K6-2 450 MHZ AMD: Compatibility problems???

Santilli

Santilli

0
Joined
Jan 27, 2002
Messages
5,451
Hi
I'm trying to build a cheap, old, computer out of spare parts.
Kid around here has the amd K6 Tannin used to love, 450 MHZ version, failed dot.com complete package.

Have to find out what mobo, if he knows. I'll be using a
TNT video card, with DVD hardware decoder from my dell, a couple Quantum Lm's I had laying around, 512 MB of ram, and standard failed dot.com cd, and floppy.

Purpose is either a Linux box, or a cheap, working, 2000 box.

Any comments from the resident experts would be great.

gs
 
It all depends on the board, Greg. The CPUs themselves are just about bulletproof. If the board is VIA-based, you should be pretty right. The only real worry with the VIA chipset Super 7 boards is power to the video card. A TNT should be fine. A Gforce 1 or a pre Ti Gforce II can be problematic, but the old TNT-2s were plug in, work in 95% of cases.

If, on the other hand, it's an Ali chipset board, such as the
Gigabyte 5AA or the ASUS P5A-B then, from my experience you have (as we say here in Australia) two chances.

Time seems to think he has a secret Voodoo method (the black magic, I mean, not the video card) to make TNTs work with Ali chipsets. He swears it works.


-----------------------------------------

* "Two chances": traditional Australian expression meaning, more or less, "not a chance in Hell". It derives from the still common expression "Buckley's chance". I'm too lazy to look up William Buckley on Google to refresh my memory, but as I recall he was a convict who escaped into the surrounding countryside, which was not thought fit to support human life, but against all expectations, he not only survived but actually prospered. He was found ten or twenty years later, living happily with the Aboriginal family he had married into and, despite being given a free pardon, was reluctant to return to the white man's world. Unless I've totally ballsed up the story, of course. No doubt Time will correct me if I'm wrong.

Anyway, although it doesn't make much sense, "Buckley's chance" came to mean "no chance" and is universally understood as such. Often it is abbreviated. For example:

"Hey, mate, lend me $50 till payday?"
"You got Buckley's."

So what does this have to do with two chances? Well, it just so happens that there is a long-established and rather posh department store in Melbourne, which had branches in the other states too, I think. Sort of an equivalent to Harrod's or Macys. And the store is known by the name of its founders: Mr Buckley (no relation), and Mr Nunn - Buckley and Nunn.

Thus we get (at last) to "two chances": Buckley's and Nunn.
 
That's funny. I was did a Google search about a week ago for something computer related. One of the hits was for a William "Bloody" Buckley and the synopsis meantioned Oz. Since it was Oz related but not related to the question at hand, I didn't investigate the story any further but filed it away in the noggin.

Would this be the same person?

PS How is 2K going to work on a k62? I am about to come into posession of one myself/
 
I ran a K6-3 450 Mhz (overclocked to 560 Mhz at times) under both Win2K and WinXP. It was a great little processor and the system always seemed fast and smooth (with only 256MB SDRAM) . I only wish the M/B I was using (Epox EP-MVP3C2) had 1-2MB cache rather than a paltry 512KB. In all other respects though, that VIA MVP3 motherboard was just awsome. None of the PCI/ISA slots were shared, AT/ATX power, ATA66 support, and solid quality.

You should have no problems as long as you have a relatively new board/BIOS.

Clocker
 
Yup. Win 2000 should be no problem. I've certainly run NT 4.0 on a number of FIC VA-503+ boards (same chipset as your Epox, Clocker) without the slightest problem. My burner box at the office is a K6-III/450+ on a 503.

503s can be difficult with Win ME though. On the other hand, not working with Windows ME ain't a feature in my book, it's a bug!

mnapostles.jpg


Clippy: yes, that sounds like our man. Damnit, I will refresh my memory. Hmm....

Born 1780, fought against Napoleon, transported to Australia for receiving a stolen bolt of cloth. Sent to the new penal colony at Sorrento (later a seaside resort, now almost a suburb of Melbourne). He escaped with 18 others just before the colony was abandoned because it was considered impossible to hack out a living in that bleak Victorian landscape, and they tried their luck in Hobart instead - now the capital of Tasmania. (Ha! These days a small building lot in Sorrento would sell for maybe half a million dollars! A very posh place. I sure couldn't afford to live there.)

They planned to walk 500 miles to Sydney but ended up going in the opposite direction, along the rugged but stunningly beautiful Victorian coastline (picture above). 13 gave themselves up because it was impossible to live off the land. (Again, how perceptions change! Victoria is the most fertile state in Australia. Ever eaten Australian beef, cheese, butter? Ever worn Australian wool? Ever drunk Australian wine? Same part of the country.)
The other six set off westwards. Five dissapeared and presumably died. Buckley walked 300 miles or so and made friends with the locals, settled down and married, lived there for 32 years.

