! I generally don’t understand what you’re talking about but I enjoy watching and trying to learn regardless
who else just died inside when he broke that chip. omg. "this is rare, albeit ill destroy it for you guys"
It's a teardown.
9:00 This is borderline porn territory. There's slugs going in and out of a hole. And copious amounts of lube
continues at 20:48 ...
Rule 34 always applies. Always.
@@ahsmeg4069 what about Godwin's law? The question I have is does it run Crysis!?
It looks like a mini nuclear reactor.
First thing that came to my mind when I saw it open was this:
www.thingiverse.com/thing:2720375
I've been wanting to make one with metal instead of printed plastic.
I wonder what would happen if you replaced the pistons with fuel pellets...
@@TauCu
Reality may get altered
Tumdéaux Looks like a Teminator 1 brain from the future.
@@srfrg9707 Nah it's the Star Trek Enterprise computer
The fact that this level of manufacturing precision was available in the 90's just blows my mind to what is available these days.
Really? lol.
IBM were the best of the best in computer manufacturing magic, starting with Appolo-era computers, back in the 60's. We can recognize their design philosophy in this module. Lots of squares, all seemingly identical, but with different underlying functions.
High-precision manufacturing is still available, and still used. It is just expensive, so it doesn't go into your ordinary household electronics.
The machining to make a modern cellphone is fully automated and as good as, or better than what built this.
Mechanical watches has existed for a good long while bub. Mechanical manufacturing precision has existed for well over 200 years now
23:20 I understand what and why you're doing but my heart, like the part, broke that instant. Great video though!
I absolutely loved this teardown video. I was a user of the IBM 3090 ES9000 systems back in the days when this was the latest thing. I did stress and performance testing of Prodigy's systems using IBMs Teleprocessing Network Simulator (TPNS) which was a VTAM application that I used to simulate up to 256 users dialing into our Point of Presence systems or thousands of users logged into the service. I was both a hardware engineer and programmer and I really liked your enthusiasm about this technology.
5:46 The CPU ASMR coupled with Dave giggling has got to be the most pure thing ever
finally, someone else who gets it!
I guess every Auzzie has his day, I mean Dave's always happy, but this, this is an all new level of euphoria.
i mean look at that thing
he's suffering from Covid Lockdown Syndrome! That's where the most inane event seems exciting due to lack of stimulus. I love you Dave.
@@muzzaball Dave is in NSW : he isn't locked down. Only those virus spreading morons in Victoria are.
@@Spookieham well that is harsh. Most of the eastern seaboard heavily populated areas suffered. Anyway, we are all here for fun and education.
I've never seen him so happy.
2770 pins? Laughs in Treadripper with it's 4094 pins. :D Jokes aside, that thing is insane! 9 Gigs of RAM in 1990 is just incredible. That O-Ring noise sceams for RELEASE THE SCHMOO!
Mainframes still supports insane amounts of RAM. IBMs latest z15 support 40TB, which is ten times that of an x86 server.
@@LeonardTavast Why are you comparing current mainframes to x86 which hasn't been used for like 15 years? Current x86-64 CPUs support 256TiB
@@frother What's a TiB? Is that like a titanium boron byte? Is the Ti better for Schmoo Releasing? So much Schmoo.
@@rickfeith6372 It's the more correct term for 2^40 bytes. 1 TB is 10^12 bytes (1000 GB or about 931 GiB which is what Windows/Linux shows to the user).
This isn't an electronics porno, it's a snuff film.
right?!
Quite moist indeed.
You destroyed a thing of beauty
@@evil-wombat right?!
@@_QA_ NICE ONE!
That moment when a CPU from 1990 has more in common with a car engine than the current generation of CPUs
Wow youtube algorithm actually recommended a good channel.
Pb is lead... so it's 97% lead, 3% tin.
Dave Sings!!!
No. No I don't.
@@EEVblog Everyone sings! You mean you don't even sing in the shower? Come on, fess up!
Queen Fran!!
