K6-2+ dickering limits

[zappazootus]

After sitting comfortably at 550 on a 450 K6-2+ for some weeks now, (5.5x100 @2.1v) I thought to my little self, "self, lets see about jacking things up a notch." First I tried to pop my bus to 112, but I got nothing but black screen, back to 100, I then increased my multiplier to 6x. Result the same: black screen. OK, Ill increase the voltage to 2.2. Result, BIOS comes up and tells me the chip is posting at 556 (?!) but wont start windows. Sure you say "keep going, incr4ease thy voltage further!" but alas, no, I have a thermal take for a socket seven and a case fan, but I am weary of the long term consequences......So I am sitting back at 550 and thibnking I might just sit tight.:beer:

[azhari]

That's too bad, but why not try fsb settings between 100 and 112?

[theflyingrat]

I see his problem; most old Socket 7 boards don't have any settings between 100 and 112; many are jumpered, and it's just one giant leap of 12 MHz.

Is it possible to take the heat cap off of a K6-2+, so he could get better thermal contact between the die and the heat sink?

[eobard]

My K6-2 450 (non-plus) took 2.8v 24/7 for several months no problem, I'm running it now at 2.7v. To get to 124fsb I had to up my i/o voltage from the board's (MSI 5184) default 3.3v to 3.6v. Maybe you should up yours a bit, that might get you 112fsb. Or it might nuke your chip, your choice, your fate BUT remember, this is overclockers.com, if you're here you have a duty to push your system beyond sane limits. Limits are for the meek, safety is for the timid! Damn it Mr. Scott, I said full power on the warp engines!!

[eobard]

Sorry, lost it there for a minute.

[littlerichie]

two mobo`s that support 133 mhx front side bus are.
soyo sy-5ssm
tekram
both are socket 7
p.s. does anyone know if there are L bridges under that
cover of the k-62 processor.
please post back with results
david

[eobard]

No L bridges under there.

[minoukat]

my Socket 7 mobo, from an OEM, donno which one tho, as it doesn't say the name or anything, supports 133FSB

[zappazootus]

Yup, 112 is the next step up, only other option above 112 is 124 on the FIC VA503+. yikes. I don't really think it is a cooling situation, since I can't even get off the ground. I bet if I increased my voltage to 2.3 or 2.4 it would probibly go, but..., I do not have any hardware utility for temperature checking, and I am on a low budget, so I am keen not to blow this chip even though it is only $35. keep in mind that the "+" series is speced at 2.0 volts not 2.3 or higher like reg K6-2's so making a big jump could prove fatal. Also I have PC100 Kinston Memory, and it may just not like 112 etc...:(

[minoukat]

A K6-2+ costs alot more than "just" $35. As it's only for notebooks, and as they run cooler, are made of .18 micron, and have on-die L2 cache, they cost and have a much greater value than only $35

EDIT : Plus, the K6-2/3/+ aren't made anymore, so only this reason pushes the prices even further higher

[eobard]

My board goes 100, 112, 124, 133. I've got 100mhz noname brand memory and my system is 100% stable at 124mhz. Even at 133 the problem seemed to be either the hard drive or the motherboard cache, not the main memory. I say bump the i/o voltage just a bit and see what that does for you. That's what I did to hit 124 X4.5.

[muddocktor]

I built a K6-3+ 450 system using a tyan s1598 mb for my brother and it's running at 600 mhz(6 X 100) with 2.4 volts vcore and no heat problems. The board does have a socket thermistor and it rarely gets to 100° F. I'm using a cheapo Coolermaster Athlon cooler on it with a 7000 rpm, 50 X 15 mm YS Tech fan that I salvaged off of a Tennmax P2 cooler. You could probably use an inexpensive Athlon cooler like a volcano 2 also. The K6-2+ and 3+ chips can handle 2.4 volts OK if you cool them really well. So far, it's run like a champ for a year now in this configuration.