See: http://www.greatoceanrd.org.au/highlights/legends/wbuckley.htm
 
The two best motherboards ever produced for socket-7 processors were the EPoX EP-MVP3G2 and Tyan Trinity S1598. The Tyan was a bit quicker, but had less features. Both were also available with a 2MB cache version (EP-MVP3G5 and Trinity S1598C2), but the 2MB cache versions (particularly the one from EPoX) are very hard to find.
 
FIC VA-503+. Marvelous board.

I had serious compatibility problems with the Trinity 1598 and normally trouble-free AWE64. Given that was the only board I ever had a problem with the AWE64, I think it's probably Tyan's fault on that one.
 
What Intel chip was the K6-2 supposed to compete with? How well?

Is the cache chip upgradable on these boards like on pentium boards or are these really just pentium boards that will recognize the K6?
 
Probably it was supposed to compete with the Pentium-II. It does surprisingly well, but ended up being more of a competitor to the Celeron, simply because there was more parity in pricing.
 
Clippy, you can get a reasonable idea of the way the K6-2/450 performed, together with its competitors (which are listed above and below in more-or-less accurate speed order) from this page here. OK, I grant you it was only Tannin that wrote that, I would have done a much better job myself, but the old bugger ain't too far off the mark.
 
Tea said:
Clippy, you can get a reasonable idea of the way the K6-2/450 performed, together with its competitors (which are listed above and below in more-or-less accurate speed order) from this page here. OK, I grant you it was only Tannin that wrote that, I would have done a much better job myself, but the old bugger ain't too far off the mark.
Click to expand...

A very thorough synopsis. The new site design? Any chance of adding chip power information. I'm trying to shoot low-middle of the performance road but with low power requirements. Oh, and inexpensive. I was initially thinking the most recent cyrix chip, but one of the older AMD chip might fit the bill.
 
Yup, all new HTML, and lots of new pictures, some revisions to the text here and there too.

Hmmm .... Power consumption. That would be a useful thing to add, though I think it might mean a lot of work to track down figures for so many chips. Plugging the figures in would not be difficult, but I'd have to add two extra items of information to make the tables lay out sensibly. That would be no problem though: for quite some time I've been thinking about adding the process technology. Oh and the Voltage.

Which makes three and buggers up my layout again.

Sigh.
 
There was a website (I'm sure it's still around) that listed just about ever processor and how much power they consume. I'm sure if you search hard enough you could find it.

It was mentioned several times on SR before 'The Crash'.
 
Tannin said:
Ahh, you are a Chris Hare's Page fan as well, Mercutio. I might have guessed. It's a goldmine, that place, I've been going there for years. Though I haven't see the voltage charts. I better check again.
Click to expand...

Tell you what Tannin. I'll collect the info for Intel processors if you will do for AMD. I have the information I was looking for Intel but I'm having a heck of a time finding info on AMDs site.

From my perspective, the process technology used is unimportant. Why? The only reason knowing the process technology would be important is if someone had two processors that ran at the same MHz and the same name but performed differently. This never happens. Process change is usually seen in processor evolution not revolution. Voltage and wattage is enough.

Sorry Merc, I couldn't find any useful information on that page.

While not exactly what I was looking for these are interesting:
http://www.macinfo.de/hardware/strom.html
http://www.via.com.tw/en/viac3/pcb.jsp
 
Actually, I like to know the process technology, it tells you quite a lot. Knowing that, for example, the 486DLCs were made on a much inferior process to the AMD and Intel 486 parts tells me that Cyrix were doing it hard: a tiny start-up company making their first venture into full-scale CPU design, competing with the big boys, and trying to do it with outdated wafers thick enough to eat your lunch off. Now that is a lot more interesting than knowing that the Celeron 1000 uses 28% less power than the P-III 1000 isn't it?

(I just made that number up, by the way, no need to correct it.)

But I guess that shows my bias. I have unashamedly lost interest in computers as practical tools. Any practical computer problem I meet these days really comes down to a simple question of "how large a bucket of money do I need to pour over this to make it work right? Small? Or large?" Well, almost any problem.

These days I'm mostly interested in the older ones. In fact, I just spent a happy hour and a half playing with my MFM drive collection. I just successfully fired up a Miniscribe 8425 made on 24th August 1989. Hot damn, that thing was a cheap crappy drive, the 1989 equivalent of a Seagate U 5 or a Western Digital Protege, and it's been sitting in a damp shed these last three years and it still works!

Then, the icing on the cake: a friend gave me a Control Data 94205 the other day: 5.25 inches and several pounds of high-cost, high-class voice coil engineering, and that still goes too.

Err .... I seem to be off topic.