@@EEVblog The mineral oil dissolves solder joints apparently - there is a video of Linus Tech Tips from 2-3 years ago where they revisited a Mineral Oil build and the SMCs were literally falling off from the MoBo and other parts of the system.
At 19:33 you wrote Tin/copper, the image says 97Pb/3Sn -- that's 97% Lead, 3% Tin. :P
Great video though.
PD: I believe it's usually pronounced mo-lib-d'num, or something along those lines. I'm not english native, so excuse the crude pronounciation guide. :P
beat me to it! good old plumbum and stannum,
In the US, it's usually pronounced mo-lib'-di-num.
Mostly-"Plumbum" solder is probably very soft. So given such small sized pads underneath, it was so easy to slide the cap chips off.
@dothemathright 1111 they had to bring those over just to have something halfway safe
@@pacman10182 Emphasis on halfway. They may be generally sociable, but you don't want to make a sheep angry.
Yeah, back in the day, we used to "force" technology. Our company used "hybrids" to attach power devices (dice) for audio output directly to silver traced (like a PCB) ceramic substrates. Then we wire bonded them and encapsulated them in a special epoxy. This was in the 1980's. We used Beryllium oxide substrates for their thermal conductivity. They actually conducted heat better than aluminum. Today, I could design the same thing using off-the-shelf parts and it would be better.
Of course these (in this video) parts were soldered down, but the oil bath improved thermal conductivity. What a mess! Our company actually used peanut oil in an aluminum tray (aluminium to you Australians lol) in a resistor heated bath to regulate the solder bath carefully. Odd to think that worked, but it did very well!
I'm so glad those days are over!
In my first job in back in '94, there was one guy in our shop who worked out of the office. I didn't really understand what he did, but as the rest of us were field service warriors, staying in one place looked pretty nice. One day I heard him quip, "well, I just earned my salary for the year!" Curiosity piqued, I asked what he meant. Turned out that the customer he'd been working with had been informed (by Big Blue no less) that they were going to have to replace a processor module in one of their systems, but this guy (along with others, no doubt) had figured out what the real problem was, and got the customer back in the air. No processor module needed. When he told me how much these things cost, I understood!
Really great vid, by the way!! Possibly Dave's most epic geek-out to date!
Although its "mips" rating is quite low in comparison to chips today (and even back then) its seriously not the whole story... like the S36, S38, AS/400, RS/6000 etc. its all about the sum of its parts. With disk controllers (additional processors) that could sync the rotation of multiple disks (to stripe the data so that multiple spindles returned the data at the same time) and memory controllers (again additional processors) that could be asked for data and then left to do the job without main processor intervention (and interrupting the main processor when done) and with data buses that were "uncoupled" (again separate controllers; picture a 60 lane wide Ethernet for a rough and ready visualisation) the "processor" was not much more than a mega calculator and also had an instruction set that was really specific to transactional processing (think lines on an invoice, stock levels, bank transaction) and also included calculation instructions that were "fixed point & length" processing with half adjust and truncate/overflow for calculations specific to currency work and item levels. The processors were brute force machines, data in - calculate - data out, that left the rest of the work down to other processors/controllers. This meant that the spec-perf/transaction processing scores were way out in front of other processors of the day, even ones that were a lot faster... mind you, you paid for this in real money! You could also "lpar" the machines into virtual machines (even by fractional "core" metrics) many years before PC's and the machines also had "service" processors that would watch over the whole box, and if the machine failed to boot would give you the exact reasons why (disk 14 failed; co-processor under voltage; database not in consistent state on shutdown; jon smith smoking in the computer room; hello dave). I know I sound like a fan boy who has gone all stalkerish over computers, but these machines were light years ahead of everyone... even now PC's are only just (well a few years back) starting to get some of the IBM stuff from way back when (like error codes not beeps; service processors (ipmi); AMD chiplet design with uncoupled memory controller; etc.).
That was very accurate and fair representation of the real IBM "affairs" in computing back then. IBM was what it was for a reason.
