Back in 1987, I briefly worked for an organisation that had a Cray. I was never allowed anywhere near it, but one of my friends worked on the VAX that was gatekeeper for the Cray. There was a strict rule that any job submitted for the Cray had to be prototyped on the VAX first - because Cray runtime was too scarce and expensive to waste on jobs that might hit errors.
The ultimate no-no, of course, was to put the Cray into an infinite loop, at £1,000 per minute or whatever its runtime cost. Rumour had it that anyone who did that would be fired on the spot, but it never happened as far as I know.
Is it any different now? If I tell you that I want a $500,000 grant for 10 million CPU-hours on your supercomputer, surely you would like to be persuaded my code has some research value.
One question I have for the up-to-date supercomputer experts on this forum is about the architecture, which will affect the types of calculations I can make most efficiently. For example, imagine that I divide space up into a 2d or 3d or 4d lattice or have other requirements on how data needs to flow from one node to another. I have a big room full of processors, and they can’t all talk to each other at the same time at the same speed. To what extent does the compiler help me out? What sort of preliminary optimization/tuning do I need to make before running my code on the big iron?
My latest laptop purchased in 2020 came with 16GB of main RAM and 6GB of GPU RAM. My habit of having too many Firefox tabs kept filling up the main RAM, so I upgraded it to 32GB. These days it tends to stay just over half full.
It’s only convenient if you still have optical media to use it with. Future generations, heck, young people of today, haven’t, so they won’t miss anything at all.
I last used one maybe a decade ago? I’m so glad I don’t have to deal with that shit anymore. I still have a USB DVD writer in my closet somewhere, but haven’t needed to pull it out for a decade. I remember thinking how crazy it was Apple was selling Airs without optical drives in the mid-00s, and within a few years I was like, wow, I hate optical media too!
I still have a (portable) optical drive for my computers, but only because I sometimes still use to buy music CDs, mostly special editions, that I can’t get streamed and I need the drive to rip them as I have done with all of my CDs. And there are still many data DVDs from back in the day with movie files I got from torrent sites (I don’t do that anymore). Else I wouldn’t need the drive.
ETA: and it’s sometimes convenient to boot from a disc. Bootable USB drives can be flaky.
In many respects, CD’s are the perfect medium for exchanging files. Sure, you can use a thumb drive–but who needs a medium with multiple gigabytes when dozens of pictures will fit on a CD? And you can buy dozens of CD’s for the price of 5 or 6 thumb drives.
Sure, you can upload files to Dropbox or Google Drive–but both of those are flaky in terms of permissions. You have to be really careful with how you set permissions, or the intended audience will not be able to even see the files, much less download or edit them. CD’s avoid that issue altogether.
I deal in pictures, and USB and online exchanging have been the easiest method of exchange. I switched to USB delivery of photos c. 2012 and all-download in 2016. There is an option in my contract for USB delivery, and not a single person has opted for it out of maybe 200-ish. It’s never been been flaky or problematic for me. I could not say the same of DVD delivery, which sometimes had bad burns or got scratched on the customers end of what not.
It’s also a lot faster and easier to stick some images on a USB stick for physical delivery than burning an optical disk.
It was possible to hook up an external CD drive, but I never did. Didn’t miss the drive back then, don’t miss it now. One of my favorite machines ever. <3 lbs and ran all the stuff I needed it to at the time.
I actually have one of those today. I acquired it in the early 2000s, when I collected “vintage” computers. From what I understand you could buy an “expansion box” for it, which allowed you to add peripherals to it, including floppy drives.
I took an embedded systems class in college, circa 2001. Our class project was to write the software to control a toy truck, which had been modified by the instructor to be controlled by a little Atmel AVR demo board strapped to the back. While I was working on that project it hit me that little microcontroler powering our truck had about the same computing power as an Apple II.
Around that same time I had an internship at IBM. One of my first tasks was testing different memory configurations on a server they were developing. I was working with 1GB DIMMs, and thinking that was an absolutely ridiculous amount of memory – at the time 1 GB DIMMs were rather expensive and something one would only find in a larger server. I think that server I was working on supported maybe 32 GB of RAM if you populated every slot. Nowadays you could probably get a laptop with that much RAM.
Beat you to it. For the record, it wasn’t that it didn’t have a few options, but that by the time my father bought it, after it ended it’s production run, you couldn’t get any such devices for it, since they fundamentally weren’t in the shops anymore.
I can’t seriously yell at my dad, even at those end-of-life sales, that would have been around 100 bucks US, or around 275 now according to an online calculator. Not universe ending cash, but quite a lot to spend on an 11 year old especially when to non-techies of the time, it was still a ‘you have a what?’ thing.
I actually forgot about that expansion port on the right. I was thinking of the Peripheral Expansion Box, shown here:
But yes, if you couldn’t buy one anymore the fact that they existed is moot. Ironically it’s probably easier to find one today, with eBay and similar sites.
What’s also neat is how the power of embedded systems has grown over time. A couple decades ago, I was using little PIC microcontrollers, some with as little as 16 bytes of RAM and a couple kilobytes of program memory.
Today, I’m using ESP32 modules, which cost about $5 on a carrier board with PSU, etc. built in. They have 4 MB program memory, ~0.5 MB RAM, and WiFi and Bluetooth built in. Oh, and they run at a much higher clock rate, are 32-bit, etc. I can program them over the network and they can interface directly with my phone via Bluetooth. Not to mention the usual set of microcontroller features, like multiple UARTS, ADCs, etc. My little hobby projects are much more powerful because of this.
I had exactly the same thought myself. I even brought up a similar point in another thread, about how easy the internet makes collecting missing pieces / rare hobby supplies, compared to back in the day trying to mail order anything that wasn’t in a local shop or chain.
@pulykamell - sorry, I meant that YOU beat @WildaBeast to the details of the peripherals, not that you hadn’t read my post. I just wanted to make sure I linked it to show the glory and madness that the full expansion entailed!