[bobt17]

u could always lower your multiplier and then up the fsb... that would tell u if your memory is accualy limiting u or if it is your chip

[zappazootus]

Ok- Im convinced- I will try increasing to 2.3 and 2.4 volts to see what I can get out of the CPU. While K6-2+ and 3+ are in wicked (maine talk) short supply, and cannot be found through pricewatch, I bought mine from Tiger Direct for $34.99. I suspect they are either unaware of the value or, the general public is unaware of the value and instead opts for the old K6-2 550 for 50 bucks. Ayuh.

http://www.tigerdirect.com/applications/SearchTools/item-details.asp?sku=CP2-K62-450P%20C

I have to do some finals work on the system today, so I will probibly wait until tonight or tomorrow to jump things up.;) :burn:

[zappazootus]

Ok Folks- here are the results of the latest round of finagleing-

2.3 v (112 x5.5) Started, ran windows, posted only 500mhz- showed some instability. even AMDCPUID only saw 501mhz.

2.3 (124 x 5) black screen nothing
2.3 (124x5) BSN
2.4 (112x5) bsn
2.4 (124 x 5) bsn
2.4 (124x4.5) bsn
2.1 (112x4.5) bsn
2.3 (100x6) bsn
2.4 (100x6) bsn

So at this point, I am just a tad freaked, like "hey man, did it just totally crap out?" But upon setting to 2.0v (100x5.5) everything seems just ducky- in fact it is a lower voltage than I was running before (2.1v) and seems just as stable so far- Did I miss any combos that would be worth trying? I think I probibly need to settle on 550 for the CPU and start dickering with the Radeon 7000........:cool:

[eobard]

Seems like thats your limit. Does your board have an adjustable i/o voltage? That may help. My chip will not do above 600 stablely, and 600 itself (100 X 6.0) is at best mostly stable. Sometimes you just have to accept that you won't get anymore out of your system. And I was so looking forward to shooting for 800 (133 X 6.0) in February (Canadian cooling at it's best.) :(

[RoadWarrior]

I guess you might try disabling the L2 cache in BIOS to see if that's what's crapping out. But the mobo might just disable the onboard cache.

You can take the cap off a regular K6-2 really easily, they look like the same caps on a plus, but I dunno what's under there. On mine (non plus) there were various bridges with resistors on them, I taped over them to avoid any nasties. I got some sticky feet from the dollar store to stop the sink wobbling, and I had to shave them down a little to make it sit tight. How I got it off was by working at the corners with a single edged razor blade, sold many places now as a scraper blade. Here I get them in dollar stores for a buck canadian for five. Your cutting through this stuff that's like grey sealant, I guess it has particulate aluminum in or something.

The result of that on that CPU wasn't very spectacular, it's a 333, and would only do 350 stable. Now it does 375ish stable. Can't get it stable at 400 though, although it posts and boots. I may need a better sink, but it's not really getting all that hot.
Anyway, your milage may vary as they say, you might get a post at 600 with the cap off, you might not.

Oh, a tip I saw somewhere that is really subtle but does actually make a stability difference. Is to ground the heatsink and/or spreader plate. Why it works is because currents can be induced into these by the CPU, which then can re-induce spurious currents into other parts of the processor. These have been measured at values of 100mV, which is 0.1v, when 0.1v can make a stability difference on an overclock, then 100mV of random interference doesn't sound like a far fetched cause of instability. I have seen it make a difference myself. It's real easy to strip back the fan wire some and wind it around the fixing screw.

regards,

Road Warrior

[eobard]

Originally posted by RoadWarrior
Oh, a tip I saw somewhere that is really subtle but does actually make a stability difference. Is to ground the heatsink and/or spreader plate. Why it works is because currents can be induced into these by the CPU, which then can re-induce spurious currents into other parts of the processor. These have been measured at values of 100mV, which is 0.1v, when 0.1v can make a stability difference on an overclock, then 100mV of random interference doesn't sound like a far fetched cause of instability. I have seen it make a difference myself. It's real easy to strip back the fan wire some and wind it around the fixing screw.

regards,

Road Warrior

Can I get a link to wherever you saw that at? That sounds interesting and I'd like to check it out in depth.

[RoadWarrior]

ack, took me a while but I found it, it's kind of buried in this article (http://www.lostcircuits.com/cpu/peltier/) at lost circuits. It attempts to find out why peltiers offer seemingly better stability than other methods of achieving the same cooling, and the author discovered the induction phenomena.

regards,

Road Warrior

[eobard]

Much appreciated, thanks a lot. :)