View Full Version : Got a real funky 1.0a, over here!
Just installed a Tualatin 1.0a in my Abit ST6R. I have been running a coppermine 733 on this board at a 148 bus with crucial pc133 cas2 at 2-2-2-7/9, stable.
I've been struggling to get this 1.0a above 1100, but with no luck whatsoever. I did the vid pin mod. At any speed between 1150 and 1400, it always posts, but hangs when loading Windows 98. This is at any voltage up to 1.8v. Win98 always locks at the same spot (just before the desktop loads).
I can boot into, and use Windows in safe mode at 1.4, no problem.
The funky part about it is that, even though Windows fails to load at 1150 at any vcore up to 1.8v, Windows loads and runs just fine at 1100 with 1.2v. YES, 1.2 vcore. Seems odd to me!
Right now, I'm running it at 1100/1.2v (Hardware Doctor is reporting 1.12v). I'm using a medium duty cooler, which is probably comparable to Intel's stock cooler. I rewired my fan for 5v. Been running Prime95 for the past hour with no errors, and Hardware Doctor shows my cpu temp at or below my case temp (37c)
It's nice to have a quiet pc, but this 110 mhz bus bites the big one!
At this point, I'm not sure if I got a bogus cpu, or if something else is going on here.
Was thinking about replacing my hard drive with a spare blank one, and attempting a fresh Windows installation at a higher speed (1300-1400) just to see what happens. I feel it won't work, but stranger things have happened.
BTW, I'm using standard, run of the mill, PSU. I was thinking about getting a better PSU, someday, but I'm not sure if it will help this situation much.
What do you guys think?
oldfart
01-11-02, 12:46 AM
Just a stab, and I dont know the ST6. Did you clear the CMOS?
Just a stab, and I dont know the ST6. Did you clear the CMOS?
No, but I can try it real quick.
Just tried cmos reset, and no go.
I also recorded a bootlog file to see where windows is stalling. I then compared the log to one from a completed load, and only two additional lines are present:
Init = Display Resources
InitDone = Display Resources
These last two lines are missing from the failed bootlog. This is where it always hangs.
BTW, I did try to use the Windows standard VGA driver, but it did not help.
I really don't think these 1.0a Celeron's are gonna be as overclocker friendly as we hoped them to be....The 1.2 still seems to hold the overclocking crown....
Still, you can't complain when you pay bugger all for basically a chip of a better spec than a 1.0 gig P3...
Make sure the "speed error check" is disabled in the BIOS. Also, make sure you have good cooling. Maybe the contact between the CPU and cooler needs to be rechecked. Hopefully you're using a good thermal paste and not some crappy cheap heat tape that often comes with CPU coolers.
Still, you can't complain when you pay bugger all for basically a chip of a better spec than a 1.0
I agree! I'm an economist overclocker. I'm always looking for more, for less money. On one hand, I'm happy to finally get above 1ghz with full P3 performance (on a budget), but this slow bus is bogus. Even though I didn't really trust the overclockability of these newest Tualerons, I still had high hopes for my little chip.
I'm going to work on this some more tonight, and throughout the weekend. If this is the best I can get out of this chip, I'm gonna have to consider an alternative.
Make sure the "speed error check" is disabled in the BIOS. Also, make sure you have good cooling. Maybe the contact between the CPU and cooler needs to be rechecked. Hopefully you're using a good thermal paste and not some crappy cheap heat tape that often comes with CPU coolers.
"Error check" is disabled, and I'm using standard silicone based compound. I don't think it's a heat issue. I can run Windows safe mode at 1400/1.65v with prime95 showing no errors, and temps stay normal.
Once again, this chip is stable at 1100 with below default vcore. At 1150 on up 1400 at any vcore, it fails. I would think that if the chip fails right at the tail end of loading windows at 1150, it should fail at an earlier point running at 1400. It always makes it to the same point at any speed. Maybe this don't mean squat, though.
