Sign in with
Sign up | Sign in
Your question

AMD K6-3 vs. K6-2

Last response: in CPUs
February 17, 2003 6:58:03 AM

I just threw together a machine to watch DVD's. It has an AMD K6-3 333MHz processor o/c'd to 375MHz (won't make it to 400). It also has an ATI All-in-Wonder 128 PCI graphics board, and 256mb PC133 ram. DVD playback is ok, but has a few glitches once in a while (may be a problem with the ATI software though - hard to tell).

For those who are familiar with these Socket 7 processors, do you think I will gain or lose speed if I switched out the 375 Mhz (o/c) K6-3 for a 533 Mhz (o/c) K6-2 (and I suppose I may be able to o/c the K6-2 to 550 Mhz, but I'm not sure). From what I've read, it's hard to tell which will give better performance (especially for watching DVD's).


More about : amd

February 17, 2003 5:22:37 PM

i didnt know k63s came in speeds that low....

wasnt 450 the slowest? 0.o

anyways.. hard to say

the k63 has full speed on die cache, makes a big difference in performance. the k62 doesnt have any of its own cache. plus the k63 is slightly faster per clock because it was a revised version of the k62...

but, i had a k62 500@550 and it ran very very well for what it was. i ran windowsXP on it, and played alot of games including quake3 and others...

the slow clock speed of the k63 might hold you back. only way to find out is to put one in, try it, then shut er down and slap the other cpu in =D

its a 1 minute process =)
Related resources
a b à CPUs
February 18, 2003 12:24:14 AM

Yes, like Lonelymaggi said, you'd be better off simply to get a DVD decoder card. I have a DXR2 and it rox, you can even use the TV-Out to record DVD's to VHS.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 18, 2003 6:43:07 AM

I'll answer everyone in one post! Yep, it's a 333MHz, O/C'd to 375. They made it as a laptop chip. Very, very low power consumption. Quite unfortunatley, it's not in Tom's new processor comparison.

As for just trying the K6-2... that would mean I would have to buy it! It may only be $26, but I would hate to throw that much money away if it turns out to be slower.

The MPEG card ideas are interesting. But the ATI All-In-Wonder 128 has all sorts of hardware DVD jazz... would those MPEG cards really make a difference compared to the ATI AIW? Do they generally include DVD playback software to utilize the card's functions?
a b à CPUs
February 18, 2003 7:02:21 PM

I'm going to let you in on a little secret: Your CPU can take more voltage. 2.4v might be ideal for overclocking. And you can attach an Athlon cooler, they do fit.

My DXR2 card allowed me to get professional quality output on a Pentium 133! It does ALL the data conversion, it takes the load COMPLETELY off the processor. And it has handy features for the TV output, for instance you can fix the screen width and height with far more flexibility than the output of a video card allows. You can bypass Macrovision and use it to record to VHS. You can change the aspect ratio like a pro and have a letterbox image fill the entire screen any way you like (handy for making VHS tapes). In essence, with my DXR2 card you could make store quality VHS tapes and sell them on a street corner in new york for $5 each, and the quality would be so high that you'd put the other street leaches out of business.

The AIW128 is good, but it needed a powerfull processor to get perfectly smooth results.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 18, 2003 7:06:42 PM

So you were that guy I saw up there in Times Square!

I had a dxr2, but had to get rid of it. It worked with almost every card on the market, except my voodoo 3.

<A HREF="" target="_new">;/A> Southeast Pennsylvania gets an Anime Convention!
a b à CPUs
February 18, 2003 7:19:48 PM

Really? Hmm. It's an overlay card, so it should work even if the desktop display was from an entirely different system!

