Byte magazine (1984): Benchmarking Unix | Hacker News Reader

Byte magazine (1984): Benchmarking Unix

archive.org

86 pointsrsecora44 comments

Some links inside the article. Byte magazine was plenty of ads, and articles can continue after jumping 300 pages

Start: https://archive.org/details/byte-magazine-1984-08/page/n137/...

Continuation: https://archive.org/details/byte-magazine-1984-08/page/n405/...

Results: https://archive.org/details/byte-magazine-1984-08/page/n411/...

I love looking trough the ads! An Apple II, and a color printer from Apple. Compuserve's CB Simulator. Ads for drives that hold Megbytes of data. So many different peripherals and PC makers. UltraPlan Spreadsheet Calculator. An ad for an Assembler. One for Unix Sys V from AT&T. Modem after modem.

Compuserve's CB Simulator sounds like it must've been awesome!

From the beginning of Byte magazine until about 1984 are really fun to read. The ads were probably annoying at the time of release, but not they are fun and interesting to read. I like the 1983 issue with imaging where they are talking about movie effects for "Tron" and "Revenge of the Jedi".

I wished they would of had more UNIX in the issues, however it was usually unaffordable. But a fun look, at least for me, to the history of computing. And who does not like an ad for a 8MB, $3999 disk drive subsystem.

The ads were not annoying at the time of release. I really enjoyed reading those ads and finding out about products specs and prices, remember that we didn’t have the Web back then and Byte magazine was my main source of information about what was available.

Indeed. Complaining about the ads in Byte would have been like complaining about the ads in the yellow pages.

Byte almost was reverse Playboy: you almost didn’t read it for the articles.

Even more true for Computer Shopper! At its peak it was probably an inch thick every month, and 99% ads. But in pre-Internet days, very useful.

Really! the articles where the main point - just whish I had had the nerve to use the $25 diy modem Steve Garcia for one project instead of paying £300 for an answer and £600 for an answer originate.

>Steve Garcia

Steve Ciarcia

https://en.m.wikipedia.org/wiki/Steve_Ciarcia

His BYTE electronics DIY project articles in his Circuit Cellar column were so well-written that I used to enjoy browsing them even though I'm not an electronics guy, gleaning what I could about the topic. He even did one about a 32-bit computer, when 8 and 16 bits were still the standard! The Definicon motherboard/CPU, I think it was about.

¿Steve Ciarcia? https://en.wikipedia.org/wiki/Steve_Ciarcia

I think it “creative computing” or nibble magazine that had a postcard where you could check off interests and send it back. I did this in middle school and would occasionally get mailings about products (and credit card offers).

Magazine ads where the way to get targeted views.

That was a standard system in the magazine trade back then. It was part of the service to advertisers. They would collect all the queries for each advertiser every month and pass them on.

Those ads were unbelievably lucrative. Each page cost thousands of dollars, and there were hundreds and hundreds of them, plus smaller classifieds. Freelancers were paid fairly generously, but that outgoing was a drop in the ocean compared to the monthly ad income. Byte was posting profits of around $10m/year in the 80s.

Edit to add: many readers saw the ads as informative rather than intrusive and distracting. This was partly because many were creative and fun, but also because they went into far more technical detail than you'd see today, so they were almost a supplement to the official written copy.

"Revenge of the Jedi" - you mean "Return of the Jedi" right? :D Although "Revenge of the Jedi" would be awesome!

The original title was actually Revenge of the Jedi and there are still old posters and ads around for it, eg https://entertainment.time.com/wp-content/uploads/sites/3/20...

A later (earlier) chapter became "Revenge of the Sith."

Seeing the article about the "Transputer" was interesting especially given the announcement of the UK government's investment into quantum computing.

The International Journal of PoC||GTFO [0] includes a lot of old tech ads if you're into that thing. A nice touch alongside the hardcore engineering feats you see in the publication.

[0] https://www.alchemistowl.org/pocorgtfo/

I remember using the CB Simulator when I was a young teen back in the 80's! That was my first introduction to multi-user stuff. It was cutting edge for the time.

Same here, and it really sharpened my typing skill.

CB Simulator was basically IRC before that became a thing. It even had channels. Probably blew a few minds in 1980 though!

Thqnx for that, my first pc is on the next page, the apricot. Paid around 9.000 gulden which is around 4.300 euro for it.

The Byte Benchmark suite can still be downloaded, compiled and run today. It's in pkgsrc, FreeBSD's ports, and can be trivially compiled on common GNU/Linux distros.

