My first PC was made by Timex and had 2K of RAM. Oddly enough the processor they used is still in use today. The Z-80 microprocessor was used in everything from handhelds to consoles. There is not much point to programming games on a Z-80 but I still recall some of the code. Before Visual C++ came along I was deeply entrenched in the Borland Turbo Pascal vs Turbo C wars. Many programmers got bloody noses over that battle. I used to program in Turbo Pascal on my Tandy 1000 HX. Those type of systems are long gone with the TI-99/4A and Commodore 64.

I doubt you can find many vintage 286, 386, or 486 mobos out there these days. I recall telling my wife how powerful my 486 with 256 KB of RAM and a 512 MB HD was back then. I even had a SVGA monitor. When Windows XP came along, it changed everything. I shut down the dial up BBS and started a new life of just not caring what the heck was going on inside my machine.

btw my rant is better. 🤣

[typo edited: 256 KB not 256 MB]
Excuse me I am an old fart with glasses! 😁

I assume your comments about programmers getting "bloody noses" meant that woke up one day and everything they new was obsolete.I don't work in the computer field, but I suspect it's tough..... and probably a lot tougher than it used to be. It changes so fast.... and can change on a dime.

Borland made both Turbo C and Turbo Pascal. There was a great deal of "fanboys" on both sides of a heated debate of which was the better language and compiler. As I saw it, both had pros and cons. C was and is the more accepted in business environments, if you were going to get a job programming. Pascal was educational, easy to read, learn, and understand. C had ability to handle mixed data-types. Pascal compiled faster and compressed better. In the end, neither side proved their language of choice to be the better one. Since both were designed to be used in the DOS 6.x OS they discontinued both.

I enjoyed programming back in the days of DOS. The problem with programming in today's environment, you are forced to learn the API of Dynamic Linked Libraries (DLL) calls. In the old days of DOS we normally just opened a file with a line like;
FileOpen ("C:\mytext.txt",readonly)
or
AssignFile(MyFile, 'C:\mytext.txt');

Today you have to ask Windows to open it for you.
Like this;
HFILE WINAPI OpenFile(
_In_ LPCSTR lpFileName,
_Out_ LPOFSTRUCT lpReOpenBuff,
_In_ UINT uStyle
);

I guess it is the difference between tinkering on a minibike or investing in a motorcycle.

ncmark wrote:

The programming windows 3.1 books I salvaged need to go on the trash. Yes, they show you all the nitty-gritty details, how it REALLY worked, how to do it the hard way before visual basic and visual C++ came along. Problem is, no one programs that way anymore. I suspect even visual C++ is outdated.

You only need one.... "Programming Windows" by Charles Petzold. The 5th Edition covers Win32, but the 3rd Edition covers Windows 3.1x and Win16. Plenty of applications are still written in good old C (the procedural way). MFC and VCL (Borland's OO frontend for Win32) are quite bloated for most tasks. Most of the Windows ports of popular emulators have their interface written in straight C doing Win32 API calls. If one does OOP on WIndows and isn't too worried about portability, just use a .NET language and Win Forms.

Effrafax of Wug wrote:

I enjoyed programming back in the days of DOS. The problem with programming in today's environment, you are forced to learn the […]
Show full quote

I enjoyed programming back in the days of DOS. The problem with programming in today's environment, you are forced to learn the API of Dynamic Linked Libraries (DLL) calls. In the old days of DOS we normally just opened a file with a line like;
FileOpen ("C:\mytext.txt",readonly)
or
AssignFile(MyFile, 'C:\mytext.txt');

Today you have to ask Windows to open it for you.
Like this;
HFILE WINAPI OpenFile(
_In_ LPCSTR lpFileName,
_Out_ LPOFSTRUCT lpReOpenBuff,
_In_ UINT uStyle
);

The standard C method of working with files is still used on Windows, mostly because its portable. Turbo Pascal was replaced with Delphi.

I don't play around with Super 7 systems anymore just because it's less aggravating to build a 440BX system and the performance is better. Super 7 only did as well as it did because it was MUCH cheaper to build a system than Slot 1. Now, though, Slot 1 is the better value because it's less buggy and as mentioned above, PII-PIII offers better performance than K6-2 and 3.

