Not about photography, but computers changed around 1980

"Perhaps something odd about your system somehow?"

Maybe just that we were shuttling off a lot of data over Ethernet to be processed by a second computer that then returned results. It was a constant stream of data with no let-ups. I don't know. We never really figured out the problem.

At a later time, I found our programmers liked one particular USB driver, but our electrical engineers liked another because of the support in the APIs they liked. Both sides refused to adapt to what the other side preferred, so we were shipping systems that had two USB different drivers installed and that that would cause intermittent hang-ups. Our customers for that product were fine with just rebooting the computers at the star of each shift and that seemed sufficient to avoid problems.

Both our electrical engineers and programmers would show me their part of the software working fine day after day with no hang-ups on their production equivalent test computers. Neither side could not seem to understand I was not interested in how nicely things ran with on systems with only one USB driver if we were shipping customers systems with two USB drivers.

I am glad to hear you were able to get reliable 24/7 operation with Labview. NI always acted like it should work, but we were trying to get into a world I don't think we really belonged in as we didn't know people operating 24/7 in production lines that could advise us with lessons learned.

OpticsEngineer said:

"Perhaps something odd about your system somehow?"

Maybe just that we were shuttling off a lot of data over Ethernet to be processed by a second computer that then returned results. It was a constant stream of data with no let-ups. I don't know. We never really figured out the problem.

Click to expand...

Possibly. The space craft system was digitising data from a spectrum analyser constantly, 16 bit, 200MHz BW but it only went to the local PC no network involved as such. The NI control and digitiser were ISA slot based as I recall.

OpticsEngineer said:

At a later time, I found our programmers liked one particular USB driver, but our electrical engineers liked another because of the support in the APIs they liked. Both sides refused to adapt to what the other side preferred, so we were shipping systems that had two USB different drivers installed and that that would cause intermittent hang-ups. Our customers for that product were fine with just rebooting the computers at the star of each shift and that seemed sufficient to avoid problems.

Both our electrical engineers and programmers would show me their part of the software working fine day after day with no hang-ups on their production equivalent test computers. Neither side could not seem to understand I was not interested in how nicely things ran with on systems with only one USB driver if we were shipping customers systems with two USB drivers.

I am glad to hear you were able to get reliable 24/7 operation with Labview. NI always acted like it should work, but we were trying to get into a world I don't think we really belonged in as we didn't know people operating 24/7 in production lines that could advise us with lessons learned.

Click to expand...

It seemed to be a way we could get some systems working quickly. Later we started using AgilentVEE as they had panels created for some of our kit making things quicker still. Similar idea I guess.

Erik Kaffehr said:

I was a postgrad student inte early eighties, although my field was reactor physics, most of my work related to computers, that was the era of the micro processor.

At that time, I didn't appreciate all the work in the previous era, being immersed in the new era.

In a few years I would work at Studsvik, a nuclear fuel testing facility. We had some mainframe from CDC for our calculations and the measurement computer was a HP 1000F.

Three years later, when I left, my department was running a Sun Sparc Server that did the work the CDC needed four days for in four hours.

What really happened was a revolution in how computers were built, with cables being replaced by curcuit boards and circuit boards being replaced by large scale integrated circuits and programmable gate arrays.

Understanding the revoution that I have been involved with is a fascinating experience.

Best regards

Erik

Click to expand...

It has flipped a couple more times...

Many PCBs were replaced by cables in the Cray1.
cray1 - Search Images

Then there was a transition to PCB backplanes to replace the cables (Density).

Now, open your favorite image search tool and start searching for images on "NV72 cable assemblies" to see what is in the latest NV GPU racks connect chip to chip between boxes used in AI.... with a optimized compute fabric that is not Ethernet (albeit, there are also scale-out paths that are Ethernet for things that tolerate the latency)
NV72 cable assemblies - Search Images

Today's cable systems have a Nyquist frequency close to 56GHz... and pass data at 224Gbps using PAM4 modulation.
Next Gen-Data Centers | Molex



What is old is new again... and again... and again...