The most popular modern descendant of this benchmark suite is here: https://github.com/kdlucas/byte-unixbench

The scripts still run, my ancient (1st or 2nd gen) Raspberry Pi gets 0.54 seconds for the multi .. 6 benchmark

That means the VAX-11/760 (120.000$-160.000$) was 27 times slower for that benchmark. And the PC-XT running PC/IX was 260x slower.

The VAX was also the size of a wardrobe. Interesting to see that it was still leading the pack, more or less, even though it was a 1977 design compared to the smaller and cheaper micros from the 80s - including the Lisa, which clearly was not a fast computer.

There were desktop VAX systems. The MicroVAX II, for example:

http://gunkies.org/wiki/MicroVAX_II

The VAX ISA was implemented in a wide range of technologies, from TTL (the 1970s VAXen) up through true microprocessors:

https://en.wikipedia.org/wiki/MicroVAX_78032

VAX 11/780's were large machines. By the mid 80's you could get a MicroVAX which weren't quite as powerful, but much smaller. In the late 80's, I knew a guy who had one at home.

Several wardrobes :-)

They write it's "interesting to note" the the IBM PC XT 8088 performed better than the PDP-11/23. I suppose the PDP was a much more expensive machine.

The 11/23 was one of the lower-priced PDP-11s, based on the F-11 chipset. I'm pretty sure it would have cost a fair bit more than an XT, though, especially fitted out with the necessary memory and peripherals.

That being said, Heathkit offered earlier LSI-11 machines (which used bit-slice processors) for several years. As a student at a high school with an 11/34, I would have loved to have an H11 or H11A at home, but the Apple II+ that we did have served the family well.

A factory-assembled and tested Heathkit H11A, sans memory, storage, or terminal, was $1895 in 1970s money.[1] Buying the kit would save you $700, which would have been well worth it since the CPU board was a stock DEC KD11-HA [2], already assembled and tested anyway.

[1] https://heathkit.garlanger.com/hardware/systems/H11/ [2] http://gunkies.org/wiki/LSI-11/2

The PC also had no MMU or OS overhead.

The 68K had no MMU or even segment registers, so the 68K machines that can run UNIX had, at the very least, some kind of external memory offset adder. I did not know the Apple Lisa had this- it's interesting because Macs didn't have it and could not run UNIX.

The Motorola 68020 has a MMU coprocessor, and the 68030, 68040 have an on-chip MMU. The Macintosh II, SE/30, Quadra, and Centris series were able to run A/UX [1] since 1988. A/UX was a SystemV with X and Finder.

[1] https://en.wikipedia.org/wiki/A/UX

> The PC also had no MMU or OS overhead.

No MMU, sure, but the whole point of this was to benchmark an OS. It's all about the OS overhead.

In the Docker container on my laptop, the multi.sh 1 2 3 4 5 6 takes 0.031 seconds.

https://github.com/Katzmann1983/performance_test_shell/blob/...

The Byte benchmarks are still very useful today. They’re also old enough that it’s doubtful that compiler makers are going to accelerate them (we know Intel does this), because barely anybody uses them any more.

Laying out code in variable width font like a boss :). I love reading these BYTE scans from archive.org, they are from such a different computing world. Yet UNIX is the same.

> Yet UNIX is the same.

Use V7 Unix in an emulator and say that. Unix didn't have networking, mmap(), or job control shells back then, and that's about the level some of the benchmarked Unix variants were at.

Fair point. My notion was mainly stemming from the looks of the shell scripts in the article.

Heh - I worked on 2 of those systems .... remember these are systems with ~1/2Mb of ram and 1-2 mips of CPU. Most are swapping systems, no paging.

In the 6a. shell benchmark, what's the modern equivalent of the "+" option in sort from the below line?

  od sort.$$ | sort -n + 1 > od.$$

-k POS

From the FreeBSD 2.1 man page [1]

  +POS1 [-POS2]
       Specify  a field within each line to use as a sort
       ing key. The field consists of the portion of  the
       line starting at POS1 and up to (but not including)
       POS2 (or to the end of the  line if  POS2  is  not
       given). The  fields  and  character positions are
       numbered starting with 0.

[1] https://www.freebsd.org/cgi/man.cgi?query=sort&apropos=0&sek...

I think that +1 translates to -k 2. And, at least, FreeBSD's sort still supports +N.

Right beside an ad for an Modula-2 compiler, the irony.

Reply on news.ycombinator.com