So far, I feel I got a dud, but I want to rule everything out, first, before I give up on it.
jmsandrsn
01-11-02, 09:01 AM
About the only thing I could possibly think of would be the dividers. When you're trying 110 FSB, are you using a 1/3 dvider or 1/4 divider.
oldfart
01-11-02, 09:18 AM
Yeah check obvious stuff. AGP divider is 2/3, not 1/1. PCI divider is 1/3? Up towards 133 FSB, 1/4 PCI and 1/2 AGP.
ol' man
01-11-02, 10:01 AM
Originally posted by Rooski
"Error check" is disabled, and I'm using standard silicone based compound. I don't think it's a heat issue. I can run Windows safe mode at 1400/1.65v with prime95 showing no errors, and temps stay normal.
Once again, this chip is stable at 1100 with below default vcore. At 1150 on up 1400 at any vcore, it fails. I would think that if the chip fails right at the tail end of loading windows at 1150, it should fail at an earlier point running at 1400. It always makes it to the same point at any speed. Maybe this don't mean squat, though.
So far, I feel I got a dud, but I want to rule everything out, first, before I give up on it.
Hmm you can run the chip in safe mode at 1400MHz and 1.65v????? If you can run the chip at this in safe mode then it is not your chip!!!!!!!!!!
I really don't think these 1.0a Celeron's are gonna be as overclocker friendly as we hoped them to be....The 1.2 still seems to hold the overclocking crown....
Don't spread fallacies without knowing for sure. I have seen one do 1400MHz on default vcore. I think they will do just fine. I think mainly some of it is the inexperience of the OC'r.
Yeah check obvious stuff. AGP divider is 2/3, not 1/1. PCI divider is 1/3? Up towards 133 FSB, 1/4 PCI and 1/2 AGP.
The st6 has no setting for agp divider. I can only assume that it shifts along with the pci divider. When I try at 115 bus, my "cpu/sdram/pci" is set at "3/3/1". This is supposed to give me a 115 cpu bus,115 memory bus and 1/3 pci. All of wich I know is stable.
Well, it's been a couple years since I did a fresh win98 install. Who knows what kind of old crappy, useless vxd or dll I might be loading. Like I said earlier, the next thing I want to try is install one of my spare hard drives, and see if it flies with a fresh install. I also have 98se and win2000 I could try. I'll do this tonight when I get home from work (unless Mrs. Rooski has other plans).
UPDATE!!!!
Ok, I hooked up one of my spare IDE hard drives drives. I set the cpu to run on 133 bus with 1.675 vcore. Formatted the drive, and proceded with a fresh win98 (original) installation. Installation went without a hiccup, at 1330!
I then istalled my Lynksys pci nic, soundblaster live and DX8. Then I installed nVidia 23.11 drivers for my gf3 ti200. So far, no lock ups!
I then installed Prime95 and 3dmark 2001, for testing. Makes it through 3dmark just fine. Ran prime95 for about 20min with no errors. Hardware doctor shows cpu temp at around 38c with load. Heatsink is very cool to the touch. Idle temp showed at around 35c (as a reference, at 1.0ghz/1.475v, idle temp shows 33c).
I then experimented with 1.4ghz. Settled in at a 1.65 vcore. It finishes 3dmark no problem, and ran prime95 for about 20 min with no errors. cpu temp showed around 41c under load. Heatsink was still very cool to the touch Just for the hell of it, I unplugged the fan, while running prime95 at 1.4ghz, for a few minutes. The cpu temp rose and hovered around 48-50c. Heat sink finally started to feel a little warm. I didn't want to leave the fan off for too long, though. I wonder if you can do this with an AMD.
Anyways, I decided to try 1.45 real quick, but even with a 1.75 vcore, it wouldn't finish 3dmark.
Well, I still need a lot more testing, but it looks like I'll be settling at around 1.4. I'm happy with that, and I'm especially happy that the chip is not bogus, after all. After what I've read, in the recent past, a cool 1.4 is what I was expecting to get. I may still fight to sqeeze a bit more, but I also want to keep the temp down, and be able to run the cpu fan on 7v, or maybe even 5v. I like a quiet pc! I also might just invest is a high quality PSU, and see it it can help out a bit.
Now I have to figure out what to do with my "chip-tripper" win98 install. I have an idea on how to salvage it, but I may just have to start over. This is something I've been thinking about doing, anyways, but didn't want to hassle it. Oh joy!
