How much science could you do with a 40yr old personal computer? A homage to the IBM PC at its 40iest birthday

The broad availability of affordable computers, or microcomputers, has fundamentally changed many aspects of society, work life in particular. Science is no exception, and indeed, has been altered in several ways. Not the least, because the way how knowledge is obtained, distributed, generated, discussed - skills that make a successful scientist - have been altered fundamentally. In this opinion piece, I reflect on the history that created the to date most successful computing architecture, as I believe some of the principles apply to modern science as well. Further, I do the self-test. I take a computer that builds on the original IBM PC architecture, an Amstrad PC1640 of 1987, and ask, how much useful science one could still do with it. I come to a surprise conclusion: you can do more with an IBM PC compatible today, as you could have done at the time.

 

Why does the success of the IBM PC matter until today?

The release of the IBM Personal Computer (PC), some 40 years ago now, in late 1981, is considered one of the key milestones in the history of modern computing.  The release of this architecture is significant for at least two reasons. First, the fact that IBM produced a personal computer, meant that in the early 1980ies the leading corporates started to take microcomputers seriously. Before that, it was mostly garage companies driven by enthusiasts (some of them like Apple and Microsoft are now famous), that did drive the key developments. Second, the IBM PC architecture survived until the present day, while other architectures, some perhaps more powerful, had a much shorter lifetime.  Indeed, the heritage of the IBM PC still dominates office and home computing.

 

IBM PC 5150, released in 1981. Image CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=51833

It is worth reflecting why the PC architecture survived, and why others, didn't. A lot of the history of the architecture is connected to Gary Kildall, in the mid-1970ies a young professor of computer science. Kildall realizes, that the 8080-microprocessor made by Intel can do much more than driving washing machines. He writes a programming language for it. Then, having a programming language at hand, he needs an operating system. He writes one too. It becomes known under the name CP/M. And Kildall incorporates a transformative idea into CP/M. He writes the OS in a multi-layered way, so that it can be ported to other architectures. Sounds trivial today, but back in the days, software was typically written for just one specific type of computer. The big investments were made into the hardware, software often came second.

 


Both aspects make CP/M critical.  First, CP/M showed that relatively cheap microprocessors like the 8080 can be used to create a useful computer (History says that Kildall offered CP/M to Intel, but in the late 1970ies, Intel did not yet realize the value).  Second, we now know that software is key that led users to stay with a computer architecture, and that drives innovation You're happy to buy a new PC, laptop, or mobile phone every few years, and you may love the better hardware specs. But for most users counts much more that you still can use access the files, images, text documents. And most people want to be able to work with the new computer immediately - and not to have to take, let’s say, a 3-week long training before being able to do that. So, to make a computer successfully over time, you need to be able to port its operating system and software to new hardware, without major changes in the user experience. The possibility to retain software turned out to be very important in the 1980ies and the 1990ies, were the hardware developed very fast.

 

Despite its transformative importance, CP/M played no major role in the long term. Ironically, it was IBM itself that spelled the death of CP/M. They offered their customers a choice to buy a 7-times cheaper operating system instead of CP/M. This Disk Operating System (DOS, the name goes back to nomenclature used by Kildall) was produced by an at the time garage company, called Microsoft. What we know from today’s perspective is that Microsoft actually didn't have a DOS product to sell at the time they signed a deal with IBM. They rushed to license a system, that that in part contained loosely re-written CP/M source code. Indeed, despite CP/M died, some of that code, and lots of its heritage, still sits insight what is now called Microsoft Windows.  

 

The next game changer: IBM PCs could be cloned. An early taste of open architectures and open source

 

A third factor at play was the for the time unusually open architecture of the IBM PC. Once IBM decided to enter the rapidly developing microcomputer market, they had to gain speed over their typical developmental cycles. The development team therefore decided to build the personal computer from off-the-shelve components, rather than using IBMs own hardware. They did so - with the important exception of the Basic input output system (BIOS), the piece of software that after Kildall's revolutionary ides, communicates between hardware and the other modular layers of the OS. That meant, that everybody could re-build an IBM hardware compatible PC - if he or she is able to reverse engineer the BIOS, that is. This situation turned out to become an unintended game changer for the computer industry. At the time, the computer industry was very protective, and typically IBM or anyone else would have sent army of lawyers after any company that would have copied a computer architecture. However, IBM could not do much against the hardware copying, so they took the other approach, and started to promote their PC architecture as a `standard' in personal computing, and licensed the architecture quite openly. And Microsoft had retained the right to sell their OS to other computers as well.

