anyone know where to buy an AMDK6-III+ online?

[carolyn32909]

I have seen them on e-bay, but I wanted to know if anyone has seen them anywhere else online. Thanks

 

[Cliptin]

I have seen them on e-bay, but I wanted to know if anyone has seen them anywhere else online. Thanks

You should also check Pricewatch.com.

Welcome to the forum.

 

[CougTek]

In the United States you mean? Or elsewhere around the globe?

 

[Cliptin]

In the United States you mean? Or elsewhere around the globe?

FL? 32909.

 

[timwhit]

Try here:

http://www.computeradjusters.com

I didn't see anything on pricewatch.

eBay is probably your best bet.

 

[CougTek]

FL? 32909.

Sorry, at first I thought it was a birth date or something (didn't count the # of digit). Here, our postal codes both include letters and numbers so I'm not familiar with your US addressing.

 

[cas]

If you are interested in the K6-III(+), you may want to take a look at this recent thread.

 

[Groltz]

#1

#2

#3

#4

 

[cas]

If I am not mistaken, all four of the above links, point to 0.25µ K6-IIIs, rather than the 0.18µ K6-III+. The 0.18µ parts were not really intended for desktop use, and were pretty scarce, even when new.

 

[CougTek]

#1
#2
#3
#4

It's aweful to see that a K6-III is just as expensive as a 1600+ Athlon XP.

 

[CougTek]

An old review at Lost Circuit comparing the K6-2 and K6-III+. Unsurprisingly, the later wins. But both would be wiped by any modern CPU (yep, even the Celeron) in almost any task though. Please remember this before making your purchase.

 

[Tea]

No no, Coug. It's great to see that the wonderful new Athlon XP 1600 is so cheap that you can pick it up for around the same price as an old K6.

 

[Tea]

Any FPU intensive task, Coug.

 

[CougTek]

Any FPU intensive task, Coug.

Any task, Tea. No matter how poor the IPC of a Willamette P4 is, at 1.7GHz, it will kick the ass of a K6-III. Especially considering the platform it uses has far less bottlenecks than the old MVP3 chipset.

MS Office, photo-editing, games, F@H or even web surfing. Anything will be faster on a Celeron than on a K6-III. We are talking about a two CPU-generations gap, with several small increments in the meanwhile. Pushing a K6-III against a current CPU is beyond rationality. I know you liked it back then, but nowadays, it's time to let the bone go. Drop the seat of the pant argument and run some benchmarks. I don't have both setups here (I refuse to sell P4-Celeron when the Athlon XP is so competitive), but the outcome is easy to figure out.

Currently, the cheapest Willamette Celeron computer I could assemble would have an SiS645-based motherboard and PC2100 DDR SDRAM (I would have a hard time to find i845 non-D motherboard and they would be dearer than the SiS645 anyway). If you think that a K6-III on anything (your best bet would be the 2MB L3 cache TYAN Trinity board that has been released late at the sunset of the SS7 days) could defeat a Willamette Celeron on a SiS645 motherboard with PC2100 RAM, please show me. (hint : you won't)

The most miserable results I saw from the Willamette Celeron was its Winstone results, but even then, the K6-III wouldn't beat it I'm pretty sure.

 

[cas]

...your best bet would be the 2MB L3 cache TYAN Trinity board that has been released late at the sunset of the SS7 days

The large L3 cache really won't help. Even proper PC133 implementations offer higher measured performance than the L3 cache of a K6-III system.

 

[Tea]

Not a task that fits inside the K6-III cache and doesn't require the FPU, Coug. And I don't count these castrated Pentium 4 thingies as Celerons. It's a completely different chip, for the love of Mike! If there is any such thing as a "Celeron 1700" then I'm typing this on a K6-4/1800. :mrgrn:

 

[carolyn32909]

Thanks for all the info on the AMD K6+!

This computer that I am putting together is an old baby AT that has been sitting around the house collecting dust. Its so old that it has 30 pin ram in it. It was a 486. I wanted to put a small amount of money into it so I can learn how to build a computer, overclock and upgrade bios, etc. I didnt want to learn on my main computer which is an AMD Athlon 1.2 Ghz.