Guess it was Windows, you'll get a tad more soon :D
Originally posted by ol' man
I have seen one do 1400MHz on default vcore.
I haven't seen one in person, but I stumbled upon a japanese site dedicated to celeron overclocking. They have a 1.0A doing 1.4 GHz on vcore as well.
@Rooski:
Congrats! :beer:
IMO it's best to only to get your data from your old 98 installation and start with something fresh. I'd make a drive image of your boot partition after you've got everything in place.
oldfart
01-12-02, 07:54 AM
Rooski, Congrats! I'm glad it's going better. You had pretty much the same results as I did. 1.4 Ghz @ 1.65 Vcore. I could never get mine to even post @ 1.45 Ghz.
UPDATE!!!!
I won't bore you guys with details of the hours I spent trying to salvage my win98 install, but in the long run, upgrading my current win98 to "second edition" cured the problem! I'll use this for now, til I decide to start fresh with something.
I'll keep it at 1.4ghz/1.65v for now, and keep testing stability. So far, so good. Current cpu temp with prime 95 running shows about 1c-2c above case temp(with fan running on 7v). Later, I'm gonna do some more testing the limits of this cpu.
I'm still considering upgrading to a better PSU, and see it it can give me some headroom. I'm using a "run of the mill" 250w PSU that came with my "run of the mill" case.
Have any of you guys experienced an increase in overclockability after upgrading to a better PSU?
Well, my adventure has come to an end, for now.
Thanks for all your input, guys!!
oldfart
01-12-02, 01:06 PM
My setup has a 300W Antec 303X PS.
Don't spread fallacies without knowing for sure. I have seen one do 1400MHz on default vcore. I think they will do just fine. I think mainly some of it is the inexperience of the OC'r.
That's the biggest load of crap i've ever heard......
Is it your 'experience' that helped you stuff 2 or 3 Tualatin based chips?!
I'm sorry ol'man but you don't know my abilities from a bar of soap and the statistics are stacking up AGAINST the Celeron 1.0a as being a good overclocker AND it stands to resason that intel would turn the lower quality of the yeild into 1.0 gig chips....Intel dosen't give a crap about how well you can overclock their chips! You make it sound like they take it into consideration!
Up until recently i've thought of you as a pretty cool bloke....Mabye I was wrong....
Please stop flamebaiting....
I really don't think these 1.0a Celeron's are gonna be as overclocker friendly as we hoped them to be....The 1.2 still seems to hold the overclocking crown....
I'm sorry ol'man but you don't know my abilities from a bar of soap and the statistics are stacking up AGAINST the Celeron 1.0a as being a good overclocker AND it stands to resason that intel would turn the lower quality of the yeild into 1.0 gig chips....Intel dosen't give a crap about how well you can overclock their chips! You make it sound like they take it into consideration!
Sorry Flu!d, but I disagree!
Determinig the overclockability if a cpu has nothing to do with top speed. It has to do with the ratio between it's rated speed (sold as) and it's overclocked speed.
Your sig says you run your 1200 at 1500. That's a difference of 300, or 25% over rated speed.
So far, all the 1.0a's I've heard of can fairly easily do 1400 with medium air cooling (as my 1.0a does). That's a difference of 400, or 40% over it's rated speed.
My 1.0a clocks 40% above default.
Your 1.2 clocks 25% above default.
So, it would seem that my 1.0a is a better overclocker then your 1.2.
In order to get a 40% overclock on a 1.2, you would have to run it at 1680. Good luck with that, especially with cheap air cooling.
The other factor in overclocking is bus/memory speed. With your chip at 1500, your slow bus speed is crippling your overall performance. My chip at 1400 combined with the faster bus/memory speed will be overall faster then yours. This is why I held out for the 1.0a, over the 1.2.
So, based on these factors, I would have to say the 1.0a (based on what we know, so far) is a better overclocker then the 1.2, and I'm sure most would agree.
The 1.0a holds the current crown!