 

The success of the IBM PC architecture was driven by its clones, 'IBM compatibles'. One of the most successful clones was the the Amstrad PC1512 / PC1640, equipped with an 8 Mhz Intel 8086 and 512 resp 640kb of RAM. Sold after 1985, the Amstrad PCs had an appealing design, were silent, and much cheaper than then IBM PC.


It is reasonable to assume that this largely open architecture was the real key behind the success of the IBM PC architecture, over the longer time. Because it was lucrative, multiple companies soon started to sell PC compatibles. This included Amstrad, a UK company that built the machine this text is written on. Indeed, for a short time in the mid 1980ies, the Amstrad PC clones dominated over IBMs own PCs in the European home computer market. Why did cloning drive innovation? IBM compatibles had to be cheaper than IBM (which wasn't that difficult), but soon, they had to be more innovative. And trump each other in marketing, which amplified the visibility of the PC architecture. The PC became the de facto standard in office computing, and later, at home.

 

Eventually, the armada of IBM PC compatibles crated a self-driven life that became independent over IBMs own decisions. Once IBM decided to change path and introduced the (no-longer existing) more protective PS/2, architecture, the industry kept developing the PC architecture. People kept buying them, software developers kept writing for IBM PC descendants, Intel, AMD and others kept developing ever faster PC CPUs maintaining the instruction sets of the 8088 CPU. IBM and Microsoft tried a while to shift users to more modern OS types, Xenix or OS/2, for instance. Without success. The PC and the DOS heritage survived is fully alive till the present day.

 

How much science can you do with a mid-1980ies IBM PC compatible at the verge of 2022?

 

In 2021, the PC descendants are a planet faster, have a universe of more memory available. For instance, in 1987, a 8 MhZ 8086 CPU, 640kb of memory and a 20 MB hard disk was considered a well-equipped computer. Further, modern PCs have 40 years of global software development on their shoulder. Yet they are still PCs, as I learned firsthand.

 

My journey starts with the aforementioned, arguably most successful IBM PC clone, the Amstrad PC. The Amstrad model PC 1640 was released much later than the IBM PC Extended technology (XT), in 1987, at a time when IBM had already moved to faster but more expensive Intel 80286 and 80386 CPUs (Advanced Technology - PC-AT). Yet, with an 8086 CPU and a well-balanced set of peripherals, the Amstrad PC was a huge economic success, and together with its predecessor, the PC1512, sold better than IBM PCs in Europe. Mainly, because the Amstrad PCs were way cheaper than the IBM PCs, looked nicer, and didn't need a noisy fan. Indeed, I compared from today’s perspective the Amstrad XT with a PC-ATs. True the hardware had developed a lot from the original PC to the AT, but effectively, from a user perspective, you could do the exact same things.

 

A had an Amstrad PC as a child, and once I could recover it from my parent's attic, a question did rise: Could I make it work again and do something useful with it? In doing so, I learned that you can do much more with this old PC in 2021, as you could have done in 1987. The main reason is that, because modern PCs still based on the very same principles, you can make the old machine talk to the modern world, exchange data, indeed network them, and do things that you could not possible imagine in the early 80ies.

In part, this is possible because there are true enthusiasts out there, that have backported some major developments, like TCP/IP to MS-DOS, so you can use your IBM PC compatible. For instance, I was able to hook up my 1987 XT over a software called ETHERDFS, which gives me a network drive that I can not only access from any modern computer in my network, but indeed synchronize to cloud storage.  Second, as aforementioned, a computer is only as useful as the software you have for it. Back the 1980ies, software was hugely expensive, and difficult to obtain, so you would have only a minimal amount of software running on your PC. Nowadays however, a lot of the 1980ies programs have long outlived their commercial value and have been made freely available. You can easily fill your 1980ies PC with software, that would have cost you a true fortune at the time of its release.

The old PCs were of high quality, my Amstrad powered up without any issues. The most difficult part was to get started with the software, and indeed, I faced the same obstacle as the personal computing pioneers in the early 1980ies: How to exchange data between different computers if there are no common disks or file systems?  The Amstrad came with two, 1980ies typical, 360kbyte 5 1/4" floppy disk drives (for those that remember them, 5 1/4 inch diskettes really are 'floppy', that's why the name - in difference to a 'hard' drive; and as many computers had two, they were called A and B, while the hard disk that came later, got the letter C); but you can't connect any 5 1/4 inch floppy drive to any modern computer. Eventually, I found that a company called Gotek sells an USB floppy emulator. I managed to install one, however in the absence of a respective adapter, it needed to heat up my soldering iron. You can't imagine the eureka moment, I managed to load a file from an USB drive on an the 1987 computer!