From the Economist, April 22, 2025

"Generative AI is creating a new boom in digital tape storage. In 2023, tape manufacturers shipped 153 exabytes of - that is, 153 billion gigabytes - of tape storage, a record" ...... "Google is one of the world's biggest purchasers of modern tape libraries."





Back around 1970, Dad told us kids after dinner, let's go over to the computer center and watch them run payroll. It was kind of a show, checking in with security, then watching a room of 40 tape drives with the reels spinning fast, slowing down. advancing slowly, backing up, then taking off again, over and over, the tape flowing up and down in the vacuum columns.

OpticsEngineer said:

From the Economist, April 22, 2025

"Generative AI is creating a new boom in digital tape storage. In 2023, tape manufacturers shipped 153 exabytes of - that is, 153 billion gigabytes - of tape storage, a record" ...... "Google is one of the world's biggest purchasers of modern tape libraries."

Back around 1970, Dad told us kids after dinner, let's go over to the computer center and watch them run payroll. It was kind of a show, checking in with security, then watching a room of 40 tape drives with the reels spinning fast, slowing down. advancing slowly, backing up, then taking off again, over and over, the tape flowing up and down in the vacuum columns.

Click to expand...

Rather an interesting metric for tape, I'm used to measuring magnetic tape by the metre, well actually foot. I used to buy 3,600 feet of tape to run at 7.5 ips for 96 minutes. No idea how any if that relates to digital data because I've never used tape storage for anything personal.

For many years we used 1/2" tape cassettes for Quick Access Recorders, they recorded more parameters than the mandatory FDR and were, as the name implies, more accessible. Later we moved to 1/4" cassettes for the same purpose. We'd replace the tape pretty much every day when the aircraft was at its home base. Current technology sees the data being transmitted over 3G or recorded on PCMCIA cards (yes we're behind the curve, usually at least one generation when the design is frozen).

My path has been from wire-wrapping Z80s to IBM 7090 to [a_bunch_of_other_stuff] to well... best shown in this video.

NVIDIA Blackwell: The Journey From Die to Data Center

(4 minutes)

GusPanella said:

My path has been from wire-wrapping Z80s to IBM 7090 to [a_bunch_of_other_stuff] to well... best shown in this video.
NVIDIA Blackwell: The Journey From Die to Data Center

(4 minutes)

Click to expand...

amazing tech, yet 40 years later our kids will never own their own home unless we help them, companies now require 10 office staff yet 40 years ago 1 office staff was required per 10 factory staff. a 20meg hd20 could store 1 million transactions now 20 meg cant even store the screen saver. what have we advanced in

When I first saw a computer that rendered/displayed an image that looks like a real photo with many hues and colours.... JAWS DROPPED!!!!!

There's also the hard drive revolution. From 2-men carry metal disc, to 5.25" hard disk 20MB, to 3.5" hard disk, to 2.5" for laptops.

Also circuit boards from through pins soldering to SMD components (very difficult, to non-repairable)

Movies with computer generated effects were mind blowing! HAHAHAHAHA

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

JimKasson said:

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

Click to expand...

I think a cavity magnetron would qualify as a vacuum tube (valve this side of the pond).

GeoffRG said:

JimKasson said:

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

Click to expand...

I think a cavity magnetron would qualify as a vacuum tube (valve this side of the pond).

Click to expand...

I want to see the design of an ALU using cavity magnetrons!

I recall my Dad coming home from work one day and saying the EE R&D department had a tube-versus-transistor meeting. Young and old, every electrical engineer and programmer (but no upper level managers). There was a lot of discussion of pros and cons.

Collectively they decided from then on out, everything would be designed with transistors.

All the older engineers saw it as a challenging situation. With tubes you primarily designed with current but with transistors you designed to voltage. So they all worked really hard to understand the theory about electrons and holes and npn versus pnp. With vacuum tubes, you really needed to understand the physics inside to design with them effectively. So they thought transistors would be the same.

A decade later my Dad said he didn't understand why they wasted all that time on electrons and holes. It wasn't helpful in designing circuits at all.

JimKasson said:

GeoffRG said:

JimKasson said:

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

Click to expand...

I think a cavity magnetron would qualify as a vacuum tube (valve this side of the pond).

Click to expand...

I want to see the design of an ALU using cavity magnetrons!