(PS: the link: no picture yet. I have a half-dozen 8425s and 8438s, I just have to test them all and find one that is not working so that I can whip the cover off and photograph it. Maybe this weekend.)
 
Tannin has gone retro.

Poor old fellah. I think the stress was getting to him.

Any of you with technical questions relating to modern hardware had better address them to me instead. As usual, if I don't know the answer, I'll just make it up.
 
Ahh, thankyou Buck.

I've been having a lot of fun with reflections these last few weeks, but I'm fast running out of trees to photograph drives under! (I only have a small garden.) I might be forced, next nice day, to take a collection of opened hard drives for a drive out into the countryside to find suitable greenery to reflect in them.

Not too many nice days left though, it's May already and the clear blue skies of late spring are a long, long way away. Hmmmm ..... but, if I can find one of those magic still blue winter's days, perhaps a tracery of bare-limbed trees might come up well, particularly if it is a drive from a manufacturer that has fallen by the wayside.

At the moment though, for no reason that I can put my finger on, I am rather fond of the Deskstar 34GXP picture - it makes no sense at all, but there is just something I like about it.

Also, the surrealistic and more or less accidental look of my Pentium 166 MMX seems popular. I like the green ones best though.

PS: I've spent quite some time trying to get my chief photographic assistant to hold still long enough to become famous for her beauty as reflected in a hard drive - but she only seems to be interested in motherboards:


woof.jpg


Oh, and mice, of course.
 
I think the contrast in the Deskstar 34GXP picture is what brings it to life. Although, if the acoustic dampener would have refleted that same brilliant blue as in the Quantum Fireball CX picture, we'd have one for the museum.
 
I just looked at your CPU pages Tannin. I hadn't actually read all the way through before.

A few things I can say (besides my usual snide comments about anything Cyrix): Yes, the C6 worked flawlessly without a fan. Acer made a low-profile, legacy-free machine around the C6 that was roughly the size of a cereal box, with only a 40mm PS fan and a heatsink on that CPU that today probably wouldn't serve a northbridge chip, and they worked really well. I really liked IDT's chips. I'm glad it's their technology, not anything Cyrix, that makes up the modern VIA chips.

...which leads me to ponder exactly how much NexGen is in Athlons. I'll bet quite a lot. I digress.

Also, Intel DID make a 266MHz Pentium MMX in a desktop, socket interface. It may have been adapted from the mobile version of the same CPU, but it was made. I know I've worked on machines that had them, although they were generics. I don't know whether the P266MMX is an acknowledged part or not. They certainly were out there, though.
 
Also, I've seen vast numbers of MII-366s. They certainly were a shipping part. A horrible pox on technicians and heatsink/fans everywhere, but they absolutely were a shipping part.
 
Actually, that's how I came to find Chris Hare's page in the first place, many years ago now. I was doing an Alta Vista search one day to see who was linking to redhill.net and I followed the link backwards. It's been a valuable resource for me ever since.
 
Thanks Mercutio. I've updated the M-II 366 entry appropriately. Do you know which of the two clockings was the one that actually shipped? Or was it both?

IDT and P-266MMX I'll do later. Perhaps the 266MMX was, like the K6-III+ an "official bootleg" part. I'm pretty sure it was never an official Intel part for desktop machines.

As for the NexGen involvement in the K7, I daresay it was pretty substantial. NexGen's chief technical guru became AMD's #1 guy for quite a time, as I recall.

Let's see if I can rember the sequence ... The AMD design team were working on the K5 replacement when they bought NexGen. But the Nx6x86 was so promising that they threw their own design out the window and the NexGen team continued work on the Nx6x86 which eventually hit the market as the K6.

Meanwhile, the AMD design team, freed from K5 replacement duties, went straight on to start work on the K7. So, according to that, the Athlon was the work of the K5 people, not the NexGen ones. But I daresay there was a lot of cross-pollination and team switching and so on, so it was probably pretty hard to assign responsibility for it to either company. (Well, they were one company by this time, but you know what I mean.)
 
The MII-366s I saw were mostly on the various PC Chips boards at 100MHz. Given how limited support was for 83MHz bus was, even on the generic boards that were practically made for the dreadful things, I'd hazard to say that was probably the more common part. Even if it wasn't, knowing the builders/integrators around here, once the HSF was on, who knew one way or the other?
 
Yup. We shipped our share of 83MHz bus systems - not M-II 366 parts, 6x86MX-266 - and they were really, really hard to get right. You needed good RAM. decent main board (not PC Chips!), and care with your choice of video card. Even then you had to hold your tongue just right.

Why did we bother? In a word, price-performance. The 266 was just so much faster than anything else in its price range. And yes (refer my remarks in some thread or other yesterday) in those days a speed grade was still a speed grade, not one of these itty-bitty modern things.