When the 1.0a Celeron was first released most people (including myself) believed they would be capable of 1500 mhz @ 150 mhz FSB quite easily....Now the chips are starting to make their way into circulation it appears 1500 mhz is a big ask, especially without what I consider to be very high vcores (and therefore higher heat output)....Therefore I believe it is a bit of a dissapointment, especially considering 1.5 gig is well within the Tualatin core.
As far as memory bandwidth is concerned....I get 437/450 on my Gigabyte mobo....That's faster than My P3 1000EB @ 150 mhz FSB CAS 2,3,2,7/9 (right now i'm running 125 mhz FSB CAS 2,2,2,5/7) - the ram speeds @ 150mhz FSB were the highest I could achieve with my RAM at the time. I really don't believe a FSB advantage of 15 mhz is even going to be noticeable on a chip that isn't terribly bandwidth dependant (unlike the P4 and Athlon cores) anyway....And, as you can see, I get fantastic memory bandwidth.
I have had a lot of experience with P3 chips running a default 133 mhz FSB (which therefore overclock at about the 150 mhz or higher FSB reigon) and I have found that high FSB's on P3 chips don't return the peformance advantages many people believe they do, i'm not saying you don't get peformance advantages in SOME (definatly not all) areas of computer useage, i'm just saying they are no where near as great as many believe them to be....This is also reflected in the comparitivly low peformance gains generated by a P3 on a DDR platform not to mention the ill fated P3 / RAMBUS (i820) platform - You're lucky to get a 5% advantage over an SDRAM platform!
Yes....Percentage wise your 1.0a Celeron gains a higher overclock than my 1.2....But as you can see by the voltages, it's definatly working harder to get there....My chip barely even raises a sweat with the standard Intel cooler....
funnyperson1
01-13-02, 04:54 PM
Originally posted by Rooski
Sorry Flu!d, but I disagree!
Determinig the overclockability if a cpu has nothing to do with top speed. It has to do with the ratio between it's rated speed (sold as) and it's overclocked speed.
Your sig says you run your 1200 at 1500. That's a difference of 300, or 25% over rated speed.
So far, all the 1.0a's I've heard of can fairly easily do 1400 with medium air cooling (as my 1.0a does). That's a difference of 400, or 40% over it's rated speed.
My 1.0a clocks 40% above default.
Your 1.2 clocks 25% above default.
So, it would seem that my 1.0a is a better overclocker then your 1.2.
In order to get a 40% overclock on a 1.2, you would have to run it at 1680. Good luck with that, especially with cheap air cooling.
The other factor in overclocking is bus/memory speed. With your chip at 1500, your slow bus speed is crippling your overall performance. My chip at 1400 combined with the faster bus/memory speed will be overall faster then yours. This is why I held out for the 1.0a, over the 1.2.
So, based on these factors, I would have to say the 1.0a (based on what we know, so far) is a better overclocker then the 1.2, and I'm sure most would agree.
The 1.0a holds the current crown!
i agree.....i have a 46% oc on my hands....i want 50:mad: grrrr
When the 1.0a Celeron was first released most people (including myself) believed they would be capable of 1500 mhz @ 150 mhz FSB quite easily....Now the chips are starting to make their way into circulation it appears 1500 mhz is a big ask, especially without what I consider to be very
high vcores (and therefore higher heat output) ....Therefore I believe it is a bit of a dissapointment, especially considering 1.5 gig is well within the Tualatin core.
Well, the problem was that people just assumed the 1.0a and 1.1a's were going to be exactly like a 1.2, but locked at different multipliers. From what I understand, even though the multipliers are locked during production, Intel does have the capability of speed testing their chips, and setting multiplers AFTER testing. I would assume this is done before the chip is attached to the PGA. So, the 1.0a's are probably the lower speed yields that were swept up from the assembly line floor, and packaged for sale as a new product. I feel that the new lower speed Northwoods are gonna be a simular situation.
as far as the higher voltages go ......Yeah, I didn't want to go as high as I had to (at 1.7 now to keep stable), but it still runs cooler then any p3 I've used in the past. Hell, If I blow it up, so be it. It's inexpensive enough to take the risk.