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 18, 2003 7:22:48 PM

yeah, but when you would hook up the 2 withthat jumper cable, the comp would not boot. It was a pian in the butt, but I got over it and got a better system. 900 athlon classic has no trouble with dvd decoding :) 

<A HREF="" target="_new">;/A> Southeast Pennsylvania gets an Anime Convention!
February 20, 2003 5:51:00 AM

Outside of just bumping up the multiplier for old socket 7 CPU's, I've never played with overclocking. I didn't realize they could handle more voltage to help overclock more. How do you know the voltage range for a processor? In specific, the K6-3?

As for the dxr2 board, I'll take a look at them. They are a pass-thru board, so your video signal always travels through it. Although I have a cheap PC, I have an expensive monitor (you only get one set of eyes). And I wouldn't want a crappy signal messing up my nice monitor. Are there any downsides to the dxr2 boards? Maybe if I crank up the K6-3 a little more, I won't even need it. All I want to do is be able to watch a DVD movie smoothly on my computer monitor... nothing fancy.
February 20, 2003 11:12:23 AM

Just adding my 2 cents:

Here is a comparison I had before,

I had a K6-III 450 + GeForce256 (ASUS) and the software decoder it came with (think it used the gfx card a bit though).

My sister had a Athlon 550, + ATI-All-In-Wonder 128 (32MB) with their "Excellent" DVD decoder.

Both systems had 256MB of PC133 memory, and ran windows 2000.

My System:
Skipped maybe once or twice per DVD.

Her system:
Skipped maybe once or twice per SECOND.

Athlon XP 1600+, MSI K7T PRO2 RU (POS), 2x256 MB CRUCIAL PC2100 CL2.5 memory, Asus V6800 DDR Delux (GF 256) video card, 6.4GB+27GB WD HD, 40GB IBM HD (all 7200RPM). My computer is an acronym
a b à CPUs
February 20, 2003 2:55:10 PM

I "go by instinct" on voltage and haven't burned anything yet! There are things to consider such as die process size (how large the tiny components of the CPU are) and heat. It just so happens that the K6-2 and III will reach a certain point where adding more voltage doesn't help you get any farther, and that point is well below burnout voltage.

AMD for example rates the K6-2 as having 2.5v max voltage, yet some of these use 2.4v stock. I've found that the K6-2 will not burn out at 2.8v core, over several years of use, as long as you keep it cool. The K6-III is a little more fragile, so I keep my recommendation at 2.4v max.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 21, 2003 8:12:51 AM

That's a real interesting comparison. I've got the combination of those two systems: the K6-3 with the ATI AIW 128 (PCI).

I wonder if I have the stronger two components or the weaker two!?! I've only played a couple DVD's, and one keeps on skipping in the same spots. It could be scratches, but I doubt it, because it skips exactly the same way in each of the spots in which it skips (at exactly the same frame). If it had scratches, I would think the skips would have a little randomness to them. Also, the DVD looks perfectly clean and scratch-free.
February 21, 2003 8:28:30 AM

Overclock it. I've had experience overclocking a K6-3+ 450 to 900 MHz (0 Stable) and 1 GHz (Wouldn't make it through a day) and it was running at 28-30C. This was with an Athlon cooler of course.

See if you can get her up to 500, or even 550. Of coursem you want to keep your FSB and RAM at the same clock (PC100 400, 450, 500 ect)(PC66 466,499,533,566 ect) for aid in stability. Of course your motherboard may be an issue. If you have an ASUS p5a you'll do fine, otherwise you may not get some of the higher clock speeds.

My Duron 1300 eats P4's for dinner. Oh, the Celerons make a nice snack too.
February 21, 2003 8:55:36 AM

Thanks for the info on voltage.

Here was my process in the past for overclocking. First, get a good CPU HSF. Then clock it up one step at a time until it wouldn't boot or was too hot for comfort. Once I had found the approx maximum for the CPU, get in there and adjust the multipliers and CPU clock speed so that the buses (incl PCI) were running at good speeds as well (without making the CPU go faster than the maximum found earlier).

Your suggestion of modifying the voltage is new to me (but perhaps old news to frequent overclockers). How does it get added to the above process? TIA.

