Agreed. When you program in machine language and optimise every single step, you can do amazing things with low resources. Sadly, today much of the software we use seems to grow to take up all the resources, with terrible optimisation. That said, the software is far more complex I guess. My first computer at 1k of RAM... was amazing how clever some people got writing code for it!
3270 terminals were _highly_ intelligent,_ having their own CPUs and "display programming language". Combined with CICS, they were an early form of client-server web computing, where the end user types everything into the terminal and then pressed the XMIT (aka Send) button. That transmitted the whole block of data to the mainframe. Until you pressed XMIT, the mainframe completely ignored you. This is in stark contrast to minicomputer OSs like Unix/Linux and VMS, where the OS must process each and every keystroke and cursor movement. Bottom line: web-based client-server computing is just a new and horribly inefficient version of a programming paradigm that's been around for *55 YEARS.*
The IBM mainframe does not have to keep an separate active program for state for each user. When I used to program CICS, back in the 1980s, you could store context in hidden areas on the user's terminal and/or in something call TSQ - Temporary Storage Queues. When the user pressed SEND, the CICS program would fetch the context based on the data from the user, do the work and return a new screen image back to the user. I do some PHP web programming these days and my programs are designed very much like a CICS program since there are cookies that come from the user and I can thus restore the context, do the work and return a page back to the user. The web server needs to remember little of a particular user's transaction. The web server has the equivalent of TSQs with session info. The comment about OS processing every keystroke is possibly true for a UART connection but I think Ethernet sends info in blocks(?) It is definitely not "... horribly inefficient ..." Computer folks are not going to make badly designed web servers.
Very nostalgic. My early education and professional careers were “punched out” on systems such as this. Unix systems became my bread and butter later and remain so. Thank you for the nice trip down memory lane. It’s amazing how many important systems were birthed on this type of hardware.
This was just before my day. I started college in 1980. I don’t remember the Winchester drives in my schools computer room. What I do remember was the PDP-11 in the math department rather than the data center and I had no idea why. Just watching that printer fold shut brings back memories of my work study days changing paper; vacuuming the printer and bursting the outpuyrv
@@jblyon2 honestly, I think "They" expected operators to all run out, and press the Bid Red Halon Button _as they were leaving._ Of course, sometimes idiots put the halon button deep inside the room, far from the door.
@RonJohn63 Lots of bad designs out there. I saw someone with a story posted somewhere that the halon button at the loading dock entrance was next to 2 other buttons, one of which was also red and commonly used. Naturally it got hit by mistake one day and everyone had just enough time to get out, but they all got out thankfully. I've also heard of similar bad designs for the quench buttons for MRIs.
@@jblyon2oh, they still use fire suppression systems with similar properties to halon in new installs. They are just hcfc or hfc based not CFC based. I have to deal with one at work.
Thanks for the wonderful film of old computer tech. Today, a $15 "Raspberry PI zero 2 w" linux computer can run IBM MVS 3.8 at the same performance level of a 1979 IBM 3033.
Pretty impressive computer system. That was quite an interesting video. Incidentally, I grew up near Asheville in Western NC which is just a short drive from the upstate and Clemson. Some of my father’s relatives had a close connection to Clemson. One was a secretary for years in the college of engineering and two others graduated from Clemson. Thank you for this video.
In this day and age of solid state storage, are spinning rust disks considered sequentially accessed media? You still have to wait for the disk with the correct data to rotate and pass under the read/write heads, whereas solid state truly is random access, as there is effectively zero latency for access.
It is mainfram storage so yes RAM. 8 MB was expensive. The 3033 was at its launch in 1977 100 % faster than the previous fastest processor. Expect the leasing cost over 4 years for 8 MB to be millions.
The singing printer is awesome
And it was stunningly loud (even under the insulated box).
Lol of course it's Tiger Rag
🤣
All I could think was "This is a pretty nifty music box, it even prints...oh...oooooh...that printout is the demonstration...ahhhh"
It's amazing to think that a computer with 8mb of memory could do all those things and support all of those agencies.
Agreed. When you program in machine language and optimise every single step, you can do amazing things with low resources. Sadly, today much of the software we use seems to grow to take up all the resources, with terrible optimisation. That said, the software is far more complex I guess. My first computer at 1k of RAM... was amazing how clever some people got writing code for it!
3270 terminals were _highly_ intelligent,_ having their own CPUs and "display programming language". Combined with CICS, they were an early form of client-server web computing, where the end user types everything into the terminal and then pressed the XMIT (aka Send) button. That transmitted the whole block of data to the mainframe. Until you pressed XMIT, the mainframe completely ignored you.
This is in stark contrast to minicomputer OSs like Unix/Linux and VMS, where the OS must process each and every keystroke and cursor movement.