But when the K6-266 came out, our 6x86MX-266 buying days were over. OK, the Cyrix part was marginally faster, and cheaper too, but the K6 was just so much easier to get right.

The MII 300, by the way, came later. It was no faster than the 266, just a 66MHz bus, but it was one of the easiest of all CPUs to work with, as easy as a Pentium MMX or a K6-2/450, and we sold heaps of them.

Pity you never got to know the Cyrix chips properly Mercutio. They had their own peculiar way about them, but you could have saved your customers a massive pile of money if you added all those systems up. The thing with them was not so much that they were unreliable or difficult, they were just different. In particular, you'd usually use a completely different motherboard for them. Put a 6x86 on your best-choice board for a Pentium MMX and, sure, it would work, but it wouldn't work particularly well. (I did benchmarks on this once. Very interesting.) Put it on the right main board though, and you'd blow the Intel chip right into the weeds and have enough change left over to think about more RAM or (our favourite trick this) a 7200 RPM hard drive.

And yes .... Alas, far too many of the really cheap and nasty builders used Cyrix chips. That is where so much of the Cyrix reputation came from: not the actual chip, the dreadful crap people used to ship with them, and the horrible things they used to do to the chips along the way, like running them at the wrong bus speed and multiplier, or 0.2v too high because they didn't know how to set a jumper on the mainboard.

Actually, I could divide them into three categories, the Cyrix CPUs, no, make that four:

Work perfectly but dog slow

486SLC

Horrible unreliable things unless you did exactly what they needed

486DX/2-80 (tricky bus speed) (actually, the AMD one was the worst, by the time the Cyrix ones arrived 40MHz on a 486 was not so hard)
5x86-120 (tricky bus speed)
6x86MX-233 (tricky bus speed, general weirdness)
6x86MX-266 (very tricky bus speed)
MII 366 (from your report)

Wonderful performers that only required a little specialised understanding

6x86-120 (cooling, voltage)
6x86-133 (tricky bus speed)
6x86-150
6x86-200
6x86MX-333 (three versions - some were excellent some were dreadful)

Just wonderful performers, totally fuss-free

486DLC-33
486DLC-40
486DX
5x86-100
6x86-166
6x86MX-166
6x86MX-200
6x85MX-300
 
I've worked with them throughout time. The biggest beef I've had with Cyrix chips is unquestionably heat. Cyrix chips are the reason I started carrying around spare HSF combos. When the first 6x86s were released, NOTHING could properly cool them. You'd throw them in a case and eventually they'd overheat, even with the most monstrous HSF combos available at the time.
Then you'd get all the fun that comes from overheating. Data corruption, weirdo video problems, instability, complaining users...

Cyrix chips were very often mated with the lowest-quality boards (PC Chips, mostly, or whatever ALi and SiS were making at the time) in the "double-zero" (sub-$500) machines resellers were just rolling out in 1996 or so. By 1999 they had entirely replaced everything, locally at least, in the budget segment. K6/K6-2s became a "luxury upgrade" for people trying to get a PC on the cheap, and invariably large numbers of Cyrix machines ended up in places they didn't belong, like computers used for business.

I actually had a boutique business going around to replace those awful Cyrix machines' motherboard/CPU with K6s and K6-2s. I couldn't always convince people who had bought $500 PCs to go in all the way for complete board/CPU changes, but I found that moving to a $50 K6-2 and decent fan solved a lot of problems for a lot of people (sometimes adding a case fan or two did the trick as well. I wasn't replacing every CPU I ran across). In 1998 and 1999 I probably replaced 50 or 60 Cyrix CPUs of varying stripes with the cooler and more reliable (IMHO) K6-2/300. I recall most of what I dealt with as being 200, 233, 300 and 366MHz parts.

Subjectively, I also thought the AMD K6-2/300 was faster than the Cyrix366 as well.

I don't recall any specific objection to their 486 chips. That was a long time ago, though.

BTW, ever work on a Cyrix-branded PC? They were sold here in the US for about a year in 96 - 97.
 
No major problems to be had there with the right board. We have 50 or so boxes with the 450Mhz version of the K6-2, and another 200 with the 500 using an ATX board and midi tower case instead of the AT mini tower. All use Soyo boards - 5EHMs for the 450s (AT) and 5EMA+s for the 500s. Problems have been few and far between with all of them - occasional BSODs due to insufficient voltage to the CPU (easily fixed with a Vcore increase to 2.3 or 2.4V from 2.2V) plus the odd board going west because of bad cache RAM. As mentioned they aren't too happy with some of the newer video cards, but ours are used mainly for office apps and the like, so a basic card like the ATi XPert 98 works fine. Incidently, I built a box with one of the K6-266s more than 3 years ago, and it's still going strong (lent it out to a student friend of mine and she still has it :D ).

GM
 
You must log in or register to reply here.
Back
Top