(BTW, it's running on an AT power supply... does that matter? I have an ATX ps that I can swap in no problem (mb accepts both)... I've always liked the AT supplies more for machines that don't get used too much because when they are off, they are really off, and not sucking a few watts all the time like ATX power supplies.)
February 21, 2003 9:01:42 AM

Why does the motherboard make a big difference?

I think it's an FIC 503A. Runs real stable right now, but I need a little more performance boost. I'm running the 333MhZ mobile K6-3 at 375mhz and have the PCI bus up to around somthing like 41Mhz, if I recall (running PCI video... ATI AIW 128). The board uses a combo of jumpers and the BIOS to set voltage and clock speeds, so I have to look at both to check the details. I'm running a pretty standard generic $10 HSF, and have a retail official AMD HSF that I can use instead.
February 21, 2003 9:23:14 AM

The P5A is a Super7 Overclocker's delight, it has all the options wanted, and the board itself is stable. Other boards will do fine, but you wont get 100 percent the results of a P5A, maybe like 80 percent.

I overclocked a K6-2-350 to 450 once with a lame DFI mobo. When I plugged it in a P5A, I was able to squeeze 500 out of it, though it wasn't 100 percent stable.

Does your MOBO have onboard heat sensors? Keep your temps below 80C to be on the safe side.

As for voltage increase, it allows higher clockspeeds because higher frequencies require more power to flip the transistors (like a on/off switch) reliably. Voltage adds stability to a point. You'll find that you can clock higher if you increase voltage. If you have a spot that runs, but not reliably, jump the voltage up a tad and it will probably run stable.

My Duron 1300 eats P4's for dinner. Oh, the Celerons make a nice snack too.
February 21, 2003 9:30:02 AM

The K6-III (Sharptooth) has a maximum core temperature of about 70'C, so I would make sure it stays well below 60'C due to the inaccuracy of the onboard thermal sensor.

I'm looking for an old K6-III anyone know where I could find one?


Who's General Failure and why's he reading my disk?
February 21, 2003 9:32:08 AM

Oh yeah, I keep getting my K7's and K6's mixed up, that is twice today! Hhahahah. Yeah, keep it below 70 for sure, around 60 is fine.

Tip to the overclockers: Get an IR Thermometer, you can point at things and get the real temp.

My Duron 1300 eats P4's for dinner. Oh, the Celerons make a nice snack too.
February 21, 2003 9:51:00 AM

My old K6-2 300 ran fine at 350 but at 400 it wouldent boot, ya can't get anything more out of ya 333, the K6-3 where alot faster then the K6-2 though, the K6-3 had the L2 cache on the CPU so the L2 cache on the mobo became L3 cache which made the K6-3 alot faster, but going from 375Mhz laptop CPU to a 550Mhz desktop CPU would preobebly be alot better option!, but its all up to you, I know that my old 350 with a GF2 Ti did'nt play DVD very well at all till I sold it to my parents and got myself an 1800+

1 Set Of Morals For Sale, Never Used, Will Sell Cheap
a b à CPUs
February 21, 2003 4:31:05 PM

Voltage is a messure of electrical force. Amperage is a measure of volume. And you also have resistance.

Imagine a water wheel with scooped paddels. Each paddel of the wheel represents a chunk of data. Now imagine the water line has a filter. This filter represents resistance. Now imagine a gate that opens and closes to let the water through into those little paddels. This gate represents your clock generator.

OK, now, you set the gate to open and close a certain amount of times per second. Everything works good. So you keep turning it up little by little, unit it's finally opening and closing so fast that the paddel buckets no longer have time to can't make the wheel go any faster...unless you increase the volume of watter through the pipe. The only way to get more volume through the pipe is to increase pressure through the filter. You can't simply increase the size of the pipe, you need pressure.

OK, well, another problem is that you will eventually reach a point where the wheel can't spin any faster, because water wheels are gravity powered and gravity has a finite rate of acceleration, that's your upper limit.