It's not all about performance. Don't forget that a decent number of games simply won't run correctly on a PIII. With SS7 you can use Pentiums and cover that speed in an ATX form factor, so that platform certainly has its uses.

sliderider wrote:

PII-PIII offers better performance than K6-2 and 3.

How long is a piece of string?

Yeah, my Slot 1 PC is pretty much an Intel-based recreation of my last Super Socket 7 PC that died due to bad motherboard design.

carlostex wrote:

sliderider wrote:

PII-PIII offers better performance than K6-2 and 3.

How long is a piece of string?

I don't know but when a K6-2 or 3 overclocked to 600mhz can't even keep up with a 450mhz P-II in games because of the lousy FPU, that pretty much says all that needs to be said. 440BX plays all the period games better and you really have to screw something up badly to make it crash while Super 7 boards would regularly bring up BSOD's with error codes that few people outside Microsoft even knew existed.

sliderider wrote:

I don't know but when a K6-2 or 3 overclocked to 600mhz can't even keep up with a 450mhz P-II in games because of the lousy FPU, that pretty much says all that needs to be said. 440BX plays all the period games better and you really have to screw something up badly to make it crash while Super 7 boards would regularly bring up BSOD's with error codes that few people outside Microsoft even knew existed.

The K6 FPU is not lousy. I've never had BSOD's with SS7 then and i'm not having them now. Yes, 440BX was a very good platform indeed. But your premise is wrong.
The K6 FPU could actually complete FPU ops faster than PII because the latter had a much higher latency in typical FPU instructions like FADD, FSUB and FMUL. The PII FPU however because it was fully pipelined it could complete 21 instructions at once against 11 of the K6. The tradeoff is that the PII will have a big penalty in a pipeline stall event. Not only this, but the branch prediction tables were more accurate in the K6.
Now here's the thing that people mostly have ignored. Your micro architecture efficiency means squat if code is not optimized for it. Now add adly written compilers, or even code crippled compilers and you are screwed. At the point of the advent of SIMD, a K6 was aroun 60% of the PII FPU performance, because code was always optimized for the Pentium. If you reverse the roles and have something heavily optimized for 3DNow!, then the PII will be eating dust. It was pretty much the moot point for Intel SSE later. And we know how the industry follows Intel. It's just how it was and it is still like that.

Paul Hsieh stated at that time:

"The state of floating point has changed so drastically recently, that its hard to make a definitive comment on this without a plethora of caveats. Facts: (1) the pure x87 floating point unit in the K6 does not compare favorably with that of the P-II, (2) this does not tend to always reflect in real life software which can be made from bad compilers, (3) the future of floating point clearly lies with SIMD, where AMD has clearly established a leadership role. (4) Intel's advantage was primarily in software that was hand optimized by assembly coders -- but that has clearly reversed roles since the introduction of the K6-2. "

If the OP needs a faster system why not build an Athlon 64 system or even Core 2 LGA 775? If he needs legacy ISA i believe there are industrial LGA775 motherboards out there that will serve his purpose. If he needs the flexibility of speed, that is, a balance of fast and slow the K6-III and SS7 is one of the best platforms, if not the best. The Athlon Thunderbird and the Athlon XP are also faster than PIII. I have a socket A motherboard that sports an ISA slot.

This sounds to me as if you complain about your own mistakes.

First of all if you talk about 'the usefulness of retro computing' you should define your usage scenario. Also the performance discussion is completely off, since you can look up benchmark scores beforehand today and decide what to build depending on your usage scenario.
If a K6-2 is too slow then build a P3 system, for the example that your usage scenario is gaming.

If the usage scenario is historical archiving and comparing the system to different other systems then performance is a unique property of the build system. It is as fast it is. Now one can start to explore historical usage scenarios, like desktop applications, how code quality and optimization reflects performance on different CPUs in comparison, gaming, etc..
There is no room to complain here about performance, since the only expectations is that you get a working system together.

Regarding the K6 series, it was exceptionally successful and gave AMD the financial resource to develop the Athlon. This contradicts the abysmal performance scenario that is often described here.