I agree, it's not about pure processing power... But still, for a mainframe/server system, only 20 MIPS (for the most expensive model), that's comparable to an Intel 486DX at 33MHz, which was also around in 1990...
Maybe the 20 MIPS rating is meant to be per core, instead of the whole CPU assembly/package ???
@@HPPalmtopTube in 1990, we used to run 700-1500 'office' users on what wasn't the largest IBM mainframe by far, but no way you could do that on a 486 based machine. Its all about independent handling of i/o. You'd notice it wasn't actually that fast when asking cpu intensive tasks, but it remained amazingly responsive with for the time very large numbers of users.
This kind of system is not for calculating primes (or such) but to facilitate hundreds or thousands of users concurrently, all of them doing relatively low intensity tasks but needing good responsivity. Say the hundreds or thousands of office workers at a bank, air traffic control etc.
Its not so much 'per core' processing power, but taking lots and lots of work that doesn't have to be done by the cpu completely away from the cpu.
My job still uses AS/400!
@@vprincessfan12 I still think its was one of the best computers ever invented. From the "object based" OS to the ease of programming using RPG (obviously both had their roots in earlier incarnations). Such a shame that IBM didn't keep promoting it after its various name changes. Its still there, still being updated, but few outside of the current users know it exists.
That is a Freedom Hercules TCM. It was actually a backup package for the System 390 because IBM was having technology issues with the glass ceramic substrates that were originally designed to go into the System 390. They did solve the issues with the glass ceramic and used these substrates also. The vias in the glass ceramic modules were faster due to copper being used.
Never send Dave the Mona Lisa or something like that for review
It will end up as a Banksy...
In the case of Mona Lisa, i'm sure he'll make her look better at the end no matter what 🤣
"Let me just try a pair of pliers to get this speck of paint off... oh I just tore right through the canvas, there you go, 10 cm gash in the Mona Lisa... sorry Mr Davinci..."
Never send Dave an anvil either !
Imagine what "Power That Should Not Be" have in their hands at this point in history, 30 years later :D
"louis rossman eat your heart out" LOOOOOL!
yay I just started laughin so hard I had to rewind and re-watch a few minutes ;) he's gotta see that "desoldering" :D
lmao, i had to hit subscribe as soon as i heard that :))
@@stefan0ro this guy is fucking hilarious.. You'll love him if you like rossman. Both extremely skilled in their trade.. Eevblog is more diy friendly though 🙂
I have no idea who that is. I’ll have to search I guess
@@codycast not all videos are great, but it's a really good channel. Don't get discouraged if you hit one or two somewhat boring.. see for exampel ruclips.net/video/6dwqxsDHkKQ/видео.html or a bit older ruclips.net/video/w3c2dOw6Frg/видео.html
I'd love to see the instruction set / opcode list. Processor speed and all that.
Includes the "Insert Characters Under Mask" instruction.
@@tripplefives1402 this might be a reason to dust off my old pc and take a walk down memory lane.
@@tawpgk You can run the Hercules emulator on a Raspberry Pi computer. i've tried it and it works.
@@CarlJohnson-wg6wy thanks, I have many Pi's...can I make like a 32way sysplex? Hahaha. I think I may try that on a pi. Get my mind working in a happy zone. Does Hercules include a HLASM? Just dying to issue EQUs, LTORGs, DSECTs, AIFs/AGOs
2020 - Covid-19, Hercules, and reliving my 1970s youth coding in BAL. Perhaps I'm going senile...lol
When I worked at CAE in the 1990s they had an IBM 4361 System 370 that was the little brother of the System 390 that took up an entire room. We bought an IBM P/370 that was a "Personal 370" which was an OS/2 server with a plugin card that contained a S370 card to replace the 4361.