As far as memory bandwidth is concerned....I get 437/450 on my Gigabyte mobo....That's faster
than My P3 1000EB @ 150 mhz FSB CAS 2,3,2,7/9 (right now i'm running 125 mhz FSB CAS
2,2,2,5/7) - the ram speeds @ 150mhz FSB were the highest I could achieve with my RAM at the
time. I really don't believe a FSB advantage of 15 mhz is even going to be noticeable on a chip
that isn't terribly bandwidth dependant (unlike the P4 and Athlon cores) anyway....And, as you
can see, I get fantastic memory bandwidth.
I have had a lot of experience with P3 chips running a default 133 mhz FSB (which therefore
overclock at about the 150 mhz or higher FSB reigon) and I have found that high FSB's on P3
chips don't return the peformance advantages many people believe they do, i'm not saying you
don't get peformance advantages in SOME (definatly not all) areas of computer useage, i'm just
saying they are no where near as great as many believe them to be....This is also reflected in the
comparitivly low peformance gains generated by a P3 on a DDR platform not to mention the ill
fated P3 / RAMBUS (i820) platform - You're lucky to get a 5% advantage over an SDRAM
platform!
Some valid points, most of which I agree. I guess in the long run, either chip can be a good choice. Besides benchmarking, neither of us will notice a performance difference between our machines, anyways.
But, I still feel that a fairly easy 40% overclock is good for a current version cpu. We may never see another 300a, again.
BTW, I went ahead and started rebuilding my Win98 installation from scratch. Gee, it's amazing how much faster a fresh OS install can feel! :)
Some valid points, most of which I agree. I guess in the long run, either chip can be a good choice. Besides benchmarking, neither of us will notice a performance difference between our machines, anyways
I totally agree!:D
Good luck with the new install dude! I usually do a reinstall every 3 months (kinda like a major service for my computer!).
oldfart
01-14-02, 07:45 AM
Some good points brought up here. I agree on the C1.0A. I dont like to juice up a Vcore a bunch to get a good overclock. I wasn't happy with having to go to 1.65 Vcore to get 1.4 Ghz. That is why I got a PIII-S which overclocks nicely and needs no Vcore increase at all (plus it's fast as Hell)!
FSB memory bandwidth is not something that should be ignored. Personally, I would not want a sub 133 MHz system. That is why I stayed away from a C1.2. I have posted a few benches below. Sandra scores need to be taken with a large grain of salt. I get a different score each time I do one. All of the below were first run tests after a restart. All tests were done on a GA6OXET, 512 meg Crucial PC133, 2-2-2 5,7 timings. I dont have a complete set for each CPU, but enough for some good comparisons.
Sandra 2001te Mem benchmark:
125 FSB:
PIII-S 1.26 @ 1.19 GHz (underclocked): 436/427
133 FSB:
CuMine 1000EB @ 1000: 413/457
PIII-S 1.26 @ 1.26: 467/457
Sandra ref 815E: 397/468
Sanda ref BX: 415/475
140 FSB:
C1.0A @ 1.4: 429/423
PIII-S 1.26 @ 1.33: 490/477
150 FSB:
CuMine 1000EB @ 1125: 424/477 (this might have been @ 3-2-2 5,7)
PIII-S 1.26 @ 1.42: 525/512
152 FSB:
PIII-S 1.26 @ 1.44: 532/519. This is my current setup.
The effect on memory BW with a higher FSB is pretty dramatic. Memory BW is a weak spot with an SDRAM system. A higher FSB does help out considerably. The PIII-S responds very well to a higher FSB.
The only chip that really benifits there is the P3-S....The reason being is that you are actually seeing the benifits of 512k cache (which Sandra mem benchmark takes into consideration). Sandra mem bandwidth benchmarks must be peformed on a clean boot with no background programs running, I can produce my results time and time again and as you can see there is nothing wrong with my sub 133 mhz FSB memory bandwidth at all! This is also reflected in Quake III timedemo benchmarks (Quake III is very bandwidth dependant)
Really, above 120 mhz FSB with fast RAM the P6 archatecture has all the memory bandwidth it needs....Like I said before, this is reflected in the low peformance increases obtained from the i820/RAMBUS saga....The P6 archatecture is not terribly bandwidth dependant and is one of the only platforms that still 'loves' good 'ol SDRAM....
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.