Bottom line: web-based client-server computing is just a new and horribly inefficient version of a programming paradigm that's been around for *55 YEARS.*
... there was a lot of swapping and let us not forget "initiators", job batching and job prioritization
The IBM mainframe does not have to keep an separate active program for state for each user. When I used to program CICS, back in the 1980s, you could store context in hidden areas on the user's terminal and/or in something call TSQ - Temporary Storage Queues. When the user pressed SEND, the CICS program would fetch the context based on the data from the user, do the work and return a new screen image back to the user. I do some PHP web programming these days and my programs are designed very much like a CICS program since there are cookies that come from the user and I can thus restore the context, do the work and return a page back to the user. The web server needs to remember little of a particular user's transaction. The web server has the equivalent of TSQs with session info.
The comment about OS processing every keystroke is possibly true for a UART connection but I think Ethernet sends info in blocks(?) It is definitely not "... horribly inefficient ..." Computer folks are not going to make badly designed web servers.
@@ablebaker99it's not the web servers, but the slow and bloated JavaScript fed by the web server.
Very nostalgic. My early education and professional careers were “punched out” on systems such as this. Unix systems became my bread and butter later and remain so. Thank you for the nice trip down memory lane. It’s amazing how many important systems were birthed on this type of hardware.
I remember seeing pictures of a 3 MB hard drive from the 1960s that was being carried by a forklift! We’ve come a LONG way!
This was just before my day. I started college in 1980. I don’t remember the Winchester drives in my schools computer room. What I do remember was the PDP-11 in the math department rather than the data center and I had no idea why. Just watching that printer fold shut brings back memories of my work study days changing paper; vacuuming the printer and bursting the outpuyrv
I was there, worked as an operator in martin Hall. Breagan terminals and 3033's with 8MB of storage!! This was the main building, as I remember it.
And now my phone has ten times the capacity of that entire room.
11:51 Will suffocate anyone who's stuck in the computer room, but won't damage the computers.
Funny how halon stopped being permitted when the computer equipment became less valuable than the average lawsuit payout for wrongful deaths.
@@jblyon2 honestly, I think "They" expected operators to all run out, and press the Bid Red Halon Button _as they were leaving._
Of course, sometimes idiots put the halon button deep inside the room, far from the door.
@RonJohn63 Lots of bad designs out there. I saw someone with a story posted somewhere that the halon button at the loading dock entrance was next to 2 other buttons, one of which was also red and commonly used. Naturally it got hit by mistake one day and everyone had just enough time to get out, but they all got out thankfully. I've also heard of similar bad designs for the quench buttons for MRIs.
@@jblyon2 "I saw someone with a story posted somewhere".
lol
@@jblyon2oh, they still use fire suppression systems with similar properties to halon in new installs. They are just hcfc or hfc based not CFC based. I have to deal with one at work.
Wow drum style memory still existed in 1980. Ahh the 2305 was already end of sale by 1980 it came out in 1970. The last of the drum era.
Well explained in detail😁
Very cool.. we have come along way..
I think some current kitchen appliances have more computing capabilities.
The "UPs!" system!
I burned my eyes out on those monitors for many years LOL
If you want one of these for your own, just fork over 4.5Mil$$ - or run Hercules and install TK4- a pre-configured MVS 3.8j for free.
Thanks for the wonderful film of old computer tech. Today, a $15 "Raspberry PI zero 2 w" linux computer can run IBM MVS 3.8 at the same performance level of a 1979 IBM 3033.
I'm old cus I remember this. 🤣
Pretty impressive computer system. That was quite an interesting video. Incidentally, I grew up near Asheville in Western NC which is just a short drive from the upstate and Clemson. Some of my father’s relatives had a close connection to Clemson. One was a secretary for years in the college of engineering and two others graduated from Clemson. Thank you for this video.
This is 2024.... Runnnnnnn
Oh, 1980. You so crazy!
The way he said modem!
6250 BPI does not mean bytes per inch it means bits per inch
11:27 A motor generator, changing electricity from 69hz to 415hz jeezemabob
In this day and age of solid state storage, are spinning rust disks considered sequentially accessed media? You still have to wait for the disk with the correct data to rotate and pass under the read/write heads, whereas solid state truly is random access, as there is effectively zero latency for access.
There is always delay caused by capacitance so it's never zero even if drives are solid state.
3033 was announced I believe in 1977 and was a big deal back in the day.
When I was a little kid, my mom worked from home and had a big machine in our basement and she entered data on punch cards, kind of like a typewriter.
That sounds like a Telex machine... the channel CuriousMarc has a great look at one
Is 8 megabytes the ram or the CPU’s cache?
Kind of sounded like cache,the way it was described
Cache memory was measured in KBytes back then. I recall working on a Honeywell mainframe back in 1982 that had 4KB of cache memory.
It's basically main system RAM. The CPU will have a very limited instruction cache.
It is mainfram storage so yes RAM. 8 MB was expensive. The 3033 was at its launch in 1977 100 % faster than the previous fastest processor.
Expect the leasing cost over 4 years for 8 MB to be millions.
It sounds so small until you remember that csects and dsects were limited to 4k.
PL/1 is the language of the future!
Might want to edit down the tail end a bit. Great stuff otherwise.
Why ? Why the beard and the glasses ! 😂
Engineer chic of the day.