In 1996/1997 was upgraded to a P/390 that was a "Personal 390" with the same setup, a plugin card in an IBM OS/2 server. IBM ran all the peripherals, disk drives, tapes, I/O, etc. virtually in OS/2. To the board it was a full System 390 that was running, we even had a 9 track reel to reel tape drive connected on a SCSI card. It was very impressive technology, this was supposed to be used for development where you could build and run your system with being on the production main frame.
So the System 390 was the computer class and you bought the capabilities you needed. If I remember the software prices were based on the number of CPUs you had. With one CPU, as we had in the P/390, FORTRAN was $10,000 to $20,000, a bargain compared to the $50,000 we paid for an AN/AYK-14 assembler program.
What you have there is a beast, all those CPUs together on one massive module would have cost a fortune. Mainframes were all about the I/O and parallel computing more than raw MIPS on a single thread.
I believe System/390 is the processor architecture, and ES9000 was a specific server built around the System/390 architecture.
Correct. S/390 is rooted in the S/360/370 systems architecture with a few upgrades along the way
Excellent video. Deeply impressive technology which is now 25-30+ years old. Absolutely spectacular.
This was excellent! Thank you!
I love the idea of having a part two with some special imaging to see into the ceramic layers.
@EEVblog
This is 97% Lead+3% Tin solder. You can see it on a document you pulled up at 24:42 . This explains why it is so soft, it is mostly Lead.
Yes I confused it with standard 97/3 lead free
Great teardown, it's really a thing of beauty! But isn't that solder 97% lead and 3% tin? It mentions 97Pb/3Sn in the text you showed.
Oops, I mistook it for the 97/3 lead free Tin/Copper stuff. Exact same ratio. D'oh.
Never heard of 97/3 Leaded stuff.
@@EEVblog2 That's why it is so soft and you are able to rip the caps off!
@@EEVblog2I think it is used here because for the bga balls cause its more forgiving to thermal expansion/contraction (i.e. softer)
www.researchgate.net/publication/267584624_High-Lead_Solder_Ball_Characteristics_in_BGA_Package
(Doubt the higher melting point is relevant)
Tin makes the solder stronger but stiffer, they must have wanted something reaallly compliant, it seems unusually high in lead even considering. No wonder it's so easy to remove, it's basically pure lead, which is ten times weaker than tin
Edit: whiteboard chalk... I need to sleep XD
Double edit: hardness =/= strength
Yeah, basically pure lead. You know, that stuff you want to play around with your naked fingers and maybe take a nice lick.
@@RealNovgorod I used 80-20 solder for years in the 70s no drainbamage hear................
“No i didn’t just crush the pins it has like tabs in the bottom” also they were already crushed 😂
Dude, your references are out of this world, everybody knows it.
Interesting ! I've actually worked in the french IBM assembly line back in '95 and with that kind of processor. There where some air cooled, other with freon.
First time I've see one dismantled. If a chip (or something else) was not working, the whole CPU assembly won a free ticket to USA.
Thanks for the back-in-time travel !!
System is just the code name for their instruction set architecture. Series is their keyword for architecture series. This was usually correlated to the OS e.g. Z-Series with Z/OS. ES was shorthand for Enterprise Systems. So System and ES are interchangeable.
I love how back then it was state of the art, and now it is a little more powerful than a 2$ Arduino Nano
Well not really so far as an Arduino Nano, but I Raspberry Pi I sure imagine beating it.
It's also interesting how they work, as far as I know those mainframe CPUs (and mainframes in general) are highly optimized for absurd parallelism of operations, they're closer to graphics cards in how they operate (sort of... maybe quite far still tho... Mainframes are weird).
@@Kalvinjj
It might beat it in MIPS, no way your Raspberry Pi would beat it for it's IO ability. That's what Mainframes were/still are excellent at, handling thousands of user terminals, ATM's etc all with extremely quick response times.
@@Mark.Brindle Indeed, for IO ability it still probably takes another mainframe CPU to beat it.
Maybe with PCIe interfacing, but then again what kind of pre-processing would be required to mediate the data I have no idea.
@@Mark.Brindleexactly! apples and oranges...