It sounds as if I could get away with a celeron 300A or just an AMD K6-2+ depending on the price. A friend gave me an Asus P55T2P4 mbo that he had pulled out of his machine and it has 64 MB if Ram in it already. So basically all I need is a processor to get this thing running. Then if I crash and burn- it's all just a learning experience!!! 8)

 

[time]

In that case, settle for a K6-2. They are a lot easier to find. The + version is like hen's teeth, however, intended as Cas says solely for mobile use.

The vast majority of Celeron 300As would be on ATX motherboards rather than AT.

 

[Tea]

You will probably find, Carolyn, that the ASUS won't work with a K6-III+ in any case. Very, very few boards support the II+ or III+ - it's not just a matter of having 2.0V support and setting the multiplier right. Nor is it one of those purely cosmetic issues that result in you getting a system that reports "Unknown CPU at 66MHz" or some such garbage when, in fact, you have a perfectly functional K6-II/300 or some such and the only problem is that your mainboard's BIOS doesn't have an appropriate CPU ID string to put on the screen for it. No: the II+ and IIi+ have a functional difference to their earlier bretheren. I don't know what it is, but you almost always need a BIOS update specifically for those chips.

Very few board makers did this - the II+ and III+ chips were manufactured for the notebook market after all, and not many escaped onto the general market in PGA desktop packaging. There will no doubt be a handful of others, but the only motherboards I know about for sure that definetly work with a II+ or IIi+ are:

FIC VA503+ (requires an undocumented beta BIOS)
FIC PA-2013 (requires a different undocumented beta BIOS)
Gigabyte GA-5AA (you can use the latest official BIOS with this one)

OMG! I just looked up what the ASUS board you have actually is! Ouch! An HX!

Hoolie Doolie, Carolyn, I'm going to have to introduce you to Tannin and his extensive collection of 286 CPUs and MFM hard drives - that thing is right out of the Ark. Successfully running a K6-III in that relic would catapult you right into our world-famous Pointless Overclocking Hall of Fame - right there on the podium alongside Sol and Clocker.

As a practical use for a K6-III+, it is a travesty, a crying shame, a waste of a wonderful little CPU. So I think your idea is stupid. ....

(Tea looks thoughtful for a long, slow moment. The corner of her lip twitches, and a sly grin gradually spreads across her face.)

.... Which seems to me to be an extremely good reason to do it. :wink:

After all, there would be a certain pleasant symmetry to it: the last and best of the Socket 7 chipsets, mated with the last and best of the Super 7 CPUs. (No, no, no - don't tell me the TX came along later - the TX wasn't a Socket 7 chipset, it was a pile of mule droppings heavily disguised as a chipset. The HX was the last real Socket 7 product. Or the last one from Intel at any rate. And unquestionably the best of them.)

May I suggest some suitable other "last and best of" items from more or less similar eras to round out the system?

• 
  A Seagate Decathlon 850 hard drive. (Last Decathlon. Best IDE drive of any from that era.)
  
  A Tseng ET4000 chipset video card. Or possibly a Diamond Monster Voodoo I.
  
  A Honeywell 101WN keyboard. (Last Honywell product before the horrible Windows keys arrived and, with an IBM or two, best keyboard ever made.)
  
  A Logitech Trackman Vista. (Last Logitech trackball to be designed by a human being with actual hands and a brain, rather than a marketing droid with a claw full of red and blue LEDs and the firm belief that "ergonomics" is something to do with the extraction of crude oil from the Lybian desert.)
  
  One of those wonderful NEC floppy drives with the super-dooper spring in them - the ones that used to fly half way across the room when you pressed the eject button.
  
  A Panasonic 562 2-speed CD-ROM drive. (Nearly the last of the proprietary interface CD drives, and far and away the best of them.)
  
  A Sound Blaster 16. (Best, most compatible, most reliable ISA sound card ever made, and the last wholly good product ever to leave the designing room at Creative Technology.)
  
