View Full Version : celeron 667->1000 OC problems
Link_flap
05-15-01, 11:11 PM
Hi all,
I'm trying to OC a cel 667@1000. 75 MHz & 83 MHz FSB is ok except it turned out that the new generic PC133 I bought didn't work properly (not even at 66MHz)
The problem is the CPU seems to need at least 2.0v to POST. I haven't applied thermal compund yet as I forgot to buy it. Now I'm thinking: if that chip needs 2.0v just to POST at the big 1G, does it really stand a chance to reach 1GHz stable?
What I'm getting at is that I will probably have a hard time returning it if I've lapped it. Is lapping even advisable on a FCPGA?
System:
-----------
mobo: aopen ax6b
mem : hyundai PC100
proc. : cel. 667 cC0 sl4nz
HSF : GW FOP32-1
I read in the CPU database that all other sl4nz have reached 1GHz, though some of them were running the chip at 2.2v
How long could I expect this chip to last on 2vthe cooling being a GW FOP-1? A week, A month, Same as the accesstime of my RAM :) What about 2.2v?
Thanks for reading this far, and hope you'll reply.
/link_flap
YIKES!! I wouldn't even TRY to o/c it withOUT any thermal paste!! Frankly, I wouldn't even try to RUN IT w/o thermal compound. Especially at the high vcore settings you're trying to push thru that chip!!
Slow down a bunch.....first off, get some thermal compound. the Radio Shack stuff will suffice until you can get the good stuff (Arctic Silver).
Dou you have ANY means of monitoring the temps?? With no compound, they've gotta be high. Make sure the cooling is GOOD before trying to o/c to the extremes you are attempting.
I wouldn't reccommend running more than 2.05v. Extremely high vcore settings for extended periods of time cause a condition called "electromigration". This is damage to the circuits of the chip itself.
I can get my C600 to POST at 1062 @ 2.1v, but I wouldn't run it there....I back it off, first off, to where it runs stable, and then at the fastest stable point at 2v or less.
Don't cook your chip....get some thermal goop on it, and keep the vcore down to around 1.95v The chip will last a lot longer.
Mr B
Yeah, what Mr. B said. I would not lap the CPU if there is any possibility it'll ever be returned. I would either lap the heatsink or get a better cooler. Not all those chips will hit 1 gig. Just because some people did and posted to the database, doesn't mean they all will. Most people don't post their failures, I'm guessing.
First, work on cooling. Until you get excellent cooling, I would not go over 1.9v. Personally, even with really good cooling, I never exceed 2.0v for any length of time, maybe 2.05v for short periods of time. For normal usage, I usually run 1.9v, but I also have lots of cooling mods done.
I have overclocked a 667 celeron once, it was extremely smooth. If after applying the thermal paste and good hsf you are still having these issues, welcome to the 20% who can't overclock this particular celeron (those aren't my numbers, they're overclocking standards). I happen to be one of the lucky ones (the 80 percentile) that got great results: 1.75v 38C with stock intel hsf. The chip I bought overclocked without breaking a sweat, must have slipped through the cracks at Intel.
Can't guarantee you'll get the same results, but you should.
Link_flap
05-16-01, 03:02 PM
Thanks for the feedback!
Heh, I know running without goop isn't a good idea, but when you have just received your goodies and all thats missing is "the white stuff", it's oh so hard to not try it :)
I wasn't expecting to abe able to run it, but more trying to check how high it would post. In my experience, cooling will mostly help to keep the chip running, not help it post.
Anyway, tried getting some thermal compound today. No luck! Small towns suck sometimes :( One computer store said "they didn't need it" and the other didn't sell it, but had some thermal glue type stuff I could use (stayed clear of that, though)
Some good news: I get the PC133 RAM stable (so far) by changing slots.
Still thinking about lapping the CPU... How is it done on a CuMine? How can I keep the CPU parallel to the sandpaper/glass? Can't see how you could possibly get it flat...