 

The next steps were then simpler, although also they took some time. The most important task for me was to establish network connection, because a majority of modern computer work involves the internet connectivity.

Was it possible to make the Amstrad talk to the modern world, and access the internet, that was invented much later?  Luckily, Intel and co had introduced the Ethernet standard already in the early 80ies, and backward compatibility has been kept until the present day. I found an 8-bit Ethernet card (a Novell NE1000) at ebay, as I did an RJ45 to Coaxial (2base10) converter. Here, the software was key. An enthusiast, Micheal Brutman, has written a TCP/IP port for MS-DOS, that did work well on the old Amstrad.

 

Some other changes I made on the system were more out of curiosity. I replaced the Intel 8086, released in 1978, with a NEC-V30, an early 1980ies development that gave some more speed and instructions to the old machine, added an arithmetic coprocessor (an Intel 8087, that accelerates floating comma operations significantly, but unfortunately, only for software that supported the FPU), and a 1990ies VGA graphic card, so that I could connect a modern Monitor. Over time, I also managed to install a soundcard (a mid-1990ies Soundblaster that still works in the 8-bit bus), and to connect the parallel output to a Rasperri PI, that converts the printout, so that I can now print on a modern network printer.

 

Can you work scientifically with a 40 year old computer?

 

From a biochemist’s perspective, certainly, you cannot do any modern machine learning tasks with the old machine, nor any useful image processing. A memory of 640 Kbyte and an 8 MhZ CPU frequency with an 8 bit external bus are simply not sufficient. However, some of the tasks that may take a large fraction of your daily work routine are surprisingly efficient. Via telnet, I accessed Alpine, and with it, the old PC is highly efficient for reading and replying emails. I now do sometimes my late evening correspondence with the old PC. Writing letters and documents with Word Perfect is highly efficient as well, and the white-on blue text screes is not tiring to eyes. No distractions by reading news while writing!. Setting up a simple database with DBASE 3 works well too, and I have seen colleagues sharing Excel tables that were far less practicable than a DBASE database. And, while complex machine learning tasks do not work on the old machine, simpler forms of data processing work quite well. DOS versions of Matlab run on the old PC, the Excel predecessor Lotus 1-2-3, facilitates quite sophisticated table calculations; and there is a plethora of plotting software available, even to generate regressions. A surprising lesson I learned how simple it is to transfer the files from the old machine to my modern computer. The key file formats, including those from Word Perfect and 1-2-3, are still read by modern software, so it suffices to save them on the network drive and open them on my new computer in an instant. Educational aspects are also great on the old PC, learning the basics of programming, for instance. And finally, entertainment. Certainly, complex graphic heavy games don't work, but there are hundreds of nice DOS games available for download, quite a few of them are still played on mobile phones. And there is a form of social media. Some enthusiasts have never switched off the pre-internet Bulletin board systems (BBS) systems. You can connect to them, now via internet, download software, leave comments, read news, and have a nice chat with fellow enthusiasts. On some you can even tweet!

 

Reflections for science

I believe, there are some reflections possible and some things we can learn from this recent piece of history for the way we do science now. First, the PC architecture has not survived because it was the technically most powerful, but effectively, because of its design was, in contrast to competitors, openly available. In this respect, science is moving in the right direction, it is more and more required to make scientific results, data, and publications, openly available. The scientific and pharmacological industry may be more secretive but may eventually learn from these experiences. And there are key discussions ongoing if our way of protecting scientific discoveries i.e. via patents has stated hindering rather than protecting innovation. Often, being overprotective of IP seems may spell the end of your innovation.

The second lesson to learn is that the computer industry moved on the produce and sell much more powerful hardware, long before the possibilities of the existing PCs were exhausted. In the bioscience, we tend to do effectively the same, and keep recording new data, before we have exhaustively analyzed what is already to our disposal. True, new developments are a key source of innovation, but at times of shrinking resources, we could also learn to better make use of what we already have.  

 

 

 

 

 

 

Markus Ralser is a Professor of Biochemistry at the Charité Universitätsmedizin Berlin. The first version of this article has been written using WordPerfect 5.1, released in 1989, on an Amstrad PC1640, released in 1987.

Comments