  A Hewlett-Packard Deskjet 500? A Laserjet II? An Epson LX-850? A Stylus 400? No, no - a Star NX-1000. Utterly indestructable, and harder to improve upon (short of a complete fresh sheet of paper) than a Volkswagon Beetle or a DC-3.
  
  A 14 inch NEC Multisynch.

 

[carolyn32909]

I'll check those FIC motherboards out!

Yeah, I know Tea -it is a stupid idea - but of course I am going to do it! Ya know it's always an Adventure! I'll check out some prices online for those motherboards. Once I get an idea how much both the processor and MOBO will cost, I can figure out where I want to go from there.

I do happen to have an NEC multisync monitor and a Sound Blaster 16 sound card - so I guess that's a start. I do need to get the Vodoo I card too. The system already has an oooold Trident card in it. I havent checked into that because I planned on just getting the voodoo card.

 

[timwhit]

The Voodoo card is an add-on. It is only for 3D, so you will need a 2D card anyways.

I think I still have one somewhere, I think it is a Diamond Monster 3D.

You know you could build an almost new system for a little more money. Something to consider, you could still use all those old crappy components in it and get a newer motherboard and a Duron or something similar.

 

[Tea]

Oh, Timwit. Don't be so ... so practical.

 

[timwhit]

DAMN IT TEA you did it again. TIMWHIT TIMWHIT TIMWHIT

 

[P5-133XL]

If your goal is to learn about configuring an older machine it should not be any newer than a 486 on an AT motherboard with only ISA/LB/ESIA/MC slots (No PCI), no integrated I/O (requires a Multi I/O card and to properly learn about I/O cards it should not have a manuel giving jumper positions or if it has a manuel, the printed jumper positions should be wrong). Several cards should conflict in one or more IRQ, Memory location, and port locations. The network card should be totally incompatible with the video card and the sound card is incompatible with every card. The HD should be MFM/RLL/ARLL and should not match the computer BIOS drive table, requiring a custom bios from the HD controller manufacturer, which has long since gone out of business. Both floppy drives should be jumpered as B: with no instructions on how to make them both work on an untwisted cable or twisted cable. To top it all off, the CMOS battery is intermittantly failing and the clock chip runs 1/hr. per day faster than it should with the customer adament that the computer keep proper time.

Can we have fun, or what?

 

[Mercutio]

Geez Mark, you forgot that she'll need to install her SIPPs one at a time.

Personally I always enjoyed dealing with wide SCSI drives on SCSI-1 controllers (the adaptor is HOW much?)

 

[P5-133XL]

Merc,

You ought to know better - Some Sipps were installed one at a time, some were in pairs and some machines had four banks that had to be equally populated. The last 486 I worked on didn't use sipps at all, but actually 4 banks of actual chips (16K, or 64K chips). I have had to deal with each and every one of those problems before. And yes, I've had my fun with SCSI too: It is just too much fun dealing with terminator resistor packs that are not included with the drive and incompatible with the spares one has on hand. I think my favorite though is configuring unfamiliar multi-I/O cards without a manual: The number of trial and error permutations to make it work properly is just a spectactular time waster. No, it is not totally trial and error, but it is close.

Living in those times was truely a learning experiance that newbie's don't get now days. I suppose each era has its' own crosses to bear.

 

[P5-133XL]

Oh, by the way, once your done making that system work, the customer will upgrade to a 10GB IDE HD and the requirement to install NT4 on the machine. You too can discover the joys of breaking the 512MB barrier without the use of a PCI IDE adapter.

It's stuff like this that makes it all "fun"

 

[timwhit]

Ya, It's called drive overlay software. Takes about 2 minutes to install.

 

[P5-133XL]

Ya, It's called drive overlay software. Takes about 2 minutes to install.