More to come...
/Link_flap
I have the same chip. I have been able to run at 1ghz but cannot play games. I need to get a better hsf, currently have golden orb. What I had to do was burn in for a short period of time at 66mhz at 2.0v. Then I was able to post at 100 fsb. I just can't get it stable. Runs at 833 mhz, 1.7v with no problems at all. I wish my mb had something between 83 and 100 fsb. I would settle for 950 mhz.
Link_flap
05-17-01, 04:46 AM
Sorry to hear that Campy.
Have you tried:
*a brand slocket?
*moving your mem to different slots
*open case with large household fan (to see if the GOrb is the problem)
*bump I/O voltage a tiny bit if you can
Whats your S-spec?
/Link_flap
Link_flap (May 17, 2001 04:46 a.m.):
Sorry to hear that Campy.
Have you tried:
*a brand slocket?
*moving your mem to different slots
*open case with large household fan (to see if the GOrb is the problem)
*bump I/O voltage a tiny bit if you can
Whats your S-spec?
/Link_flap
I will try a couple of them. I have sl4nz (s code).
Thanks for the suggestions
Trying another slocket as stated above would probably be a good idea. With my Celly II 600 I went from unstable at 1008 with 2v using an Abit sloket to completely stable at 1008 with 1.9v using an MSI 6905 Master sloket. Seemingly insignificant changes can sometimes make a huge difference.
Lapping the CPU is risky. Lapping the heatsink is safe and does help a little. Definitely wait until you get some thermal paste before doing much else. Sounds like that might be a good chip, no use burning it up.
Link_flap
05-21-01, 04:42 PM
Ok, here's an update:
After applying thermal compound (0.8W/kM), lapping HSF and slightly lapping the CPU, it will now POST at 1.95. I will fold proteins for 14 hours+ using 2.1 Vc. no lockups.
Like Campy I'm having game problems though. Removing crappy aureal soundcard and dropping the video acc. slider a notch will allow QIII to run. Delta Force LW will run if I underclock the GF256.
Also, the position of the RAM modules seems very important.
Really feel a bit lost right now. The actual CPU seems to handle serious number crunching just fine, but the peripherals complain. Seems strange that the GF would give me trouble when I'm not running the AGP out of spec.
Would additional cooling of the GeForce be beneficial? Perhaps using the retail celery HSF? I've also tried feeling the temp on the northbridge but it really seems cool (maybe this is an indication that the green HS is fastened poorly to the northbridge chip?)
Oh yeah, using the abit slotket. Funny I used to hear it was one of the best...
mb:aopen ax6b
snd: generic aureal
gfx: creative GF256
mem: winbond 7.5 ns (supposedly ok, can anyone comment?)
HSF: GW FOP32-1
abit slotket!!! (never heard anything bad about it until now, maybe I'll have to try another one...)
Dammit, so close now!
Suggestions please :)
/Link_flap
Link_flap
05-22-01, 05:13 AM
Hmm an additional note:
While looking at hardware monitor, I can see what looks like horrible voltage fluctuations. Especially the 3.3v can occasionally drop to 3.09v. Wonder if it's a mobo or a PS problem...
If anyone out there has a hack for upping the 3.3v to say 3.5, I'd be eternally grateful. I'm thinking somthing like "remove resistor x" or "add resisitor y"
/link_flap
Link_flap (May 22, 2001 05:13 a.m.):
Hmm an additional note:
While looking at hardware monitor, I can see what looks like horrible voltage fluctuations. Especially the 3.3v can occasionally drop to 3.09v. Wonder if it's a mobo or a PS problem...
If anyone out there has a hack for upping the 3.3v to say 3.5, I'd be eternally grateful. I'm thinking somthing like "remove resistor x" or "add resisitor y"
/link_flap
Yikes (!!) -- I couldn't find resistor X or Y on my board... sorry.
; )
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.