The programming documentation is of course obsolete. But as always it is difficult to decide what to archive and what to forget. Historically it is quite interesting to analyze how software was programmed at a certain time. It is much easier however to look into source code. A complete picture however has also the include the compiler and hardware where the software eventually was executed. Maybe somewhen in the future we have the tools to analyze old binary code the same way.

Effrafax of Wug wrote:

My first PC was made by Timex and had 2K of RAM. Oddly enough the processor they used is still in use today. The Z-80 microprocessor was used in everything from handhelds to consoles. There is not much point to programming games on a Z-80 but I still recall some of the code.

About programming games: Do you mean today or at the time? A lot of games have been programmed for Z-80 home computers and consoles.

I recall telling my wife how powerful my 486 with 256 MB of RAM and a 512 MB HD was back then.

If think you are confusing some of the details...

btw my rant is better. 🤣

I'm not sure if you are just joking or trolling.

sliderider wrote:

I don't know but when a K6-2 or 3 overclocked to 600mhz can't even keep up with a 450mhz P-II in games because of the lousy FPU, that pretty much says all that needs to be said. 440BX plays all the period games better and you really have to screw something up badly to make it crash while Super 7 boards would regularly bring up BSOD's with error codes that few people outside Microsoft even knew existed.

According to the 686 benchmark comparison for Quake II, the K6-3-600 falls between a PII-400 and a PIII-500, so its gaming performance is at least a PIII-450, and that was without the 3DNow! Quake extension enabled. With 3DNow! enabled, expect the AMD K6-III to perform as well as a PIII-600 Katmai in Quake II.

The fastest K6-3+ I could get repeatably stable was at 550 MHz, however, and as previously noted, a SS7 doesn't even come close to the stability of a 440BX. The most stable socket 7 I've come across is a 430TX running the K6-III at 450 MHz, and with 3DNow! extensions, that's about equivalent to a PII-450 in Quake II.

ncmark wrote:

After building 4-5 retro systems, I am wishing I hadn't wasted my time on K6-2. I have two board I got off Ebay sitting in the closet. Obsolete is obsolete. The pentium 3 systems I built are still useful but K6-2... forget it. I wish I had spent more time building more modern systems - like another Athlon or maybe even a core2. I still don't have windows XP, and still can't use external drives except on linux

I always say: decide what software you want to run, and then worry about the hardware you need to run it. The hardware on its own provides limited satisfaction.

The programming windows 3.1 books I salvaged need to go on the trash. Yes, they show you all the nitty-gritty details, how it REALLY worked, how to do it the hard way before visual basic and visual C++ came along. Problem is, no one programs that way anymore. I suspect even visual C++ is outdated.

Indeed, if you're going to do any kind of programming, there's not much point in limiting yourself to techniques primarily designed for systems from 20 years ago.

After trying to burn some pictures into CD-ROMs, I have come to the conclusion THAT it's not all it's cracked up to be. Okay it was MILES ahead of floppies and tape-backup, but DVD is just plain better - holds more and so fewer disks, and faster too

Why would you possibly think otherwise..?

d1stortion wrote:

It's not all about performance. Don't forget that a decent number of games simply won't run correctly on a PIII. With SS7 you can use Pentiums and cover that speed in an ATX form factor, so that platform certainly has its uses.

Indeed, rather than messing around with cache, you can just take out the CPU entirely and put in something slower. But again, the question comes down to whether one will have need of such capability often enough to make it worth the tradeoffs.

idspispopd wrote:

About programming games: Do you mean today or at the time? A lot of games have been programmed for Z-80 home computers and consoles.

IF you research the chip it was used in the Timex Sinclair 1000 but also popped up in the Sega Genesis as a co-processor and the Game Boy contained the Z80 instruction set. My point was and is; What may seem obsolete sometimes is never obsolete.

idspispopd wrote:

If think you are confusing some of the details...

Interesting, confusing the details of a conversion between my wife and I, How? Were you present during our conversation? Back in my time, when 486 was top model on the market, my company had a one million dollar budget to replace all 286 and 386 models within the company. It is only with future hindsight, knowing that if the company held out for 2 more years using the older models, a much better model (Pentiums) were coming out with the Windows 95 OS. At that time, I suggested we install Windows NT to the new 486 models. The company disagreed and instead purchased Windows 3.11. But many of the corporations I have worked with have never shown much reason either.

idspispopd wrote:

I'm not sure if you are just joking or trolling.