@@Mark.Brindle Back then IO was not sending TB of data per second. I'd imagine a normal CPU could handle all the IO workload with ease, not because it has a huge bandwidth, but because it can serve each request in a fraction of the time. And probably still has a much higher parallel bandwidth anyway. (I have no idea what I'm talking about and would like to know).
As a new microelectronics engineer, this is like candy for my eyes
Oh my god I love how excited you get about this. It's adorable
I started my professional career as a cobol programmer and then as a systems programmer on IBM mainframes back in 1993. We swapped out our 3090-300J for an ES/9000 9121 so really appreciate the teardown to get a look at where those electrons flowed...
Takes a lot of magic to trick rocks into thinking IBM!
I remember telling my friend about a new processor called a Pentium
and how I thought it would change computers
No galvanic corrosion AT ALL thanks to the oil. Pretty awesome / amazing after almost 30 years!
Dave, this is the first of your videos that I've ever watched. I am glad that this was the one, cause thanks to your, shall we call it "enthusiasm", I am hooked.
It seems like a portion of you hasn't ever grown up, and I hope it never does.
Im looking forward to watching all your past stuff, & future stuff too.
Have a fantastic day!
Wienerschleiden and springensproingen? Did AvE design this thing?
That precission metal work on the frame. Perfect holes, perfect threads. Machine cut.
IBM Poughkeepsie manufactured gun-sights in WW2, and mechanical tabulators always required precision machining.
Would love to see the performance of this compared to a modern processor :)
Release the SCHMOOOO!
Someone had too much AvE
He drops an AVE quote at 1:44... I always wondered where that came from!
I worked on 2 different 9370 systems back when they first came out - I believe they were the little brothers to the system this came out of. Very interesting to see this. Needless to say, this was one of those systems that sat in an airconditioned room 24/7/365.
I’ve got a couple of these at home, they came out of one of several 3090 mainframes at an Australian bank. I worked for a 3rd party maintenance company. We didn’t buy spare units from IBM, we just bought old mainframes and stripped the parts from them. The 3090 was water cooled and the TCMs had a water jacket bolted on top, to replace a TCM you unbolted the water jacket then undid the retaining screws, torquing up the screws on the new unit. The main thing was to make sure that you’d powered down the correct frame (you only powered off the processor that you were working on, the rest stayed up) I heard of an engineer who pulled a live TCM, there was a big flash and bang and the machine was offline for several weeks while a new backplane was installed. Those suckers drew a whole lot of current.
27:28 I reckon J. U. Knickerbocker was just in it for the glory.
funny
Comparing MIPS on a mainframe with a 486 isn’t quite so easy. Unless someone knows the complete facilities involved it just isn’t an apples to apples comparison. Dave makes a valid point in bringing the IO into the picture. As for the porn comment I think you’ve hit a new level of bliss in a tear down.
I was barely 2mins into the video and I had to call the Police on Dave. I hope there were no children watching! 😀😀😀
I am probably the youngest viewer here I’m only 15. I have a love of electronics and computers and I make teardown videos taking thing apart and explaining how they work.
Meet Connor and his mainframe: ruclips.net/video/wJyiHsfJLEI/видео.html
@@zf4hp24 Also see his talk: ruclips.net/video/45X4VP8CGtk/видео.html
IBM always had amazing mechanical engineering (check out the mechanical masterpiece that was the Selectric sometime, or their high speed tape drives). It makes sense that they'd see higher performance as being mostly a thermal packaging problem, especially back in 1991.
I haven't watched an EEVblog all year!! Thanks for still getting at it!
LOL, just yesterday i have been thinking about how cool was the video about that old IBM HDD :) 5 y.ago
@ 20:29 Louis Rossmann ROFL
Louis is probably creaming himself imagining all that oil as flux. XD
@@Brooo007MC lol, yes, he is so into flux :D
23:33 That's like something from _The Andromeda Strain_
YOU READ MY MIND 👁👄👁
Imagine a AMD Zen CPU package like that for datacenter applications. that would be insane.