Click to expand...

Simply pointing out that not everything has been replaced by transistors, yet.

GeoffRG said:

JimKasson said:

GeoffRG said:

JimKasson said:

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

Click to expand...

I think a cavity magnetron would qualify as a vacuum tube (valve this side of the pond).

Click to expand...

I want to see the design of an ALU using cavity magnetrons!

Click to expand...

Simply pointing out that not everything has been replaced by transistors, yet.

Click to expand...

Indeed. Analog iterative optical machines another route or some chap that likes to build them out of big cogs and shafts

The idea of a TWT or a magnetron based calculating machine is pretty novel.

When I was in school, there was a large corporation doing high power which required tubes that was offering a full four-year scholarship for EE's with summer internships for anyone who would come work for them after graduating. They could get no takers. Everyone young wanted to do transistors.

Ephemeris said:

GeoffRG said:

JimKasson said:

GeoffRG said:

JimKasson said:

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

Click to expand...

I think a cavity magnetron would qualify as a vacuum tube (valve this side of the pond).

Click to expand...

I want to see the design of an ALU using cavity magnetrons!

Click to expand...

Simply pointing out that not everything has been replaced by transistors, yet.

Click to expand...

Indeed. Analog iterative optical machines another route or some chap that likes to build them out of big cogs and shafts

The idea of a TWT or a magnetron based calculating machine is pretty novel.

Click to expand...

I'm trying to get my head around that idea but all I come up with is a baked potato

OpticsEngineer said:

When I was in school, there was a large corporation doing high power which required tubes that was offering a full four-year scholarship for EE's with summer internships for anyone who would come work for them after graduating. They could get no takers. Everyone young wanted to do transistors.

Click to expand...

Probably thinking of future employability.

GeoffRG said:

Ephemeris said:

GeoffRG said:

JimKasson said:

GeoffRG said:

JimKasson said:

silentstorm said:

BertIverson2 said:

Being 87, I went through that era of computers. Today, with a video card in a windows PC which has (I think 43 billion transistors), I try to imagine what the computer will look like in year 2050.

Bert

Click to expand...

I'm the opposite. I tend to think what would a computer with 20 billion transistors, converted back to vacuum tubes into the 1950 looks like. Even with small nu-vista tubes, 20 billion of them is just mind blowingly insane. HAHAHAHAHA

And then we look at those Ai chips, nVidia B200 with 208 billion transistors!!! Think vacuum tubes!!! HAHAHAHAHA

Click to expand...

If you could manage to construct a pc capable of running Windows 11 with 12AX7s, think about how long it would take to boot up.

Click to expand...

I think a cavity magnetron would qualify as a vacuum tube (valve this side of the pond).

Click to expand...

I want to see the design of an ALU using cavity magnetrons!

Click to expand...

Simply pointing out that not everything has been replaced by transistors, yet.

Click to expand...

Indeed. Analog iterative optical machines another route or some chap that likes to build them out of big cogs and shafts

The idea of a TWT or a magnetron based calculating machine is pretty novel.

Click to expand...

I'm trying to get my head around that idea but all I come up with is a baked potato

Click to expand...

We could used potatoes for memory. Raw is zero. Baked is one. Write once, though.

OpticsEngineer said:

I recall my Dad coming home from work one day and saying the EE R&D department had a tube-versus-transistor meeting. Young and old, every electrical engineer and programmer (but no upper level managers). There was a lot of discussion of pros and cons.

Collectively they decided from then on out, everything would be designed with transistors.

All the older engineers saw it as a challenging situation. With tubes you primarily designed with current but with transistors you designed to voltage. So they all worked really hard to understand the theory about electrons and holes and npn versus pnp. With vacuum tubes, you really needed to understand the physics inside to design with them effectively. So they thought transistors would be the same.

A decade later my Dad said he didn't understand why they wasted all that time on electrons and holes. It wasn't helpful in designing circuits at all.

Click to expand...

I agree with your Dad. I took semiconductor theory from John Linvill at Stanford, and thought it was interesting, but never used it. I just used the models for transistor behavior when I designed with them. Truth be told, that's how I designed with tubes, too.