Maybe I should pose the same question to you?

K6's were perfectly fine with games that didn't use the FP unit (basically anything not a 3d shooter).

Unknown_K wrote:

K6's were perfectly fine with games that didn't use the FP unit (basically anything not a 3d shooter).

That probably essentially means games that weren't particularly demanding on the CPU anyway. It's difficult to find benchmarks of how long a turn takes to calculate in a turn based game or some such.

3D games also tend to also be sensitive to memory bandwidth and latency and the Socket 7 platform was certainly not an asset in that respect compared to Slot 1 / Socket 370.

Unknown_K wrote:

K6's were perfectly fine with games that didn't use the FP unit (basically anything not a 3d shooter).

The cheapo K6-2 350Mhz machine with Intel i740 machine I built back in 1999 for CD burning did a respectable job running Quake 3 at the time. The machine wasn't designed for game playing, but could be pressed into service if I really wanted it to.

feipoa wrote:

The fastest K6-3+ I could get repeatably stable was at 550 MHz, however, and as previously noted, a SS7 doesn't even come close to the stability of a 440BX.

Yeah, VIA chipsets were quirky as usual (4-in-1 is the magic word), while SiS seemed to make nothing but train wrecks. It didn't help that they were usually paired with PC Chips boards! The ALi Aladdin V seemed to be pretty stable, at least it was on my Asus P5As. I still have these machines assembled and sitting on a shelf. They still work just fine too.

For general use, the K6-2/3 was perfectly fine and met most folks needs (word processing, web browsing pre-Flash Player, music, etc.). It was a bargain next to Intel's offerings and we sold quite a few machines with them at the computer store I worked at. I also had the pleasure of supporting them when things went wrong as a ton of our customers would buy/build what I called a "computer show specal" (sic, the signs at the booths were classic). Usually it was a PC Chips/Amptron SS7 board in a blood drawing sharp edged case where nothing lined up right, a dicey power supply, the fan barely hanging onto the CPU+HS, and integrated sound and video. Occasionally they would use a PCI (eck) or AGP card with some cheapo video chipset (SiS or that i740) made by Jaton.

Effrafax of Wug wrote:

Interesting, confusing the details of a conversion between my wife and I, How? Were you present during our conversation? Back in my time, when 486 was top model on the market, my company had a one million dollar budget to replace all 286 and 386 models within the company. It is only with future hindsight, knowing that if the company held out for 2 more years using the older models, a much better model (Pentiums) were coming out with the Windows 95 OS. At that time, I suggested we install Windows NT to the new 486 models. The company disagreed and instead purchased Windows 3.11. But many of the corporations I have worked with have never shown much reason either.

With a computer at that much disk space, you're probably confusing the fake "256KB CACHE MEMORY" message on POST that some of those cheaper 486 motherboards had. 486 motherboards usually can't hold more than 64mb of actual RAM, especially not in the timeperiod where there was 512mb hard drives and 32MB RAM was considered the enthusiast high-end money-burning luxury. There's also a cache-disabling issue if you go beyond 32MB IIRC....

Age does crazy things.

NJRoadfan wrote:

Unknown_K wrote:

K6's were perfectly fine with games that didn't use the FP unit (basically anything not a 3d shooter).

The cheapo K6-2 350Mhz machine with Intel i740 machine I built back in 1999 for CD burning did a respectable job running Quake 3 at the time. The machine wasn't designed for game playing, but could be pressed into service if I really wanted it to.

K6-2's are fine....... if you use 3dfx cards. Their drivers were more 3dNow! optimized than the other vendors - which kind of made the leap to Geforce2 a bit disappointing performance-wise (in 2000 before the refined detonators the following year, anyway). Not to mention there's those special 3DNow 3dfx MiniGLs...

However K6's MMX support was poor so games like Serious Sam and Battlezone II that make use of them (for transformation processing than anything else) perform far worse than the Pentium II counterparts. Not even 500MHz can cope.