And the quasar space-time friction singularity drive you'd need to power it.
www.popularmechanics.com/technology/design/a28816626/worlds-largest-computer-chip/
"This Is the World's Largest Computer Chip
It's the size of an iPad. And it's ready to unlock the power of AI"
There were up to 9 TCMs on a board. That board was multi layer ceramic (I don't remember how many layers)and cost $xxx,xxx as well.
I love your enthusiasm, i cant truly tell you are passionate about this, i as well that processor is so cool.
the chip side of those "pistons" look hand lapped and the spring side is just plain. I wonder what the finish looks like without the oil.
@Right Round That's a fair point. It would be interesting to see one dry.
@Right Round They would be sanded and polished by a machine, but not lapped. Lapping is a very specific process, and the term is usually incorrectly used by computer enthusiasts.
That was probably the last time a "Made in USA" sticker was used 😏
probably only the sticker itself...
@Lassi Kinnunen Nah, not "entertainment stuff like racing engines", but Military; that's where the big bucks are for specialist engineering in the USA these days.
@Lassi Kinnunen Er... Medical? Defense? Aerospace?
"Made in USA"
Ahh the good ole days of technology and processors :)
I remember some more too about IBM TCM's. On the 3745 they were smaller, with just one square block of pins, I don't recall how many. We'd receive a m/c from the assembly line ready for power (These were mostly "Y" configured 415v/ac 63a m/c's) and logic testing. After power testing we'd populate the m/c with all the logic, including the TCM's. We'd have to lay a static mat to stand on, while wearing our ESD straps, which were all tested before each logic population. We'd unbolt the fans and unscrew the heatsink. Populate the logic cards, then came the TCM's. They came in a 400mmx400mmx400mm shock and ESD proof box. IIRC they each cost GBP5000, back then that was more than my annual salary. We had a special gauge that was used for checking for bent pins. It was a 3 or 4mm, machined, sheet of plastic that "floated" over the TCM until it fell in to place. Each pin had a matched hole and it was easy to see if there were bent pins (There never were and you'd never cause one or lose your job). So we'd unpack the TCM, unwarp it form the ESD bags, check for bent pins and gently place the TCM on the "socket". This was a huge thing, an over grown 486 like thing. We would then insert a wrench in to the TCM and socket, the wrench having a cam like end to it. We'd rotate the wrench and the TCM would be "wound" in to the correct seating position on the socket and locked in to place. We'd attach the heatsink and fan and power on for the logic test. The "BIOS" would go out and "discover" what hardware was in the m/c and then we'd run diagnostic and soak tests. I did have one 3745 prime power box blow up when earth went open circuit! While debugging one m/c one night, a sprung loaded component struck me in one eye. The damage is still there, optometrists always ask about it, when I go get new glasses. Fun days! Thanks for the memory.
I don’t think I have ever been as joyful as him about anything for that long a time.
Wow this is amazing, I must have worked with those a couple of times but never as close to the CPU. Just hope you don't begin to repair watches.
Paperclip at the ready!
Sorry couldn't help myself.
Could you imagine doing the routing for those 63 layers? I know it would be a whole team of people doing it. But would still be a lot of traces and via for each member to be responsible for
Gets even more insane when you think about that the module was designed to keep working at its nominal capacity even after a certain number of random(!) chips failed.
@@HenryLoenwind you read the reply from
John Bellas? interesting how they used a paste to form the vertical interconnects layer by layer, alignment woes, amazing to think of these guys forging ahead into the unknown to make something so specialized, like the women who wove Apollo memory modules.
This video sparks joy. My thoughts and prayers go out to the IBM engineers.
i'm a geo by trade, and before COVID i spent over 10 years working in a greenstone gold deposit. as soon as i saw that moly gleam on the contacts my heart was aflutter. it's an indicator mineral in greenstone gold deposits. i was very happy to read the comments by the former IBM people in the comments. thanks all!