In the CPU, increasing voltage allows you to push data at a faster rate through the resistance of the circuit. But it also increases the amount of heat. And heat makes the CPU unstable. Adding even more voltage helps overcome that instability, but creates even more've reached a limit.

Generally the best voltage to use is something less than that point where all the extra voltage is lost to heat. Some processors will burn out before they reach that point, most won't. But many will burn out from a hot spot, over time. I'm not going to talk about electro migration and so forth because it's not pertinent to such a basic discussion.

Anyway, you'll find that there's a certain point where adding more voltage no longer helps you. "The wheel just can't be spun any faster".

The best way to find the limit is experimentation. Experimentation shows me that with air cooling, voltages above 2.8v don't help the K6-2. Voltages above 2.05v don't help the PIII Coppermine. Etc. Each processor varies a bit, but you can come up with some fairly close approximations.

Voltages above 2.4v didn't make my K6-III run any faster. Based on my knowledge of the K6-2, this seemed reasonable, so I recommend 2.4v or less on an overclocked K6-III.

ATX power supplies often help AGP equiped AT boards run their AGP cards with more stability, because many can take their 3.3v input directly from the power supply, instead of using the weak onboard 3.3v regulator (AT power supplies have no 3.3v line).

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 21, 2003 6:59:53 PM

Now THAT my friends is a class 'A' description of AC voltage/resistance/current/power. I wish I could have come up with that description. I quit writing because I was tired, and I didn't know if anybody would understand it when I was done. :p  (hey, 2 years in electrical/electronics/avionics training can do that to a guy)

I'm gonna copy that and print it for later use if you don't mind, I think it is a awesome explanation of frequencies ect.

My Duron 1300 eats P4's for dinner. Oh, the Celerons make a nice snack too.
a b à CPUs
February 21, 2003 7:20:15 PM

Sure. Feel free to "slap some lipstick on that pig" aka edit it for clarity and spelling.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 21, 2003 7:22:13 PM

Cool, good deal.

My Duron 1300 eats P4's for dinner. Oh, the Celerons make a nice snack too.
February 22, 2003 3:23:34 AM

Wow! Great posts! Thanks to everyone, especially crashman for his great explanation. I'm going to read that over again a few times!

I'll have to experiment with your ideas a little. I just hope I don't fry the processor or create a fire.

I never tried increasing the voltage at all to see if it could handle higher frequencies with a little more voltage. I assumed a processor had to run at a specified voltage or it would burn up (either quickly or over time). Since I'm already at 375mhz with a 333mhz processor, I doubt I'll get much more. But I'll try increasing the voltage to the next increment, and see if I get a higher frequency out of it. I'll keep a close eye on the temps (especially during processor intensive operations) to make sure it doesn't get too warm.
a b à CPUs
February 22, 2003 3:59:17 AM

Actually, processors DON'T have a specific voltage by design, one is arbitrarily applied by the manufacturer. Instead, they have a voltage range. Prior to the release of special CPU's for laptops, many standard processors were specified for laptops because they could be used at lower voltage. In fact, many of these laptop chips were underclocked desktop chips, they ran them slower so they could reduce the heat and power consumption. One method of reducing heat and power consumption is to reduce voltage.

This can easily be seen by Intel's voltage of the Celeron 566 at 1.50v, and the PIII 750 at 1.65v, both on the exact same revision of the exact same core. Later PIII's came with the voltage at 1.75v, as Intel attempted to standardize the voltage across the entire line of PIII's. Slightly updated core, same die process, didn't really need 1.75v except at the higher speeds.

Manufacturers will often specify an upper limit for voltage, but sometimes even these aren't perfect.

<font color=blue>Watts mean squat if you don't have quality!</font color=blue>
February 22, 2003 4:49:04 AM

You have amassed an amazing amount of quality information! Thanks for sharing a tiny piece of it!

Keep on talking... we're listening... and learning! :-)