I just looked at the current version of dm (2000) and you are probably right now, but that was not the case untill recently. The previous versions would not work with NTFS and NT4 did not support Fat32. The result was one could only create 2 GB FAT-16 partitions and that was where you had to install to. Once installed you could boot to NT then one could use NT to create NTFS partitions but one could not secure the boot partition because it was fat-16. Any attempt to change the boot partition to NTFS using utilities like Convert would destroy the drive overlay software.

 

[Mercutio]

I don't recall any 486-era system boards that used SIPPs actually. I think they pretty much died on early 386 systems.

Internal modem + cheap ISA soundcard + an I/O board or two + SCSI controller = a fun couple of hours for a tech. Especially when two or more of the above don't have manuals.

 

[carolyn32909]

386-486 Those were the days???? Haaa!

I agree -Newbies just don't get the exposure now a days. Technology has really come a long way in just the past 5 years. Those 386's and 486's sound really impossible and frustrating.

My PC Repair teacher was telling us that in their day those machines were pretty incredible. But I know I am lucky to be coming into the field now when machines are somewhat more user friendly. I haven't repaired computers professionally at this point, but when someone asks me a question - it's usually about an older system. A lot of times its really hard to try to explain problems to them too.

Also, I am trying to get some classes in at the local college, so I can prepare for the A+ Certification. The study questions for the exam are insane. What a mess that is. I just keep finding more questions on the web and taking practice exams. We'll see!

 

[cas]

Re: 386-486 Those were the days???? Haaa!

I haven't repaired computers professionally at this point, but when someone asks me a question - it's usually about an older system.

I used to repair(and build) systems in '91, when the described 386 and 486 machines were brand new. At that time, a large percentage of the repairs/upgrades were for original IBM 5150s, which were ten years old at that time.

While I question the economics of building a K6 based system today, I admire your initiative in learning more about older systems.

 

[Mercutio]

If you'd like, I can give you all the A+ material in the entire world. I teach it, after all.

 

[Tea]

New machines easier? Ha! Not even close. Seriously, new machines are much, much more difficult. In the old days, instead of wrestling with Plug and Play we used a system called Plug and Work. You had to know what an IRQ was and how to set it, but after that, you just plugged the card in and it worked. First time, every time.

 

[time]

Unless it was ... VESA LOCAL BUS!

:eek4:

Cracked motherboards, anyone?

 

[Tea]

VESA was a bitch. Even Tannin doesn't miss it.

 

[carolyn32909]

A+ Exam

I would like all the A+ information in the world Mercutio! I want to take the exam soon, but I want to make sure I am really ready. The testing center charges $150 for both parts. (which I understand is the going rate in my area).

 

[CougTek]

New machines easier? Ha! Not even close. Seriously, new machines are much, much more difficult.

I never realized I was such a genius by building all those new boxes without much trouble. :roll:

 

[cas]

Seriously, new machines are much, much more difficult.

New machines are much more complex. Whether that means they are more difficult depends on whether that complexity is working for or against you.

Plug and Play is an excellent idea. Novice users should not be expected to understand what IRQ lines, DMA slave channels, and IO base addresses are. That said, for those who do, this simple arrangement was far superior to the period that immediately followed it.

The first implementations of PnP worked by writing patterns to ‘known’ addresses, and checking for expected behavior. This had the frequent side effect of locking up a system solid.

Worse, the first ISA cards that I remember without jumpers stored their IRQ/DMA/IO allocations in EEPROM. Since in situ programming is the only option for most people, incorrectly programming the device once could make it useless forever (at least in a given machine).

More recent busses like PCI/AGP, and USB were designed to support PnP operation. Despite Tea’s anthropoidal cynicism, they actually work very well when the specifications have been adhered to.

If for nothing else, I am pleased that I don’t have to memorize CHS values for a dozen popular drives.

 

[e_dawg]

Heh, memories of IRQ, memory address assignments, and CHS values... If you can believe it, I had to maintain a fleet of computers with those features in 1999. I left shortly after making everything Y2k compliant, but I bet they are still using those 386sx computers at my former job. Ever try running Windows 95 on a 16 MHz 386sx with 4 MB of RAM and an 80 MB HD?